Searching \ for '[EE]: Design Challenge - low power step down switc' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/power.htm?key=power
Search entire site for: 'Design Challenge - low power step down switc'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: Design Challenge - low power step down switc'
2001\08\20@193219 by Russell McMahon

picon face
An ongoing theme is "how can I supply eg 5 volts from a much higher voltage
supply without excessive power dissipation?"

I have a design I would like to share that will meet many such needs cheaply
and simply. Rather than reveal it immediately I propose a design challenge
along the lines below. Hopefully some useful ideas will come out of this.


Requirement

Step down voltage regulator suitable for powering small low voltage circuits
from a higher voltage supply with a reasonable power dissipation.

Overall figure of merit = finished cost while meeting design specification.
Costing rules provided below.

Switching regulator for reasonable efficiency.
   Any topology.
   Efficiency need not be superb but should be substantially
   better than an equivalent liner regulator over most of the
   input range.

Power out
   In this instance "'small" - say a few tens of milliamps
   Ideally scalable as required.

Output voltage
       Fixed but selectable over say 2 to 12 volts.
       Output would ideally be useable as is for operating typical
       processor based circuits but the use of a linear post regulator
       is acceptable if required. Cost of post regulators not included in
       costing.
.
Input voltage
       Always greater than output.
       Minimum: as little as 1  volt above design output
       Maximum: More the better -
           (preferably at least 50 volts )
    A design with a varying input voltage is desirable.
    Fixed Vin/Vout (eg 24 to 5) designs would be of
    interest but less useful generally.


Output regulation
   Smaller change the better
   Ideally +/- 0.1 volts across all Vin range and across
   load variation from 10% to 100%

Isolation
   Input to output isolation would be a bonus but is not required. This
allows almost any desired regulator topology (buck, flyback, forward, CUK,
etc) as desired.

Noise / EMC
   Output noise should be "reasonable" but post filtering for noise may be
necessary and is not included in the costing. A design that does not need
any post filtering is better. Extra filtering should NOT be required to meet
regulation specification. EMC / radiated or conducted noise should be
amenable to limiting by normal means.


BASIS OF COSTS

Components used are costed in Arbitrary economic units as follows. Any
relationships to $NZ in volume is entirely coincidental. All names have been
changed to protect the guilty. No animals used in testing this product. Each
of these units is very approximately 0.5 US cents. This costing is assumed
to be proportional to all up cost in an amateur environment. ((Real costs
can be scaled by an appropriate factor to match the currency of your
choice. ))

1    resistor 1/2w 200v max rated
5    transistor, TO92 pkg, general purpose (BC337 etc)
10    FET small signal TO92.
1    diode, small signal, 75v 150 mA (1N4148)
2    diode, low frequency, 1A, 400 volt (IN400x)
10    Diode, high frequency
10    Cap plastic up to 0.1 uF
20    Cap plastic 0.1 to 1 uF
5    Capacitor electrolytic 1 uF - 10 uF
10    Capacitor electrolytic 10 uF -100 uF

50    Inductor, single winding
10    extra isolated windings
5       extra winding connected to existing winding.

25    Comparator, 4 in pkg (LM339(
30    Opamp, 4 in pkg (LM324)
100   Switching regulator IC (384X, MAX, ... your choice)
50     TO92 Linear regulator (78L05 etc)

If you don't find what you need here describe it and suggest an appropriate
cost.
Also adjust any of the above if you feel it justified (but say why).

Provide brief estimated specifications (selectable output range, max and min
input range, dropout (minimum in to out voltage), regulation with changes in
load and input voltage,  noise, limitations and advantages of design.

Example:
=======

Simple switcher / MAXIM style buck regulator, with input filter cap, output
cap, inductor, all internal components

Cost    Item

15       caps
100    IC - super whizzbang all up one chip
50      1 winding inductor
2         resistors for Vout selection

167    Total

Comments: in reality usually more "glue" components would be needed.
Most  dedicated switching regulator ICs won't work at 50 volts input.
A very simple design with low PCB area.
Power limited by IC ratings and dissipation.
Power switch and flyback diode included in IC


          ---- INDUCTOR---
         |     -----------------    |
Vin--o---|        IC          |--o---.----> Vout
         |      -----------------   |
       ==  Cin                  ==  Cout
         |                              |
---------------------------------------------




       Russell McMahon

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\21@073614 by Russell McMahon

picon face
part 1 1070 bytes content-type:text/plain; (decoded 7bit)

> Requirement
>
> Step down voltage regulator suitable for powering small low voltage
circuits
> from a higher voltage supply with a reasonable power dissipation.
>
> Overall figure of merit = finished cost while meeting design
specification.
> Costing rules provided below.
.................

OK - here's a solution that has been suggested by a friend from the NatSemi
LM336 regulator diode data sheet

Without going back to my original cost list I estimate that this is about
130 to 150 units cost. (Under $US1 probably with substitutions below). Not
too bad.
I've pretended that the LM336 here can be substituted with a low cost zener
diode - if not the price is significantly higher.
I've assumed that all transistors can be low cost general purpose ones BC337
/ BC327 as the power level specified in my challenge is low.
Also, the output catch diode could be a 1N4148 ! - not what you would
normally see in this role but again, the power level is low.

Any improvements on this?




       Russell McMahon



part 2 5663 bytes content-type:image/gif; (decode)


part 3 144 bytes
--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@083633 by Roman Black

flavicon
face
I'd say it can be done with one transistor and
a very small coil with two windings...
:o)
-Roman


Russell McMahon wrote:
{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@144336 by D. Schouten

flavicon
face
Russell,

Is this the circuit you were referring to in the thread "tiny power
supply"?

Thanks,

Daniel...

{Original Message removed}

2001\08\21@150013 by Eisermann, Phil [Ridg/CO]

flavicon
face
>
> OK - here's a solution that has been suggested by a friend
> from the NatSemi
> LM336 regulator diode data sheet
>

       That's going to be hard to beat for cost, I think. Maybe one of
those self-oscillating ones Roman mention might be cheaper (Royce
oscillator?)

       I guess for low power levels, you don't need the power PNP, so its
down to three bipolars and some passives. I've seen a similar one that uses
a NPN and PNP in a multivibrator configuration instead of the differential
pair. uses one more capacitor, but two less resistors.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@180233 by jamesnewton

face picon face
Ok, Roman, Time to put your cards on the table? How do you wind the coil and
what exactly is the circuit?

---
James Newton, Admin #3 spam_OUTjamesnewtonTakeThisOuTspampiclist.com
1-619-652-0593 VM 1-208-279-8767 FAX
PIC/PICList FAQ: http://www.piclist.com or .org

{Original Message removed}

2001\08\21@182524 by Russell McMahon

picon face
> I'd say it can be done with one transistor and
> a very small coil with two windings...
> :o)
> -Roman


I tend to agree - with reservations.
Part of the reason for my doing it this way was to bring out the good ideas.
A self oscillating "ringing choke" / flyback converter is indeed one
possible solution.

Tell us how.
How many of the design requirements are met by this.


           Russell McMahon

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@190319 by Russell McMahon

picon face
> Is this the circuit you were referring to in the thread "tiny power
> supply"?

No. But it's going in the right direction.
My circuit is substantially simpler and cheaper.
Probably also lower performance in some areas but maybe not.

Will post shortly - lets see what else comes up first. I hoped for more
ideas than we have seen so far.




       Russell



{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@190333 by Russell McMahon

picon face
> > OK - here's a solution that has been suggested by a friend
> > from the NatSemi
> > LM336 regulator diode data sheet
> >
>
>         That's going to be hard to beat for cost, I think. Maybe one of
> those self-oscillating ones Roman mention might be cheaper (Royce
> oscillator?)
>
>         I guess for low power levels, you don't need the power PNP, so its
> down to three bipolars and some passives. I've seen a similar one that
uses
> a NPN and PNP in a multivibrator configuration instead of the differential
> pair. uses one more capacitor, but two less resistors.


Getting better!
Can you post a simple circuit and comment on its capabilities.


       Russell McMahon

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@191813 by Jim

flavicon
face
What was the ripple requirement/EMI compatability
on this?

I have built this style of PS before - to power
a PRC-10 radio (old mil radio) and develop the
B+ (plate) potentials of +67 and +135 V plus the
+1.5V filament power from a +12V battery.

Used a toroid for transformer core ... it's been a few
years back now ...

Jim

{Original Message removed}

2001\08\21@194601 by steve

flavicon
face
--Message-Boundary-11782
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body

A lateral thought solution.
Since this is the PICLIST and the application will no doubt have a
PIC in it, I've taken the liberty of using one with a comparator.
Since it had to be there anyway, there's no cost to the power
supply. Same with the LED and its resistor. They were there
already.
The LED and additional diode make a simple voltage reference and
also enough voltage to get the PIC started. The PIC wiggles the
port pin as long as the comparator says it wants more voltage.

Cost 91 units.
For low power applications you could probably remove the diode
and inductor and just use a resistor, topping up the capacitor as
needed. (Cost 41 units).
I have no idea if it would work or not.

Steve.

======================================================
Steve Baldwin                Electronic Product Design
TLA Microsystems Ltd         Microcontroller Specialists
PO Box 15-680, New Lynn      http://www.tla.co.nz
Auckland, New Zealand        ph  +64 9 820-2221
email: .....stevebKILLspamspam@spam@tla.co.nz      fax +64 9 820-1929
======================================================

--Message-Boundary-11782
Content-type: text/plain; charset=US-ASCII
Content-disposition: inline
Content-description: Attachment information.

The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any another MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.

  ---- File information -----------
    File:  psu.gif
    Date:  22 Aug 2001, 11:37
    Size:  6497 bytes.
    Type:  GIF-image

--Message-Boundary-11782
Content-type: Image/GIF; name="psu.gif"
Content-disposition: attachment; filename="psu.gif"
Content-transfer-encoding: BASE64
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--Message-Boundary-11782--

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\21@224059 by Russell McMahon

picon face
> What was the ripple requirement/EMI compatibility
> on this?

Ripple was "less the better, tell us what you achieve" :-)
See below for closest part of original spec.

A circuit of your valve supply driver and comments would be of interest.
(There are lots of paper designs for this sort of things but real circuits
and people's experiences are invaluable).



       Russell

_________________________________________

Output regulation
   Smaller change the better
   Ideally +/- 0.1 volts across all Vin range and across
   load variation from 10% to 100%

Noise / EMC
   Output noise should be "reasonable" but post filtering for noise may be
necessary and is not included in the costing. A design that does not need
any post filtering is better. Extra filtering should NOT be required to meet
regulation specification. EMC / radiated or conducted noise should be
amenable to limiting by normal means.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\22@062631 by Roman Black

flavicon
face
Hi Russell, I'm too busy this week to do it, but would
really love to.

I seee a one transistor self excited device which
uses one of the choke coils for neg feedback to
continue oscillation. Basically it allows a decent max
current in normal oscillation, say 150mA, then as the
output voltage reaches a setpoint this biases the
oscillator off, giving both a fixed max current and
crude voltage regulation. But with very low parts cost
and small size.

The dual-coil could be made from a tiny "balun" 8 former,
like used in TV antenna fittings and available for a few
cents, or a cheap small toroid.

I think this would be a nice solution, as a ringing-choke
style design is max current limited to start with, and
it's easy to add another feedback wire to give max
voltage regulation. A final zener can be added if the
expected 5% voltage regulation is not good enough.
:o)
-Roman

Russell McMahon wrote:
{Quote hidden}

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@074050 by Russell McMahon

picon face
part 1 5329 bytes content-type:text/plain; (decoded 7bit)

> > OK - here's a solution that has been suggested by a friend
> > from the NatSemi
> > LM336 regulator diode data sheet
> >
>
>  That's going to be hard to beat for cost, I think. Maybe one of
>  those self-oscillating ones Roman mention might be cheaper (Royce
>  oscillator?)
>
>  I guess for low power levels, you don't need the power PNP, so its
> down to three bipolars and some passives. I've seen a similar one that
> uses a NPN and PNP in a multivibrator configuration instead of the
> differential pair. uses one more capacitor, but two less resistors.

That was quite a good surmise :-)

My solution is given below - its a classic buck regulator but without most
of the components usually used.
Don't let that stop people suggesting other circuits - the more the better.
"Cost" using the components mentioned below is about 100 units which is
pleasing for a fully stand alone unit.
A single transistor self oscillating flyback converter MAY come in cheaper
but for reasons discussed below, may be inferior overall.

This design is simple to build, forgiving, well behaved, easily scaled for
power and voltage and reasonably efficient.
It SHOULD have use in many amateur applications.

This circuit is special - it was provided by God (no kidding).

Those here who don't believe in a personally accessible deity can attribute
the design to me but I assure you that the basic concept was divinely
inspired. Anyone interested ask offlist and I'll provide more detail. I'm
not saying it hasn't been proposed by other people before now - just that
that's how I got the design. The NatSemi app note, which I 1st saw for the
first time yesterday, is the closest to it that I've seen.

The design requirement was much as stated in the original challenge with the
following changes or notes
Reliable, cheap, simple, reliable, minimum parts count, cheap.
12 to 200 volts input (!) for 10.5 volts output.
Graceful degradation of output for Vin under 12 volts.
Reasonable regulation over load and input voltage.
Minimal EMI.
Did I say cheap ? :-)


I've snipped and tidied this diagram from a larger circuit so a few
components need changing.
The P channel FET used here can be replaced by a much smaller one or by a
small pnp transistor.
The BYV26C diode can be replaced by a 1N4148 (!)
If using a transistor instead of a FET ZBUK2 and RBUK4 can be removed.
The insanely minimalist can probably remove PBUK1 and CBUK1 and possibly
even RBUK1 !!!

As shown here it provide up to an amp or so at 10.5 volts out.

Regulation is (to my surprise) about +/- 0.1 volt across wide load and Vin
variations.

Leaving all these in, changing to a 1N4148 and BC327 instead of the FET we
get a costing of .101 units (allowing 5 for the 10 volt zener). This
includes the unneeded extra diode and upper zener.

I will give only a brief overview here - if the traffic warrants I'll add
more in due course.
I believe that this should prove very worthwhile for many amateur
applications. I am personally very pleased with it and will be finding lots
of things for it to do in future.

Output voltage is approx Vzbuk1 + 0.6 volts.
Output voltage for Vin < Vout-design + 1 volt is about 1 volt below Vin down
to around Vin = 4 volts.

OPERATION:

All off.
Vout = 0
All transistors off.
Apply Vin.
QBUK2 turned on via Vin / RBUK2
QBUK2 on turns BUKFET on via RBUK3.
If a P channel FET is used then the 13v zener limits the maximum gate drive
voltage and RBUK4 provide turnoff bias.
BUKFET (or transistor) on provides current to Vout via LBUK1.
CBUK2 charges until Vout reaches design voltage.
At Vout = design voltage ZBUK1 conducts and turns on QBUK1 via RBUK1.
QBUK1 on drives base of QBUK2 to ground turning QBUK2 off. QBUK2 off turns
BUKFET off and interrupts current fed to LBUK1.

Voltage on LBUK1 "flips" and stored energy in LBUK1 continues to be
delivered to output via DBUK2 and LBUK1 (standard buck regulator operation).

Voltage out CONTINUES TO RISE due to stored energy in inductor.
When energy is delivered Vout.begins to fall and ZbUK1 stops conducting,
QBUK1 turns off and the cycle repeats.

The OFF time is affected by energy storage relative to load current and the
size of CBUK2

Size of LBUK1 will vary with power level and application.

The units in use so far have had a short circuit current limit circuit added
(not really needed, 2 more transistors) and an extra bipolar drive
transistor added to the FET to improve the drive waveform.  This helps
improve the efficiency at high input voltages which is especially important
when Vin can range over a 20 to 1 ratio!.For most applications with Vin in
the 12 to 50 volt range this is less of a concern.

A single transistor self oscillating forward converter would almost
certainly be slightly cheaper, but not by much.
Scalability of these is problematic, noise is a greater problem, the
inductor is more critical and the second winding needs designing (as opposed
to the buck inductor which is very insensitive to design changes and can
usually be an off the shelf item. Regulation of the flyback converter will
probably be inferior.



That's enough for now.
We'll see if there's enough interest to discuss it more.

Thoughts ?


       Russell McMahon


.







part 2 4235 bytes content-type:image/gif; (decode)


part 3 131 bytes
--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@090916 by Jim

flavicon
face
I'll see if I can do something with the documenmtation
tomorrow (if interest still exists) - I had built a passive
load box (with switched load, test points) and everything
for development purposes - this was in 1983. I took a
quick peek at the *paper* schematic and notes awhile
ago - this design actually uses an opamp and several
transistors - the extra components are for providing
regulation.

A power supply Eng/friend originally designed it - he
also did the PS design in the LANTIRN Radar pod for the
F-16 (in  *case* anyone needed that tidbit of information).

This design requires a multitude of windings (four!) on the ferrite
xfmr just for basic operation of the PS circuitry, namely, windings
for: 1) Collector  (to supply +V)     2) base drive/feedback to driver
xsister  3) -6V (for regulator bias) and  4) then the  'load' winding(s)
(as req'd).

The -6V 'bias' winding can also be used by external loads.

I think the 'buck' regulator, for a single output, is really the way to
go - much simpler inductor ...

The *4* outputs this supply genned (requiring such a complex set
of transformer winding) were:
1) -6V (PA tube fils)      2) +1.5 for rcv/exctr fils
3) +67 for receiver B+   4) +135 V for xmitter B+.

So, in total, this ferrite core xfmr had 7 windings - the HV
windings could actually have been tapped - if half-wave
rectification was acceptable, making the count 6 windings.

Jim

{Original Message removed}

2001\08\22@120749 by Byron A Jeff

face picon face
> My solution is given below - its a classic buck regulator but without most
> of the components usually used.
>
> This design is simple to build, forgiving, well behaved, easily scaled for
> power and voltage and reasonably efficient.
> It SHOULD have use in many amateur applications.
>
> This circuit is special - it was provided by God (no kidding).

And it is a Godsend!

Switching regulators has been on my list the last few months due to my
perpetual desire to power PC's in my vehicles. Starting out with MP3
applications but considering moving into DVD and other digital video apps.

There is a bunch of stuff out there about buck and boost theory. But two
things became apparent: No details are discussed on the error/control/feedback
channel and that everyone expects a fixed frequency timing, with just a
variation on the duty cycle to control the voltage.

You've blown both of those out of the water! The discussion of the feedback
through ZBUK1 and QBUK1 is clear and inspired. And since ZBUK1 is the sole
voltage control element, one a trivially change the voltage by replacing it.

Quite impressive.

Now on to my questions:

1) You talked about replacing BUKFET with a PNP bipolar. What about an NPN
bipolar? Remove QBUK2, RBUK3, ZBUK2, and RBUK4 and tie the bottom of RBUK2
and the collector of QBUK1 to the base of the NPN. When QBUK1 is off then
RBUK2 will turn the power NPN on. When QBUK1 conducts, the NPN will turn off.
I'm sure that the base current from RBUK2 will need to be higher so that
the power NPN will saturate.

2) Is RBUK3 the current limiting resistor for ZBUK2? Just making sure.

3) What element requires the 1V headroom?

4) Is the BUKFET and the inductor the only power elements? My goal is to build
a 5V 10A supply.

And my new design challenge:

How could this circuit be augmented so that it could output 12V from an
input between 6V and 15V? 15V: Car running normally. 6V: Car starting.
In my previous design I used a gel cell battery to steady the input current.
However the battery doesn't have enough headroom to satisfy the design
requirement.

Some ideas:

1) Create a boost/buck design where the boost stage always pumps to 16V or
so giving the required headroom. A boost design can be done from this same
circuit with a small rearrangement of the power elements. Instead of switch
(BUKFET), diode, inductor use inductor, switch, diode:

Vin ----IIIIIIIII------+-------DDDDDDDDDDD--------+------ Vout
                      |                          |
                      S                         Cap
                      |                          |
                     GND                        GND

Use exactly the same control system and Vout will rise to the ZBUK1 control
voltage.

2) Since I'm going to have a 5V circuit anyway do it in the opposite direction
and boost the 5V output to 12V.

3) Add a couple of more control elements so that the supply can switch from
buck to boost. consider:

Vin ---S1-+--IIII------+---+--------D2-------+----+------ Vout
         |            |   |                 |    |
         S2           S3  +--------S4-------+   Cap
         |            |                          |
         D1          GND                        GND
         |
         GND

In buck mode:

S1: osciallates. It's BUKFET in the original design.
S2: on. catch diode for buck
S3: off. Boost power element
S4: on. Boost catch diode D2 bypass

In Boost mode:
S1: on. Passes Vin to boost
S2: off. buck catch diode bypass
S3: oscillates: The boost BUKFET.
S4: off. Use the D2 boost catch diode

At first glance it seems like it's more work than it's worth.

Any ideas?

I plan to build my 5V version in a short time frame. I'll probably implement
suggestion #2 above unless someone has another idea.

BAJ

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@124049 by Spehro Pefhany

picon face
At 11:32 PM 8/22/01 +1200, you wrote:

>If using a transistor instead of a FET ZBUK2 and RBUK4 can be removed.
>The insanely minimalist can probably remove PBUK1 and CBUK1 and possibly
>even RBUK1 !!!

I'd say PBUK1 is the most necessary of the lot, as it sets the Zener
current.

This type of circuit was popular before PWM ICs became widespread.

I think I'd like to see a bit of hysteresis in the two bipolar
transistors to assure that it actually starts oscillating. You can do
this by tying the (currently grounded) emitters to ground through a
low-valued resisistor and altering the vales of RBUK2 vs. RBUK3 to
give a bit of snap to it (it increases slightly the input voltage,
and turns the transistors into a ST circuit.  Otherwise it could
conceivably go linear and burn up the MOSFET or transistor.
inductor and o

The complementary version with an N-channel MOSFET might be interesting.
High voltage P-channel MOSFETs are not common.

Best regards,


P.S. where there is AC involved, here is a totally different idea, but
one that uses similar components and no inductor. ;-)

http://www.trexon.com/articles/edn_Mosfetdesign.pdf


=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speffspamKILLspaminterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
Contributions invited->The AVR-gcc FAQ is at: http://www.bluecollarlinux.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@132334 by Eisermann, Phil [Ridg/CO]

flavicon
face
part 1 1628 bytes content-type:text/plain;Here it is. Very similar to your, in that it is a self-excited type.

Initially, everything is off. Apply V+, and Q1 conducts through R1. C1
begins charging through L1 until the design voltage is reached: (Vzener+Vbe)
= (R6/(R6+R5)Vout. Then Q2 conducts and dumps current into R1, raising the
base voltage of Q1 until it turns off. D1 is the free-wheeling diode of the
buck regulator. C3 reduces gain at high frequency to reduce ringing, but
circuit will work without it. CIrcuit probably doesn't need R4 either.

When input voltage is too low, then Q1 just conducts current and the output
voltage tracks input voltage within the saturation voltage of Q1. With
proper components, it can be made to work at fairly high voltages.

Now, lets make this thing work with an N-channel FET. Then we'd have a
pretty cheap housekeeping supply for off-line switchers :) I'll see what I
can come up with.


> {Original Message removed}
part 2 14435 bytes content-type:application/octet-stream; (decode)

part 3 131 bytes
--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@150628 by Olin Lathrop

face picon face
> My solution is given below - its a classic buck regulator but without most
> of the components usually used.

Hmm.  You've tested this and found it to meet your needs, but there are a
few points nagging me about this design.  I guess you get away with most of
these issues because the currents are so low.  Here are some issues:

1  -  There is no hysterisis in comparing the output level with the desired
threshold voltage.  Although unlikely, there is no guarantee this will
oscillate at all.  It could, in theory at least, revert to being a linear
regulator.  A little bit of feedback from the collector of QBUK2 to the base
of QBUK1 would guarantee instability and also make sure the FET gate is
being driven to the extremes.

2  -  CBUK1 doesn't belong there.  It will make the regulator slow to turn
on in response to a voltage drop.  It probably makes the feedback less
stable, but there are better ways to do that (note 1).

3  -  The comparator is two cascaded common emitter stages with passive
pullups.  These are going to be slow to respond, asymmetrical, and have a
low slew rate.

4  -  The FET will turn off very slowly because the only thing driving the
gate is a 100K resistor.  This means the FET will spend a proportionately
long time in the transition region.  This may be OK in your situation, but
in general this reduces efficiency and dumps power into the FET.  This would
definitely be a problem if the current or input voltage requirements were
increased.

If you want to stay with something close to this circuit, I would get rid of
CBUK1, add a feedback resistor from collector of QBUK2 to base of QBUK1, and
generally reduce the resistor values.  You're probably trying to keep the
resistors high to minimize wasted current, but I bet you'd pick up more
efficiency by speeding up the slew rate and keeping the FET out of the
linear region more.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, .....olinKILLspamspam.....embedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@180414 by Peter L. Peres

picon face
Try to design a standard Schmitt Trigger using 1 PNP and 1 NPN and you
should be there. Maybe a series switching element would be used and maybe
not. A secondary on the coil would help with efficiency but it is not
necessary for the power you aim for. Some simple schemes of switchers
using linear regulators were published by National in older catalogs. Very
few parts (fewer than you are using) but not 80V input. The Schmitt can be
built to take even 300V input with suitable transistors. The regulation is
as good as the Schmitt threshold, which is usually set to 25 or 50mV for
these applications. The reference is the BE junction of the 1st transistor
in the Schmitt. It can be temperature compensated easily.

I would add some sort of crowbar on the output. A 1W zener and a fuse
somewhere may be enough.

Peter

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@182418 by Dave Dilatush

picon face
Russell McMahon wrote...

>OK - here's a solution that has been suggested by a friend from the NatSemi
>LM336 regulator diode data sheet
>
>Any improvements on this?

[attachment "Lm336buck.gif" snipped]

Actually, that circuit really isn't a bad starting point for a
micropower switcher, even though it was designed for MUCH higher
output currents than we're talking about.  
I set about revising it for low-power operation (optimized for 5 volts
out at 20-50 mA) and low cost, and the result is shown below in
ASCII-art form, followed by a parts list.

I changed several things from the original design you showed:

First, I ditched the Darlington-connected power switch; it's not
necessary at low current, and it has high power losses both static and
switching.

Next, I replaced the LM336 precision voltage reference with an
ordinary 4.7 volt Zener diode.  This causes some degradation in output
voltage regulation and accuracy, but for powering digital circuits it
should still be adequate.

I also scaled the circuit resistance values to reduce bias currents,
which helps boost the efficiency at low output currents.

And finally, I added a transistor (Q3 in the diagram below) to shunt
base drive away from the main power switch (Q4) during turnoff to
increase its switching speed.  This reduces switching losses and
improves efficiency significantly.

The final design is shown below: 6 resistors, 1 inductor, one aluminum
electrolytic cap, one Zener diode and one switching diode, and four
"jelly-bean" transistors.

Parts list and design notes follow the schematic.

                           Q4                      Vin o--+------+-------+---E C--+--L1--+---o Vout
        |      |       |    B   |      |                 R1      |      R4    |   |      |                  |      E       |    |   |      |                  |   Q3  B------+    |   |      |                  |      C       |    |   |      |                  |      |       |    |   |      |                  |      +------------+   |      |                  |      |       |        |      |                  +--R2-------------------+      |                  |      |       |        |      |                  |      C       C        |      |                  +-----B  Q1 Q2  B--+-------R6--+                  |      E       E   |    |      |                  |      |       +   |    |      |                  |      +---+---+   |    |      |                  c          |       |    c      |                 D1         R3      R5   D2     C1                  a          |       |    a      |                  |          |       |    |      |           COM o--+----------+-------+----+------+---o COM                                                     C1 = 47uF 10WVDC Panasonic ECE-A1AN470U           D1 = 1N4732A Zener 4.7V 5%                          D2 = 1N4148                                       L1 = 680uH TOKO 187LY-681J                        R1 = 10K 5% (for Vin = 15V)                       R2 = 22K 5%                                       R3 = R4 = 2.2K 5%                                      R5 = 10K 5%                                       R6 = 1.0K 5%                                 Q1 = Q2 = 2N3904, 2N2222, etc.                    Q3 = Q4 = 2N3906, 2N2905, etc.                                                                    NOTES:

R1 should be chosen to give approximately 1 milliamp of bias current
for Zener diode D1: 7k ohms at Vin = 12V, 10k at 15V, 20K at 25V, and
so forth.  Insufficient Zener current will degrade regulation and
increase noise.  Excessive Zener current will just waste power.

R6 and R5 set the output voltage (in this case, slightly above D1's
Zener voltage).  The values chosen should give about 5 volts out, and
this could be trimmed if desired by making R6 a trimpot (ugh!).
Output voltage will be roughly Vout = 4.7 * (R5 + R6) / R5.

Changing catch diode D2 to a Schottky device (e.g., 1N5819 or similar)
will increase efficiency at higher power levels, due to lower forward
voltage drop.  For lowest cost, stick with a 1N4148.  Absolutely DO
NOT substitute a 1N4001 or other power rectifier meant for
mains-frequency operation: these have HUGE reverse-recovery times
which drastically reduce efficiency.

One short-cut I took in this circuit is that the hysteresis which
governs the switching frequency depends on Zener diode D1 having a few
hundred ohms of Zener impedance; the voltage divider formed by the
Zener impedance and resistor R2, along with the input voltage,
determines the hysteresis.  If a precision reference (e.g., an
LM236-5.0) is substituted for the 1N4732A for better regulation, a
resistor will have to be inserted between D1 and the base of Q1, and
R2 also connected to Q1's base.  Otherwise, the circuit will operate
at a very high frequency, and have poor efficiency.

As shown, the regulator should give somewhere around 70% efficiency
with 12 volts in and 30 mA output current, dropping to around 50% with
25 volts in.  For the transistors chosen, maximum output current is
around 50 mA.  Few of my PIC projects take more than that.

Enjoy...

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@220300 by Russell McMahon

picon face
From: "Dave Dilatush" <EraseMEdilatushspam_OUTspamTakeThisOuTHOME.COM> said
Actually, that circuit really isn't a bad starting point for a
micropower switcher, even though it was designed for MUCH higher
output currents than we're talking about.

I set about revising it for low-power operation (optimized for 5 volts
out at 20-50 mA) and low cost, and the result is shown below in
ASCII-art form, followed by a parts list.

I changed several things from the original design you showed:
.............

_______________________________

I especially like the extra transistor Q3 which acts as a high side driver.
Similar to my slightly more complex design.
Use of a long tailed pair (as per NatSemi & your circuit) gives a more
conventional and 'understandable" circuit. It would be worth building to see
how it compares in practice.
In terms of construction (number of parts, PCB area etc) its slightly worse
than mine but may be worth doing.

It's good to see the ideas that are coming out of this thread.




       Russell McMahon


________________________________________________________________

First, I ditched the Darlington-connected power switch; it's not
necessary at low current, and it has high power losses both static and
switching.

Next, I replaced the LM336 precision voltage reference with an
ordinary 4.7 volt Zener diode.  This causes some degradation in output
voltage regulation and accuracy, but for powering digital circuits it
should still be adequate.

I also scaled the circuit resistance values to reduce bias currents,
which helps boost the efficiency at low output currents.

And finally, I added a transistor (Q3 in the diagram below) to shunt
base drive away from the main power switch (Q4) during turnoff to
increase its switching speed.  This reduces switching losses and
improves efficiency significantly.

The final design is shown below: 6 resistors, 1 inductor, one aluminum
electrolytic cap, one Zener diode and one switching diode, and four
"jelly-bean" transistors.

Parts list and design notes follow the schematic.

                           Q4
 Vin o--+------+-------+---E C--+--L1--+---o Vout
        |      |       |    B   |      |
       R1      |      R4    |   |      |
        |      E       |    |   |      |
        |   Q3  B------+    |   |      |
        |      C       |    |   |      |
        |      |       |    |   |      |
        |      +------------+   |      |
        |      |       |        |      |
        +--R2-------------------+      |
        |      |       |        |      |
        |      C       C        |      |
        +-----B  Q1 Q2  B--+-------R6--+
        |      E       E   |    |      |
        |      |       +   |    |      |
        |      +---+---+   |    |      |
        c          |       |    c      |
       D1         R3      R5   D2     C1
        a          |       |    a      |
        |          |       |    |      |
 COM o--+----------+-------+----+------+---o COM

 C1 = 47uF 10WVDC Panasonic ECE-A1AN470U
 D1 = 1N4732A Zener 4.7V 5%
 D2 = 1N4148
 L1 = 680uH TOKO 187LY-681J
 R1 = 10K 5% (for Vin = 15V)
 R2 = 22K 5%
 R3 = R4 = 2.2K 5%
 R5 = 10K 5%
 R6 = 1.0K 5%
 Q1 = Q2 = 2N3904, 2N2222, etc.
 Q3 = Q4 = 2N3906, 2N2905, etc.

NOTES:

R1 should be chosen to give approximately 1 milliamp of bias current
for Zener diode D1: 7k ohms at Vin = 12V, 10k at 15V, 20K at 25V, and
so forth.  Insufficient Zener current will degrade regulation and
increase noise.  Excessive Zener current will just waste power.

R6 and R5 set the output voltage (in this case, slightly above D1's
Zener voltage).  The values chosen should give about 5 volts out, and
this could be trimmed if desired by making R6 a trimpot (ugh!).
Output voltage will be roughly Vout = 4.7 * (R5 + R6) / R5.

Changing catch diode D2 to a Schottky device (e.g., 1N5819 or similar)
will increase efficiency at higher power levels, due to lower forward
voltage drop.  For lowest cost, stick with a 1N4148.  Absolutely DO
NOT substitute a 1N4001 or other power rectifier meant for
mains-frequency operation: these have HUGE reverse-recovery times
which drastically reduce efficiency.

One short-cut I took in this circuit is that the hysteresis which
governs the switching frequency depends on Zener diode D1 having a few
hundred ohms of Zener impedance; the voltage divider formed by the
Zener impedance and resistor R2, along with the input voltage,
determines the hysteresis.  If a precision reference (e.g., an
LM236-5.0) is substituted for the 1N4732A for better regulation, a
resistor will have to be inserted between D1 and the base of Q1, and
R2 also connected to Q1's base.  Otherwise, the circuit will operate
at a very high frequency, and have poor efficiency.

As shown, the regulator should give somewhere around 70% efficiency
with 12 volts in and 30 mA output current, dropping to around 50% with
25 volts in.  For the transistors chosen, maximum output current is
around 50 mA.  Few of my PIC projects take more than that.

Enjoy...

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@220319 by Russell McMahon

picon face
part 1 1592 bytes content-type:text/plain; (decoded 7bit)

Here's the circuit as it actually got implemented.

A current limit has been added (not part of the original design) using QBCLL
and QBCLH, the 0r33 current limit resistor and supporting R's & C's.
I don't recommend doing this in most cases.
CCL1 can cause problems - it was intended to provide dynamic cycling
("hiccupping") under short circuit conditions (which it does) but
care needs to be taken to stop it pumping up under some conditions of normal
operation. Omitting it removes the hiccup protection but allows linear
limiting.
A crowbar or simple fuse will do as well. True foldback current limiting may
be worthwhile.

An extra driver has been added to the FET and this is worthwhile for higher
powers. It adds very little to the component count.
At turn off QBUK3 is now turned off by RBUK4 and provides substantially
increased low impedance turn off drive to the FET.
RBUK3 could be smaller for more on drive if desired but this is not a major
problem.

An extra "almost free" supply has been provided by adding another inductor
on the same core plus DBUK3 and CBUK. In this case it "stands on top" of the
original output but this could be an entirely isolated supply.

DBUK1 is necessary to provide 'low side drive" to the FET.
Another transistor could be used here (mirror image of QBUK3) but it added
little in practice.

CBUK2 which is shown as 470 UF was reduced to AFAIR 47K in practice.
LBUK1 was not 100 uH.
CBUK3 is elsewhere on the PCB and not part of the circuit proper.


   Russell McMahon




part 2 7337 bytes content-type:image/gif; (decode)


part 3 131 bytes
--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@220332 by Russell McMahon

picon face
From: "Jim" <jvpollspamspam_OUTDALLAS.NET>
> I'll see if I can do something with the documentation
> tomorrow (if interest still exists)

Still interested.
The multiple outputs are not too important in most cases but the core design
will add to the discussion.
I have friends who probably know of your friend if he did the LANTIRN
design.



Russell

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@220344 by Russell McMahon

picon face
From: "Byron A Jeff" <@spam@byronKILLspamspamCC.GATECH.EDU>

> 1) You talked about replacing BUKFET with a PNP bipolar. What about an NPN
> bipolar? Remove QBUK2, RBUK3, ZBUK2, and RBUK4 and tie the bottom of RBUK2
> and the collector of QBUK1 to the base of the NPN. When QBUK1 is off then
> RBUK2 will turn the power NPN on. When QBUK1 conducts, the NPN will turn
off.
> I'm sure that the base current from RBUK2 will need to be higher so that
> the power NPN will saturate.

I've paper designed something similar. The main problem is that the NPN
needs positive drive and when it is turned on the base needs to be driven by
a voltage ABOVE the input voltage. This can be accomplished by using an
extra high side supply. This can be provided by an extra isolated winding
on the inductor and a single diode and capacitor.

More exciting is the prospect of using a NChannel FET in this mode. The
extra
inductor supply behaves "interestingly" as duty cycle changes but can
certainly be used for this purpose. I have used an extra winding to derive a
second output voltage but didn't want to complicate the initial discussion.

> 2) Is RBUK3 the current limiting resistor for ZBUK2? Just making sure.

RBUK3 limits the negative going voltage swing on the gate of the FET
relative to Vin. The zener ZBUK2 clamps the gate to a safe value for the FET
allowing QBUK2 collector to swing to ground. It is possibly unnecessary for
special cases of Vin being less than the max Vgs of the FET but should
always be included in a "real design".

Designing the value of RBUK3 is interesting for a high input voltage. The
smaller it is the more drive available - esp[ecially important with a
bipolar high side switch. Lower R gives higher dissipation. It can be shown
that dissipation is proportional to duty cycle. Duty cycle decreases with
increasing input voltage - but not linearly as efficiency drops with
increasing Vin. Dissipation increases with the square of Vin. Overall the
dissipation rises with Vin but not as severely as the DC steady state would
c
ause. For very high input voltages RBUK3 may be rated to work well under
normal conditions but to fail open if oscillation ever stopped, thereby
stopping operation "fail safe". eg short the output and, in the absence of
other protection,  RBUK3 promptly burns out and turns the FET off.  I
wouldn't recommend this form of protection but it works.

> 3) What element requires the 1V headroom?

Mainly the FET used here which has a highish Rdson. P Channel FETs generally
have higher Rdson than their N channel companions and high voltage ones more
so. This one is rated at 200 volts and in practice survives at full power at
200 volts in - not something I'd recommend but it works.

Using a bipolar transistor there will improve the situation if properly
rated. There is also some voltage drop across the inductor and the feedback
circuit likes a small amount of headroom to operate (but not much).

> 4) Is the BUKFET and the inductor the only power elements? My goal is to
build
> a 5V 10A supply.

Not quite.
The flyback diode DBUK2 MUST be rated for a peak current capability of
several times the mean output current and MUST be a high speed part (eg NOT
1N400X). A BYV26 is rated at 1 amp  and a BYV28 at 3 amps. I use both of
these in this and similar applications. Worst case usually occurs around 50%
duty cycle due to peak power x time on considerations..
RBUK3 needs to be correctly rated powerwise but for normal input voltages
this is not hard.

> And my new design challenge:
>
> How could this circuit be augmented so that it could output 12V from an
> input between 6V and 15V? 15V:

This requires a fundamental change in the circuit.
At present it is a pure BUCK step down circuit and switching occurs when the
input voltage rises above the design voltage. There is therefore no need for
an oscillator - it is self switching. A boost or boost-buck design NEVER
reaches a target voltage when Vin is below Vout-design so would not self
switch as this does
A similar simple scheme using eg input current would be possible but it
would be different than this.

Your suggested boost and boost buck designs are along the right track.
In a one off situation or where the power level means moderate expense is
necessary for the power parts , then using a commercial control IC may be a
good idea.


regards


               Russell McMahon

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@220357 by Russell McMahon

picon face
From: "Olin Lathrop" <KILLspamolin_piclistKILLspamspamEMBEDINC.COM>
> Hmm.  You've tested this and found it to meet your needs, but there are a
> few points nagging me about this design.  I guess you get away with most
of
> these issues because the currents are so low.  Here are some issues:

This was the design I started with and I embellished it slightly for my
final application but it does work very well indeed as shown. Specific
component
values need to be adjusted for the particular application but its fairly
forgiving. RBUK4 is sized fro the circuit I snipped this from and could be
rather lower.

> 1  -  There is no hysteresis in comparing the output level with the
desired
> threshold voltage.

I was worried about this apparent lack for quite some while - it's why I
would never have thought the design up by myself :-).
I was quite unconvinced that hysteresis was present. However, the inductor
in the loop is what makes it work. While you could conceivably very very
very carefully adjust things so that the circuit was in a linear mode and
there was no increasing current in the inductor, in a real world I believe
it will never happen. Took me a while to become convinced however. In
practice you can raise the supply voltage as slowly and smoothly as you like
and when it reaches design point it transitions cleanly into oscillation. I
have tried VERY hard to make this circuit fault as it cannot afford to ever
do so in practice. I have so far never managed to do so.

> Although unlikely, there is no guarantee this will
> oscillate at all.  It could, in theory at least, revert to being a linear
> regulator.  A little bit of feedback from the collector of QBUK2 to the
base
> of QBUK1 would guarantee instability and also make sure the FET gate is
> being driven to the extremes.

I tried various feedforward arrangements but none improve results.

I will post the full actual circuit I am using some time soon. Main
difference is 1 more transistor for FET drive (as mentioned before -
improves waveform) and a current limit (which isn;'t needed.)


> 2  -  CBUK1 doesn't belong there.  It will make the regulator slow to turn
> on in response to a voltage drop.

Largely agree.
That was the intention.
The point was to add hysteresis by maintaining QBUK1 off longer than
otherwise but in practice it is in fact unnecessary. In practice it can be
small or even non existent.


> 3  -  The comparator is two cascaded common emitter stages with passive
> pullups.  These are going to be slow to respond, asymmetrical, and have a
> low slew rate.

Only the slow speed of the second is important as it affects FET drive
waveform.
Slow QBUK1 turnoff slows QBUK2 turn on and FET turn on. These are adequate
in practice. FET turn off is more problematical and is addressed in practice
by lower RBUK4 or a high side driver (see separate post). At low powers
these are not as important.

The circuit shown is intended for lower power levels. RBUK4 could indeed be
a lower value.
The division of voltage by RBUK2/RBUK4 will limit the lower voltage the
circuit will operate at due to the FET Vgs threshold. Using a bipolar
instead of the FET minimises this.

> 4  -  The FET will turn off very slowly because the only thing driving the
> gate is a 100K resistor.

agree

> This means the FET will spend a proportionately
> long time in the transition region.

agree

> This may be OK in your situation, but
> in general this reduces efficiency and dumps power into the FET.  This
would
> definitely be a problem if the current or input voltage requirements were
> increased.

Yes. OK for low power.
See only slightly more complex circuit for higher powers.





regards

               Russell McMahon

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\22@223009 by Jim

flavicon
face
I didn't get around to converting the 'paper' docs
into anything suitable for xmission electronically -

On the matter of LANTIRN - Ron Cleveland did the
PS and a man named John Finklea did the TWT Xmtr.

I knew Ron, had worked with John on the "MRCA" (Multi-role
Combat Aircraft) RADAR project for the Panavia Tornado
aircraft.

John was a protogee of Joe Bekerman from his days on the
MRCA Transmitter (LRU 2).  Bekerman went to a defense/space
contractor (Martin M.?) in Florida sometime in the early eighties.
Clark Dowell was MRCA program manager about that time ...

Jim

{Original Message removed}

2001\08\23@081229 by Alan B. Pearce

face picon face
Russell's design challenge is almost exactly what I need the results of.
However I think mine may have to be isolated. I have a +/-15V power rail
going to some analog stuff, and the ground rail of the PIC needs to be at
the ground level of the analogue stuff. I would prefer to have the pic power
generated from the 30V total of the supply so I am equally loading both
sides, and hence the reason the output may need to be isolated to get the
grounding correct. The input control to the PIC is to be through an isolated
I2C link.

Now for the fun part, the analogue output is allowed to have a 10khz (with
harmonics) on it, in fact in this application it is desirable). The level of
this needs to be controllable however, so it probably should not be
considered in the converter design, but may signal an operating frequency
for the converter, without needing heaps of filtering to keep the noise off
analogue supply rails.

I had been thinking in terms of using a fixed pulse width converter with a
minimum drop linear regulator, but would like to keep the current down if I
can. Does anyone have any experience with doing things like this?

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservTakeThisOuTspammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@120748 by Olin Lathrop

face picon face
part 1 3649 bytes content-type:text/plain; (decoded 7bit)

> Here's the circuit as it actually got implemented.

Since everyone is throwing their circuits out there, I've decided to give in
and throw out mine too.  This circuit is a piece of a personal project I am
currently working on and it was designed before this thread started.
However, it hasn't been built yet, which is why I didn't want to let it out
earlier.  It is also intended for much higher currents, but the topology
should scale down to lower currents nicely.

This circuit is intended to take 12-20 V unregulated DC at "+12UNREG" and
make a roughly regulated 10V out at "+10V" with the capability of 1.5A.
This is a rather straight forward self oscillating buck converter.  It is
actually similar in concept to the "divine inspiration" circuit, although
the source of this one was much more mundane common sense and some down to
earth engineering with careful calculations.  Some notes on this design:

1  -  D6 need not be a fast recovery diode.  The values of L3, C4, the
hysterisis, and the 0-1.5A current draw range make sure that L3 is always
done conducting before Q1 turns on for the next pulse.  This allows D6 to be
an ordinary power diode, which is very useful considering the currents
envolved.

2  -  Note that the *average* current thru Q1 and L3 is 1.5A max.  Since
this current will come in pulses, the peak current will be much larger.
This is an important point that the other designs seemed to have missed.  I
chose these parts to handle a peak current of 6A, since this is some margin
above the theoretical peak current.  Note that this also effects the peak
current rating of D6, because it will see the same peak current as L3.

3  -  Oscillation is guaranteed due to the hysterisis provided by R4 and R2.
This assumes that +5VREF is a zero impedance source.

4  -  In my project, +5VREF is actually coming from a 5V regulator powered
by the +12UNREG line.  This is OK since the current draw on the +5VREG line
is under 20mA.  You could generate a good enough reference with a zener,
resistor, and cap.  The cap must be large enough so that its impedance is
much less than R2 at the switching frequency.  With a custom reference, the
R6 and R5 divider could be eliminated and the output voltage fed directly
into the minus input of the comparator.

5  -  Q2 and Q3 provide the current amplification necessary to switch the
FET gate quickly.  Note that power FETs have considerable gate capacitance,
and the slew rate of most ordinary op amps or comparators would slow down
significantly if connected directly to the gate.  Any reasonalbe small
signal "junk box" transistors would do for Q2 and Q3.  The ones show happen
to be my junk box transistors I use for this purpose.  They are available
very cheaply from Jameco.

6  -  The choice of LM6132 as comparator is not optimal.  There happened to
be one of these left over from the rest of the circuit and it should be good
enough for this purpose.

7  -  The max input voltage is limited by the max allowed FET gate voltage,
which is 20V.  This could easily be extended with a resistor between the op
amp output and Q2/Q3 followed by a 12V zener clamp to the +12UNREG line.
The next limit is the max power supply allowed by the LM6132, which is 24V.
This could be extended another 10V or so by playing games with the negative
supply input, but that gets messy.  The right answer is to use a comparator
better suited for the job or make something from a few discrete transistors
that can all take the higher voltage.  Keeping the Q2/Q3 current amplifier
will be especially important then.


part 2 6967 bytes content-type:image/gif; (decode)


part 3 335 bytes

********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, spamBeGoneolinspamBeGonespamembedinc.com, http://www.embedinc.com

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email TakeThisOuTlistservEraseMEspamspam_OUTmitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@164324 by Byron A Jeff

face picon face
>
> From: "Byron A Jeff" <RemoveMEbyronspamTakeThisOuTCC.GATECH.EDU>
>
> > 1) You talked about replacing BUKFET with a PNP bipolar. What about an NPN
> > bipolar? Remove QBUK2, RBUK3, ZBUK2, and RBUK4 and tie the bottom of RBUK2
> > and the collector of QBUK1 to the base of the NPN. When QBUK1 is off then
> > RBUK2 will turn the power NPN on. When QBUK1 conducts, the NPN will turn
> off.
> > I'm sure that the base current from RBUK2 will need to be higher so that
> > the power NPN will saturate.
>
> I've paper designed something similar. The main problem is that the NPN
> needs positive drive and when it is turned on the base needs to be driven by
> a voltage ABOVE the input voltage. This can be accomplished by using an
> extra high side supply. This can be provided by an extra isolated winding
> on the inductor and a single diode and capacitor.

Confusion sets in on the computer guy ;-( . Everytime I think I actually
understand this semiconductor stuff, someone comes along and throws a curve:

The statement you make above makes sense for N-channel MOSFETs. But that's
not my understanding of how NPN bipolars work. First off I thought that
bipolars are current amplifiers, and that voltage wasn't really the determining
factor. Secondly my understanding was that a bipolar can be driven to
saturation with the base voltage well below the collector voltage. And that
in fact that once the base voltage gets above the Vbe that it doesn't in fact
continue to rise but that it sucks down more and more current through the base
while channeling more and more current through the CE junction all the way
to the point of saturation, where Ice/Ibe = hfe.

Now if this were not the case then  very ordinary things like relay drivers
where a PIC pin drives the base of an NPN and the CE junction of the NPN
switches the relay coil. For a 12V or 24V relay the base voltage never goes
above 5V.

I guess my only question out of this is whether or not it is possible to
oversaturate the base.

For now I abandoned the NPN bipolar for a much simpler reason, the 6W of heat
that would have to be dissapated if I maxed out at the 10A I'm shooting for.
That .6V drop is devastating when you're pumping 10A through...

My interest in the question is for the junkbox variety supply. My box has
in decreasing frequency: NPN bipolar, Nchannel MOSFET (usually logic level,
PNP bipolar, Pchannel MOSFET).

> > 4) Is the BUKFET and the inductor the only power elements? My goal is to
> build
> > a 5V 10A supply.
>
> Not quite.
> The flyback diode DBUK2 MUST be rated for a peak current capability of
> several times the mean output current and MUST be a high speed part (eg NOT
> 1N400X).

That's not a problem. The peak current was 30A and it's an ultra-fast
recovery diode.

{Quote hidden}

I have two goals in this arena: understanding and simplicity. Both give me the
ability to throw together a switching PS with readily available parts. As I
stated in my first message, I got the theory, but failed to grasp the control
mechanism. Now I get it.

For the sake of the following discussion I'm going to abrrev BUK with B only.

This prompts one last question: What exactly is the purpose of RB2 and QB2
again? I keep seeing QB1 and QB2 in a darlington configuration but I'm not
understanding why that config is required. Consider if we removed RB2 and QB2
and attached the collector of QB1 to the bottom of RB3. When Vin is applied
the gate of BF will rise causing it to conduct. QB1 is turned off by PB1
pulling the base of QB1 low. The coil charges until ZB1 conducts. This turns
on QB1, grounding the gate of the FET turning it off. Eventually the coil will
start to lose energy and ZB1 turns off, turning off QB1 and turning on the
FET again. So it'll oscillate right?

It seems to be the same exact operation. The only difference I could find is
that the MPSA42 (QB2) had a much higher Vce. But even that's confusing because
QB1 is going to subject to a higher Vce via RB2.

So anyway just wondering why the darlington config  and whether or not
RB2 and QB2 are absolutely necessary?

Great project. Keep up the excellent work!

BAJ

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservEraseMEspam.....mitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@190606 by Dave Dilatush

picon face
Russell McMahon wrote...

[snip]

>This design is simple to build, forgiving, well behaved, easily scaled for
>power and voltage and reasonably efficient.
>It SHOULD have use in many amateur applications.

Since you've built this circuit and are using it, I'd be interested in
seeing some performance data.

What efficiency do you actually get, at whatever operating point
you're using (Vin = _____, Vout = _____, Iin = _____, Iout = _____)?
What frequency does it oscillate at?  What are the rise and fall times
of the voltage waveforms on BUKFET's gate and drain?

>This circuit is special - it was provided by God (no kidding).

Watch out, there!  One always needs to be aware that God has a sense
of humor, and one of His more subtle pranks was to create Zener diodes
with weak knees.

With the resistor values shown, your circuit operates Zener diode
ZBUK1 with about 60 microamperes of reverse current; while this may be
enough for some diodes when operated at room temperature, it is
definitely NOT enough for ALL diodes over any appreciable temperature
range.

Zener diodes frequently have a rather gradual transition between the
non-conducting state (at voltages well below Vz) and reverse
conduction beginning at voltages approaching Vz.  They generally
should not be operated in this "knee" region because they can be very
noisy, have wretched temperature coefficients, and appear to have
breakdown voltages that are WAY out of spec.

A good rule of thumb is that a Zener diode usually needs at least one
milliampere of reverse current in order to reliably "do its thing".  
With only 60 microamps of Ir, you could run into troubles with this
circuit- poor regulation, large changes in output with temperature,
noisy operation, etc.- that show up in production, particularly when a
new batch of diodes arrives.

I tried simulating this circuit in SPICE; it works, after a fashion,
but it appears to operate more like a linear regulator that's decided
it would be fun to oscillate a little, than a true switching
regulator.  Others have pointed out the problem caused by the weak
turn on/turn off drive to BUKFET, and it's a real killer: in
simulation, at least, I get an efficiency that's not appreciably
greater than that of a linear regulator, mostly because of BUKFET's
long, long, LONG turnoff time.  You need sharp on/off transitions to
avoid wasting power, and that means solid gate drive.

Part of the problem is BUKFET itself; the IRF6215 you've chosen for
BUKFET is a real monster of a FET, with a huge gate charge
requirement.  If this design is primarily for low-current use (a few
tens of milliamps output), you could use something like a ZVP3310 from
Zetex which would switch a lot faster.  Total gate charge for this
device is about 0.8 nC, compared with 60+ nC for the IRF6215.

Got any measurement results to share?  I'd be interested in seeing
them, particularly if they summarize results for a number of units.

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email EraseMElistservspammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@190622 by Dave Dilatush

picon face
Olin Lathrop wrote...

>Since everyone is throwing their circuits out there, I've decided to give in
>and throw out mine too.  This circuit is a piece of a personal project I am
>currently working on and it was designed before this thread started.
>However, it hasn't been built yet, which is why I didn't want to let it out
>earlier.

[snip]

I think the feedback sense is reversed from what it should be: as is,
this circuit will latch up with Q1 either on or off, and stay there.

By deleting R2, connecting +5VREF to the (-) opamp input, and
connecting the junction of R5 and R6 to the (+) opamp input and R4,
the regulator works fine (at least in SPICE).  With 20V in and 10V out
at one amp, efficiency looks to be about 80%.  Output ripple should be
about 850 mVp-p at this load, with an oscillation frequency of 6
kilohertz.  With input voltage reduced to 13 volts, efficiency goes up
to nearly 90%.

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservEraseMEspamEraseMEmitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@201122 by Russell McMahon

picon face
> > > 1) You talked about replacing BUKFET with a PNP bipolar. What about an
NPN
> > > bipolar? Remove QBUK2, RBUK3, ZBUK2, and RBUK4 and tie the bottom of
RBUK2
> > > and the collector of QBUK1 to the base of the NPN. When QBUK1 is off
then
> > > RBUK2 will turn the power NPN on. When QBUK1 conducts, the NPN will
turn
> > off.
> > > I'm sure that the base current from RBUK2 will need to be higher so
that
> > > the power NPN will saturate.
> >
> > I've paper designed something similar. The main problem is that the NPN
> > needs positive drive and when it is turned on the base needs to be
driven by
> > a voltage ABOVE the input voltage. This can be accomplished by using an
> > extra high side supply. This can be provided by an extra isolated
winding
> > on the inductor and a single diode and capacitor.
>
> Confusion sets in on the computer guy ;-( . Everytime I think I actually
> understand this semiconductor stuff, someone comes along and throws a
curve:
>
> The statement you make above makes sense for N-channel MOSFETs. But that's
> not my understanding of how NPN bipolars work. First off I thought that
> bipolars are current amplifiers, and that voltage wasn't really the
determining
> factor. Secondly my understanding was that a bipolar can be driven to
> saturation with the base voltage well below the collector voltage. And
that
> in fact that once the base voltage gets above the Vbe that it doesn't in
fact
> continue to rise but that it sucks down more and more current through the
base
> while channeling more and more current through the CE junction all the way
> to the point of saturation, where Ice/Ibe = hfe.


Everything you say is essentially correct as it applies to other circuits !
:-)
Yyou just need to think through what happens in this case when an NPN in the
place of the BUKFET is turned hard on so that its emitter is pulled all the
way up to Vin. What will the base voltage now be relative to Vin? Think
about it first and then read the following if needs be.

[[[ The confusion arises here because of what happens when the main pass
transistor (BUKFET  replaced by an NPN) turns on.
For an ideal transistor the drop across the transistor Vce will be zero. For
a real transistor this will be in the 0.1 volt to 0.3 volt range. Much
higher than that and you probably need a different transistor for the
current levels you are using. The result is that the emitter is now
*essentially* pulled all the way up to Vin.  To keep the transistor turned
on the base needs to be about 0.6 volts above the emitter so it will need to
be about 0.6 volts ABOVE Vin. You are correct that a bipolar transistor
TENDS to maintain its Vbe at about 0.6 volts once turned on and just "suck
increasing current" as attempts are made to increase this voltage. It's
actually an exponential relationship but the curve gets noticeably steep at
about this pint and by the time you get to 0.7 volts you would normally be
overdriving the device beyond its ratings. BUT, if the source which is
providing this Vbe is a voltage source we normally turn it into something
approximating a current source by placing a resistor between the turn on
voltage and the base. This means we need more than 0.6 volts to cause the
required current to flow i the resistor. The larger the driving voltage the
larger the resistor for the desired current and the closer this approaches a
true current source which will provide "about" the desired current as Vbe
moves up its exponential curve. Typically we allow at least a few volts to
do this. Say we want 1 mA base drive and we have a 5 volt source then the
resistor required is R = V/TI= (5-0.6)/0.001 = 4600 ohms. If the ACTUAL base
voltage rose to 0.8 volts the current would be V/R = (5-0.8)/4600 = 0.913
mA. Still OK probably.

The result of all this is, when the emitter is pulled to Vin the requirted
base drive voltage required is several volts above Vin to apply enough
voltate to the base  resistor to get the 0.6 volts odd Vbe required. For
this we need a :"high side supply". The xtra winding that I show in my more
complete diagram can be used to supply this at relatively little cost. (One
winding, 1 diode, 1 resistor, 1 capacitor).

> I guess my only question out of this is whether or not it is possible to
> oversaturate the base.

You can damage the transistor by providing too much base current. This is
seldom a problem in practice in low power circuits.
A saturated transistor takes longer to turn off and some high speed
switching designs use reverse schottky clamp diodes from C to B to prevent
the transistor fully saturating. This is not applicable here.

> For now I abandoned the NPN bipolar for a much simpler reason, the 6W of
heat
> that would have to be dissapated if I maxed out at the 10A I'm shooting
for.
> That .6V drop is devastating when you're pumping 10A through...

You SHOULD be able to get Vsat down around 0.1 volts with a suitable
transitor. At those current levels you need to make sure to study the spec
sheet carefully. Many devices are specified for current gains at say 1 amp
and have markedly less gain at higher currents. It can be informative to put
your transistor on a heat sink, apply different base currents (a few volts
supply through a variable resistor) and then to apply the desired collector
current (second supply and resistor or variable supply). My experience is
that Vsat can get worse quite suddenly as base drive is reduced, that actual
current gain (beta) can be quite different than that at specified current
levels (not surprisingly) and that there can be wide variations for the same
type. Intelligent use of published data and an genetraous amount of base
overdrive will usually help. The essentially zero DC drive requirement of a
FET makes it much easier to drive in high current situations - (but you
still need to provide a means of rapidly moving charge in and out of the
gate if you want fast switching.) For a FET a drop of 0.1 volts at 10 A
corresponds to an Rdson of 0.010 ohm. - easily enough cahieved in lower
voltage FETS (and for $ in higher voltage FETS).

> My interest in the question is for the junkbox variety supply. My box has
> in decreasing frequency: NPN bipolar, Nchannel MOSFET (usually logic
level,
> PNP bipolar, Pchannel MOSFET).

**** NB NB NB NB NB NB NB  ***

I said previously that -

> > The flyback diode DBUK2 MUST be rated for a peak current capability of
> > several times the mean output current and MUST be a high speed part (eg
NOT
> > 1N400X).

Olin has suggested that this is NOT the case due to circuit time constants
etc. I'm not convinced yet but his assertion needs looking at,. The ability
to use a low cost general purpose diode here would be an advantage.

> I have two goals in this arena: understanding and simplicity. Both give me
the
> ability to throw together a switching PS with readily available parts. As
I
> stated in my first message, I got the theory, but failed to grasp the
control
> mechanism. Now I get it.
>
> For the sake of the following discussion I'm going to abrrev BUK with B
only.
>
> This prompts one last question: What exactly is the purpose of RB2 and QB2
> again? I keep seeing QB1 and QB2 in a darlington configuration but I'm not
> understanding why that config is required. Consider if we removed RB2 and
QB2
> and attached the collector of QB1 to the bottom of RB3. When Vin is
applied
> the gate of BF will rise causing it to conduct. QB1 is turned off by PB1
> pulling the base of QB1 low. The coil charges until ZB1 conducts. This
turns
> on QB1, grounding the gate of the FET turning it off. Eventually the coil
will
> start to lose energy and ZB1 turns off, turning off QB1 and turning on the
> FET again. So it'll oscillate right?
>
> It seems to be the same exact operation. The only difference I could find
is
> that the MPSA42 (QB2) had a much higher Vce. But even that's confusing
because
> QB1 is going to subject to a higher Vce via RB2.
>
> So anyway just wondering why the darlington config  and whether or not
> RB2 and QB2 are absolutely necessary?

Summary: It's NOT  a darlington - look again and you'll see ot's a little
unusual. and look at the drive polraity required by the pass FET/transistor.

Detail.     RB2 is used to provide turn on base drive to QB2. At startup
there is no output voltage and this is the only way for the rqgulator to
start. In my more complex circuit you will see that I have also added RBUKX
from Vout to the base of QB2. This allows Vout to provide some/most of
thebase drive for QB2 and allows RB2 to be a higher resistance and
therefiore limits dissipation for high Vin (recall that in my case this runs
up to 200 volts Vin).

QB2/QB1 connection: In a darlington the emitter of the first transistor
drives the base of the second. Collector current in the first transistor
becomes the base current of the second transistor. In this case quite the
opposite occurs. QB1 SHUNTS the base drive to QB2. QB2 is normally turned on
by resistor RB2 and turning QB1 on shunts this drive current to ground
thereby turning QB2 off.

Now look at the relative polarities required.
I am using a P Channel FET whose gate must be pulled "low" to turn on.
For Vin too high, QB1 is on and QB2 is off and FET gate is high and FET is
off.
If we changed to your QB1 only design the we get Vin too high, QB1 on, FET
gate low, FET on and ..... oh dear.
HOWEVER if the FET was an N Channel then this would work as desired.
However, we would then need to drive thre FET gate ABOVE Vin during FET
turnon and so need a high side supply. (as when on FET drain and source are
connected so source is at Vin so gate must be at leat Vthreshold above Vin
(plus a bit more).
Using a second winding would provide such a supply).

A note:

       My application demanded, among other things,  very low cost in
volume production and it met this requirement superbly.  Power level was
moderate (up to 1 amp out at 10 volts).  For the low power levels I
specified in my original description for this "challenge" then the circuit
shown is appropriate. For a low volume high current design I would CONSIDER
using a more conventional circuit such as the one Olin just published or one
using a commercial IC. These usually have a more controlled drive of the FET
gate. Adding the extra transistor in my more complex circuit improves FET
turn off markedly. Replacing the diode DBUK1 with a PNP transistor (relative
to QB3 connect emitter to emitter, base to base, collector to ground ) adds
surprisingly little improvement but ends up looking much like Olin's drive
circuit. I trialled this arrangement during development and decided it was
not needed.
.
Hope all that makes sense.



regards

                   Russell McMahon

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservspam_OUTspamKILLspammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@201134 by Russell McMahon

picon face
> Since everyone is throwing their circuits out there, I've decided to give
in
> and throw out mine too.  This circuit is a piece of a personal project I
am
> currently working on and it was designed before this thread started.
> However, it hasn't been built yet, which is why I didn't want to let it
out
> earlier.  It is also intended for much higher currents, but the topology
> should scale down to lower currents nicely.

Looks good. Fairly conventional (which is not bad). It has limitations for
my original application (eg voltage and less than spartan parts cost) but
should be a fine moderate cost design in many cases.

There is one major problem with the FET used as shown. This is an N-Channel
device but its gate is only driven as high as its drain during turn on. This
means that there will be a drain-source drop of at least Vthreshhold (2 to 4
volts for the IRF5305) leading to very high dissipation.
That's a VERY nice FET you have chosen - it would be a shame to not use it.
This could be achieved by adding a winding on the inductor to provide a
small high side supply . Alternatively changing to a P Channel FET and
changing the drive sense.

The IRF5305 is somewhat dear - Digikey charge $US6.24 in 1's.
Something like an IRF640 (Digikey charge $US1.48 in 1's) would probably
suffice at your proposed current levels.
The Rdson is 0.18 ohms so the dissipation would be much worse than for the
IRF5305. Still OK at your current levels probably.
This is also an N Channel device

You could consider using a P Channel FET.
This would allow your circuit to be used with minimal changes
The IRF6215 P Channel FET that I show in my circuit is sold by Digikey for
$US1.51 in 1's.
The Rdson is 0.3 ohms but at a mean current of 1.5A that's not too bad.
(I'm actually using an ON Semi 200 volt P FET in my final application with
an even higher Rdson but the circuit is unchanged).

_______________

Consider placing a small series current limiting resistor between the drive
transistor emitters and FET gate to limit FET switching times. You can get
immense currents here and FET losses can be increased by the too rapid
switching (yes - you can get too much of a good thing :-) ).

> 1  -  D6 need not be a fast recovery diode.  The values of L3, C4, the
> hysterisis, and the 0-1.5A current draw range make sure that L3 is always
> done conducting before Q1 turns on for the next pulse.  This allows D6 to
be
> an ordinary power diode, which is very useful considering the currents
> envolved.

This is an interesting claim and one I will think further about. I don't
really believe it but I hope you are cirrect as the saving in diode cost can
be significant.
Buck convertersa re usually best run in continuous current mode and this
converter has the option of being run either way.  In continuous mode
switching occurs while inductor current is still flowing so D6 has to handle
Q1 switching on.

> 2  -  Note that the *average* current thru Q1 and L3 is 1.5A max.  Since
> this current will come in pulses, the peak current will be much larger.
> This is an important point that the other designs seemed to have missed.

I agree with that requirement
In a prior post I noted

   > > The flyback diode DBUK2 MUST be rated for a peak current capability
of
   > > several times the mean output current and MUST be a high speed part
(eg NOT
   > > 1N400X).

However, the mean diode current and peak diode currents can be quite
different. Worst case is typically around 50% duty cycle. Especially in
discintinuous current mode, as duity cycle drops (increasing Vin) the peak
current rises but occurs for less time.

> chose these parts to handle a peak current of 6A, since this is some
margin
> above the theoretical peak current.  Note that this also effects the peak
> current rating of D6, because it will see the same peak current as L3.

> 3  -  Oscillation is guaranteed due to the hysterisis provided by R4 and
R2.
> This assumes that +5VREF is a zero impedance source.

Yes - this has a much more "comfortable" feeling tahn in "my" circuit as one
can trace a step by step DC path for the switching.  In "my" design the
inductor forms an essential part of the hysteresis mechanism. Your circuit
would oscillate without the inductor producing a PWM regulator using the FET
in resistance as trhe contro element (you wouldn't of course do this but it
would work after a fashion).

> 4  -  In my project, +5VREF is actually coming from a 5V regulator powered
> by the +12UNREG line.  This is OK since the current draw on the +5VREG
line
> is under 20mA.  You could generate a good enough reference with a zener,
> resistor, and cap.  The cap must be large enough so that its impedance is
> much less than R2 at the switching frequency.  With a custom reference,
the
> R6 and R5 divider could be eliminated and the output voltage fed directly
> into the minus input of the comparator.

This provide more flexibility than in my circuit where Vout is set by a
zener. The NatSemi cct with longtailed pair also provides this flexibility.
Mine is of cours echeaper and simpler (whetehr that makes it better or not
depends on application).

> 5  -  Q2 and Q3 provide the current amplification necessary to switch the
> FET gate quickly.  Note that power FETs have considerable gate
capacitance,
> and the slew rate of most ordinary op amps or comparators would slow down
> significantly if connected directly to the gate.  Any reasonalbe small
> signal "junk box" transistors would do for Q2 and Q3.  The ones show
happen
> to be my junk box transistors I use for this purpose.  They are available
> very cheaply from Jameco.

Agree. I typically use BC337 & BC327 in this application. They are commonly
available and have better specs in this application than most other TO92
transistors while often costing less. (500 mA rated, Beta of 300 typically).

> 7  -  The max input voltage is limited by the max allowed FET gate
voltage,
> which is 20V.  This could easily be extended with a resistor between the
op
> amp output and Q2/Q3 followed by a 12V zener clamp to the +12UNREG line.
> The next limit is the max power supply allowed by the LM6132, which is
24V.
> This could be extended another 10V or so by playing games with the
negative
> supply input, but that gets messy.  The right answer is to use a
comparator
> better suited for the job or make something from a few discrete
transistors
> that can all take the higher voltage.  Keeping the Q2/Q3 current amplifier
> will be especially important then.

A possible solution is to use a resistor and zener from Vin to get the
circuit started and then to feed the comparator supply from Vout or, more
probably, from a secondary winding to give you enough voltage


regards

           Russell McMahon

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservTakeThisOuTspamspammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\23@225536 by Byron A Jeff

face picon face
On Fri, Aug 24, 2001 at 11:14:45AM +1200, Russell McMahon wrote:

[BAJ]
>>>>1) You talked about replacing BUKFET with a PNP bipolar. What about an NPN
>>>>bipolar?

[Russell]
>>>I've paper designed something similar. The main problem is that the NPN
>>>needs positive drive...

[BAJ Again]
{Quote hidden}

[Russell]
{Quote hidden}

Understood. And if we only pull the base up to Vin, then we'll not get the
emitter very close to Vin. In fact it'll end up several volts below.

>
>>I guess my only question out of this is whether or not it is possible to
>>oversaturate the base.
>
>You can damage the transistor by providing too much base current. This is
>seldom a problem in practice in low power circuits.
>A saturated transistor takes longer to turn off and some high speed
>switching designs use reverse schottky clamp diodes from C to B to prevent
>the transistor fully saturating. This is not applicable here.

I was asking this question in the context of the 20 to 1 input voltage
Since the base is pulled above Vin and some base resistor will current limit
how do you design so that you don't fry the transistor and yet provide enough
base current to get to saturation at lower Vins.

[Edited bipolar selection process...]

{Quote hidden}

Since I've been dabbling in switching power supplies recently, high current,
high power, high speed rectifiers are actually in my junkbox. I'm fortunate
to be in Atlanta and there are several places that stock the required parts.

>
>Olin has suggested that this is NOT the case due to circuit time constants
>etc. I'm not convinced yet but his assertion needs looking at,. The ability
>to use a low cost general purpose diode here would be an advantage.

That's true. But I have a properly specified part in my prototype.

{Quote hidden}

Duh. It took me a while to see that absolutely everything is reversed from
the N-channels that I'm used to using. Source is positive, Drain is negative,
Gate must be more negative than drain to get the part to conduct.

I hooked the P-channel up backwards and I didn't get that QB2 acts as a sort
of inverter.

>
>Detail.     RB2 is used to provide turn on base drive to QB2.

And since QB2 is an NPN, it needs a pullup resistor to conduct.

> At startup
>there is no output voltage and this is the only way for the rqgulator to
>start. In my more complex circuit you will see that I have also added RBUKX
>from Vout to the base of QB2. This allows Vout to provide some/most of
>thebase drive for QB2 and allows RB2 to be a higher resistance and
>therefiore limits dissipation for high Vin (recall that in my case this runs
>up to 200 volts Vin).

But you still need the startup resistor RB2 right? I see.

>
>QB2/QB1 connection: In a darlington the emitter of the first transistor
>drives the base of the second. Collector current in the first transistor
>becomes the base current of the second transistor. In this case quite the
>opposite occurs. QB1 SHUNTS the base drive to QB2. QB2 is normally turned on
>by resistor RB2 and turning QB1 on shunts this drive current to ground
>thereby turning QB2 off.

Got it. And all of this is a prelude to shunting the base voltage generated by
the pullup RB4.

I also think that I figured out that RB3 is a current limiting resistor for
high Vin. I'm testing with 12-15V Vins. I left it off in my prototype.

>
>Now look at the relative polarities required.
>I am using a P Channel FET whose gate must be pulled "low" to turn on.
>For Vin too high, QB1 is on and QB2 is off and FET gate is high and FET is
>off.

And the coil is feeding power to the load. This is the buck phase.

>If we changed to your QB1 only design the we get Vin too high, QB1 on, FET
>gate low, FET on and ..... oh dear.

An avalanche. This didn't occur in my testing because I had the FET wired
backwards. I added QB2 and RB2 before I realized that I needed to reverse
the Source and Drain. Maybe you should label for us clueless ones out here. ;-)

>HOWEVER if the FET was an N Channel then this would work as desired.
>However, we would then need to drive thre FET gate ABOVE Vin during FET
>turnon and so need a high side supply. (as when on FET drain and source are
>connected so source is at Vin so gate must be at leat Vthreshold above Vin
>(plus a bit more).

The bit more is based on the current passing through the device. I saw an
equation like Rds*Ids...

>Using a second winding would provide such a supply).

I'm happy with the P-channel.

{Quote hidden}

Noted. Now for my report.

After reading (and gaining some understanding) I rushed out and bought parts.
I have an IRF9540 P-channel (Ids=19A, Vds=100V, rDS=0.2 ohms), a 10 microhenry
11A coil, a 330 uF cap, and a SB580 5A ultra fast schottky catch diode.
Since I wasn't concerned with high Vin, I used ordinary 2N2222 for both
transitors. I also dropped the values of all of the resistors since I'm not
worried about a high Vin. My concern was getting a bit better snap off on the
FET. I bought a handful of different types of zeners, but ended up using a 5V
one. I'm still trying to get clued in on why the output voltage isn't 5.6V
as you specified but 5.0V right on the mark.

I'm pleased to report that it was a total success. Once I added RB2/QB2 and
got the FET leads straight, Vout is right at the 5V mark.

I did some preliminary high current testing. Built a dummy load out of 4
0.7 ohm 25W ceramics, wired in series/parallel to give a 0.7 ohm 100W load.
Final result: Vout=4.75V @ 7.8A !!! A total success.

Next up is building a similar boost regulator to get 12V out even with a 6.5V
input. I may even try wrapping that second coil so that I can get a low current
-12V.

>.
>Hope all that makes sense.

It does. I've been trying to get switchers under my belt for a while. Your
design and this thread has taken me several light years further than where
I was. I really appreciate it.

BAJ

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email EraseMElistservspamspamspamBeGonemitvma.mit.edu with SET PICList DIGEST in the body


2001\08\24@014012 by Bob

flavicon
face
How about a VB408, a 1% resistor, a precision zener (or more if you want
selectable voltages), and a small low voltage cap?  Output can be variable
(replace zener with a pot) from 1.25VDC to 30VDC.  Input can be up to 400 VDC.
Okay, it's not an smps, but it is inexpensive, and should meet the critiria
originally set forth on this thread.

--
http://www.piclist.com hint: To leave the PICList
RemoveMEpiclist-unsubscribe-requestKILLspamspammitvma.mit.edu


2001\08\24@042410 by Russell McMahon

picon face
>This design is simple to build, forgiving, well behaved, easily scaled for
>power and voltage and reasonably efficient.
>It SHOULD have use in many amateur applications.

At the start of this I should indicate for which parts of the circuit I
claim divine input. God doesn't need anyone to stick up for him but I should
point out where my responsibilities begin :-). The key "breakthrough" was
the use of the transistor "comparator" and "zener based voltage reference" .
I had been presented with a prior design which had used a standard
microprocessor reset generator IC as a switching regulator controller and I
had to produce one which was cheaper and very preferably functionally
superior in this application. So, the core transistor based "controller"
concept using (in this implementation) two transistors and no conventional
hysteresis control were the special input. How I implement and or extend the
design is my responsibility (to the extent that anything is).

I,  like Olin, was somewhat dubious that the circuit would work (let alone
work reliably) and was overjoyed to find out how well it did. Working out
just why came next :-)

/Since you've built this circuit and are using it, I'd be interested in
/seeing some performance data.

/What efficiency do you actually get, at whatever operating point
/you're using (Vin = _____, Vout = _____, Iin = _____, Iout = _____)?

I can give you rough answers from memory. For details I'll have a dig in my
design log or possibly it may be better to whip up a low power one as per
this
particluar requirement and measure them again.

/What frequency does it oscillate at?  What are the rise and fall times
/of the voltage waveforms on BUKFET's gate and drain?

Now you're asking useful questions :-).
The system is self oscillating and actual frequency is quite variable and
dependent on a number of factors.
The on time is a function of the time that it takes for the on state voltage
to rise to the trip point and the off time is a function of how long the
stored energy in the inductor will maintain the output voltage ABOVE the
trip point. Note I say 'a function of" - these are not the only factors.
Someone (Olin?) said that CBUK1 was unnecessary (actually he said it was in
the wrong place). This ws my attempt to control the minimum off time by
holding QB1 on when the output voltage had fallen below the transistor's
normal turn on point.
So, changing the inductance affects frequency significantly. Also CBUK2.
Load is also a major factor - more load = higher frequency.
Typically oscillation frequency is very low at very low load - under 10 KHz.
This is entirely expected as it is switching "on demand" and if there is no
demand it doesn't switch !. Adding a small permanment load increases Fmin.
Loading it up takes the frequency up rapidly. Depending on various
parameters this will be up to around 100 KHz or so.

I have basic spectrum analysis facilities available but my experience is
that a VERY good initial idea of emc problems can be had with the use of a
standard AM/FM radio. This circuit is about as quiet as a switcher gets. Not
perfect - a radio placed right on top of the coil certainly picks up signal
but the radiated harmonics are minimal.

FET drive waveforms are not nice. They are, rather to my surprise, much
nocer than for the system using the reset IC.
In my fulle rcircuit (posted ?2 days ago) I show an added FET gate turnoff
transistor. This very substantially improves the turnoff wavshape and time.
A similarly connected transistor for turn on produced minimal if any
improvement in actual performance.

Peak efficiency occurs for smaller values of Vin and is arounf 85%+ AFAIR.
This is not as good as some very optimised designs but entirly adequate in
many cases where this circuit would be useful. Having a 1 ohm Rdson FEt in
my case does not help!. eg at 10 volts out and 500 mA Pout = 5 watts and FET
Rdson loss would be 250 mW or 5% before any other losses are considered,
.
Efficiency drops off with increasing Vin (a major factor being that more of
the output is provided by flyback and less by forward conduction) and it
falls to around 55% AFAIR at 200 volts in. In my applicatio the 200 volts is
an extreme condition - normal values are in the 20 to 50 volt range but it
MUST be able to survive the occasional extreme excursion and work there.

For a system with less extreme max to min values a better result could be
obtained.

The PFET that I use has an Rdson of about 1 ohm - this is truly horrible
and is a significant contributor to losses. It was chosen because it
provided a
good mix of cost and performance in my application. 200 volt plus PFETS are
not very common!

>This circuit is special - it was provided by God (no kidding).

/Watch out, there!  One always needs to be aware that God has a sense
/of humor, and one of His more subtle pranks was to create Zener diodes/
/with weak knees.

Yes. I am well aware of the soft turn on of zeners.
The somewhat ill defined voltage characteristic in this case was very
tolerable because of the cost of a zener compared to a referenc ediode.

/With the resistor values shown, your circuit operates Zener diode
/ZBUK1 with about 60 microamperes of reverse current; while this may be
/enough for some diodes when operated at room temperature, it is
/definitely NOT enough for ALL diodes over any appreciable temperature
/range.
/Zener diodes frequently have a rather gradual transition between the
/non-conducting state (at voltages well below Vz) and reverse
/conduction beginning at voltages approaching Vz.  They generally
/should not be operated in this "knee" region because they can be very
/noisy, have wretched temperature coefficients, and appear to have
/breakdown voltages that are WAY out of spec.

I largely agree with these comments. I suggest that rather than having a
gradual transition what we are seeing is an exponential rise and moving the
curve onlyt a little either way on one axis can have a major influence on
the other - in this case voltage drop versus current.

I agree that the reproducibility of voltage over various bartches of zeners
and gereral zener characteristic swould be improved by increasing zener
current. This is difficult to do in this particular layout because of the
wway that the zener is used here which is quite different to when it is used
as a "reference' as in eg the Nat Semi circuit. This circuit could arguably
be 'improved" by substantially decreasing RBUK1 and PBUK1 to increase zenera
current. The extra power loss would be insigniicant. Notice that RB!/PB1 are
NOT intended to form a significant voltage divider (PB1>> RB1) - PB1 is
there to ensure QB1 is off in the absence of zenere conduction. In practice
PB1 could PROBABLY be ommitted but this would be risky.

The zener behaviour is better than might be expected because it is driven
from a very limited swing input voltage - Vout varies little over all
operating conditions.

/A good rule of thumb is that a Zener diode usually needs at least one
/milliampere of reverse current in order to reliably "do its thing".
/With only 60 microamps of Ir, you could run into troubles with this
/circuit- poor regulation, large changes in output with temperature,
/noisy operation, etc.- that show up in production, particularly when a
/new batch of diodes arrives.

Yes. Where it was reasonably possible to do so I would use a higher current
than this. Can you suggest a reasonable way of doing so here (apart from
reducing RB1 & PB! markedly)? There will be ways to do this but they are
liable to increase complexity and cost .
And, the circuit works very well for its purpose.
Moderate changes in operating point over temperature, load, Vin and time are
tolerable.


/I tried simulating this circuit in SPICE; it works, after a fashion,
/but it appears to operate more like a linear regulator that's decided
/it would be fun to oscillate a little, than a true switching
/regulator.

In practice it is VERY CERTAIN about what it is.
No matter how hard I tried to "fool" it (and I tried very hard indeed) it
always burst into full life cleanly at a certain point. One moment it is a
linear regulator and the next it is definitely a switcher.

Others have pointed out the problem caused by the weak
turn on/turn off drive to BUKFET

The resistor values were not meant to be applicable for all applications - I
should/could have reworked them for the actual challenge (should!) but they
actually come from the working 200 volt circuit).
By all means change the drive and turn off resistors. I was sloppy when I
snipped the simple circuit out of a larger diagram. RBUK4 and RBUK3 can be
much lower for a low voltage circuit., Also RBUK4 was actually being used to
drive a bipolar gate driver and not the FET directly.
For a say 30 volt maximum in circuit RB3 for 1/4 watt max dissipation could
be as low as V^2/P = 900/.25 = 3500r - say 3K9.
This would be OK for eg a 12 to 24 volt system with peak occasional
excursions to 30 Volt in and an SFR16 resistor as RB3 (Philips - 0.5 watt
rated).
In practice RB3 actually dissipates FAR less than its static DC value (due
to duty cycle aspects) and it also divides full Vin with RB4 so you could
use an evenm lower value in this case. probably eg RB3 = 1K, RB4 = 4K7 or
similar.
That would help immensely.

/, and it's a real killer: in
/simulation, at least, I get an efficiency that's not appreciably
/greater than that of a linear regulator, mostly because of BUKFET's
/long, long, LONG turnoff time.  You need sharp on/off transitions to
/avoid wasting power, and that means solid gate drive.

My real results AFAIK are say 85% at Vin = 15 and 55% say at Vin = 200 v for
a 10 volts out say. at 600 mA.
(The system works to 1 amp but peak real load was 600 mA and is now under
half that after I redesigned some peripheral equipment for another supplier
:-)).

A linear regulator would be 66% and 5% efficient respectively.

The buck dissipates 1 watt and 5 watts respectively.
A linear regulator would dissipate 3 watts and 114 watts respectively.
(!!!!)

In my application the switcher wins quite well.

The 200 volt in condition is very very very very rare - but it can happen.in
theory for very very short periods.
(This is used in excercise equipment and if Arnold S ever decided to set the
machine on Level 0 and pedal at super super lightning speed then this might
happen for a brief period.)

/Part of the problem is BUKFET itself; the IRF6215 you've chosen for
/BUKFET is a real monster of a FET, with a huge gate charge
/requirement.

Absolutely !!
Also rather high voltage for a PFET. I am actually using an ON Semi part
rated at 200 volts which is even worse !
I don't recommend this FET unless you need the very high voltage.
I am keen to extend this circuit to NFET use at some stage (SHOULD be easy)
as there are many more much better much cheaper NFETs available.
BUT, this one works well for the purpose.

/If this design is primarily for low-current use (a few
/tens of milliamps output), you could use something like a ZVP3310 from
/Zetex which would switch a lot faster.  Total gate charge for this
/device is about 0.8 nC, compared with 60+ nC for the IRF6215.

Agree entirley.
Or use a PNP bipolar.

/Got any measurement results to share?  I'd be interested in seeing
/them, particularly if they summarize results for a number of units.

Anon.
But even better would be if someone el;se gives results for versions of this
which are closer to more normal PICLIST use.
eg Byron Jeff has built a 10A 5 volt unit - it will be interesting to hear
actual results.
He says -

> I did some preliminary high current testing. Built a dummy load out of 4
> 0.7 ohm 25W ceramics, wired in series/parallel to give a 0.7 ohm 100W
load.
> Final result: Vout=4.75V @ 7.8A !!! A total success.

I notic ehe said it had Vout = 5 volts at no load so this was 1 0.3 volt
regulation drop at 8 amps. We'd have to know where he measured this to know
how well it was actually regulating.




regards


               Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
piclist-unsubscribe-requestSTOPspamspamspam_OUTmitvma.mit.edu


2001\08\24@042440 by Russell McMahon

picon face
Byron,

Could you publish some performance figures for your version of this design
which indicate eg efficiency and regulation.
Measuring Vin, In , Vout and load for a few values of load would be
interesting. Also frequency of operation at various loads.
Also note that Vout would best be measured between near the cathode of zener
ZBUK1 and regulators own ground as at the currents you are using lead drop
can be significant.


> Understood. And if we only pull the base up to Vin, then we'll not get the
> emitter very close to Vin. In fact it'll end up several volts below.

Yes

> >>I guess my only question out of this is whether or not it is possible to
> >>oversaturate the base.
> >
> >You can damage the transistor by providing too much base current. This is
> >seldom a problem in practice in low power circuits.
> >A saturated transistor takes longer to turn off and some high speed
> >switching designs use reverse schottky clamp diodes from C to B to
prevent
> >the transistor fully saturating. This is not applicable here.
>
> I was asking this question in the context of the 20 to 1 input voltage
> Since the base is pulled above Vin and some base resistor will current
limit
> how do you design so that you don't fry the transistor and yet provide
enough
> base current to get to saturation at lower Vins.

Same problem in designing RBUK3 and RBUK4.
Short of providing a second switching regulator to provide a constant drive
voltage to the high side driver (oh no!!!) you just have to design for
enough drive at worst case and enough dissipation at highest Vin. With a FET
this is not hard as the drive current requirements are minimal. With a
bipolar itr can be annoying.

There is in fact an interesting characteristic of the secondary buck coil
volatge which makes this easier. As Vin increases the average voltage of the
FORWARD pulses tends to stay constant as their peak voltage increases but
their duty cycle (of course) decreases. careful rectifier and smoothing
design so you get the average voltage (and not the peak) can mean that high
side drive voltate stays fairly constant. This wasn't the sort of thing I
felt should be introduced with the basic design :-).

> >Summary: It's NOT  a darlington - look again and you'll see ot's a little
> >unusual. and look at the drive polraity required by the pass
FET/transistor.
>
> Duh. It took me a while to see that absolutely everything is reversed from
> the N-channels that I'm used to using. Source is positive, Drain is
negative,
> Gate must be more negative than drain to get the part to conduct.

Duh here too - I thoughgt you were referring to my lower circuit where QB1
turns off Qb2 - also not a darlington and more or less the image of ther FET
gate turnoff circuit.

> I also think that I figured out that RB3 is a current limiting resistor
for
> high Vin. I'm testing with 12-15V Vins. I left it off in my prototype.

Yes. You really always want a resistor there although ideally it will be
much smaller than I have shown for low Vin voltages.


> >HOWEVER if the FET was an N Channel then this would work as desired.
> >However, we would then need to drive thre FET gate ABOVE Vin during FET
> >turnon and so need a high side supply. (as when on FET drain and source
are
> >connected so source is at Vin so gate must be at leat Vthreshold above
Vin
> >(plus a bit more).
>
> The bit more is based on the current passing through the device. I saw an
> equation like Rds*Ids...

No - more than that.
A FET has  arated threshold voltage . Apply a voltage less than this to the
FET and it will not conduct at all (essentially). Once gate voltagev exceeds
Vthresh the FET starts to turn on but it does not do so fully until you gate
a gate voltage several volta slarger. Typically for modernish N Channel FETS
Vthreshold is 3 to 7 volts. The datasheet will have a  family of curves
plotting the drop across the FET against various currents with a curve for
each value of gate voltage.
You will see that for small values of gate voltate the curves "knee" quite
early on and above certain values of drain current the drain-source voltage
rises quite steeply. This is not usually something you want to happen :-).
As gate voltage goes up the cirves become closer to sloping verticalsih
lines - ie the FET is beginning to look like a pur resistor across its
operating range. You want enough gate voltage to keep the Rdson low during
operations so dissipation stays low so we all get home to dinner OK.

> After reading (and gaining some understanding) I rushed out and bought
parts.
> I have an IRF9540 P-channel (Ids=19A, Vds=100V, rDS=0.2 ohms), a 10
microhenry
> 11A coil, a 330 uF cap, and a SB580 5A ultra fast schottky catch diode.
> Since I wasn't concerned with high Vin, I used ordinary 2N2222 for both
> transitors. I also dropped the values of all of the resistors since I'm
not
> worried about a high Vin. My concern was getting a bit better snap off on
the
> FET. I bought a handful of different types of zeners, but ended up using a
5V
> one. I'm still trying to get clued in on why the output voltage isn't 5.6V
> as you specified but 5.0V right on the mark.
>
> I'm pleased to report that it was a total success. Once I added RB2/QB2
and
> got the FET leads straight, Vout is right at the 5V mark.
>
> I did some preliminary high current testing. Built a dummy load out of 4
> 0.7 ohm 25W ceramics, wired in series/parallel to give a 0.7 ohm 100W
load.
> Final result: Vout=4.75V @ 7.8A !!! A total success.
>
> Next up is building a similar boost regulator to get 12V out even with a
6.5V
> input. I may even try wrapping that second coil so that I can get a low
current
> -12V.

> It does. I've been trying to get switchers under my belt for a while. Your
> design and this thread has taken me several light years further than where
> I was. I really appreciate it.

Excellent - now you can start on non-divinely inspired arcane "real" ones
:-) !!

regards

           Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
spamBeGonepiclist-unsubscribe-requestSTOPspamspamEraseMEmitvma.mit.edu


2001\08\24@083136 by Roman Black

flavicon
face
Hi Russell, i've been following your challenge
for simple SMPS to give a small cheap 5v regulator.
This is the first chance I had to give it some
time, here is a quick attempt at a solution
using one transistor as previously mentioned.
:o)

http://centauri.ezy.net.au/~fastvid/smps01.gif

Please forgive me if there is a major error or
problem, this was done in a bit of a hurry and
I haven't built one yet. :o)

* R2 D2 C2 are a crude 5.6v regulator, which
clamps the base of Q1 to 5.6v via the low ohms
base resistor R1.

* large ouput cap on emitter of Q1 acts as an
ac decoupled voltage reference, so Q1 turn on
is greatly affected by this voltage.

* this gives simple negative feedback, ie output
volts will never get much above 5v as it would
become impossible to turn on Q1. Hence meets
the goal of simple voltage regulation.

* L1 and D1 are slightly unconventional, load
current is supplied only when Q1 is on, and when
Q1 turns off the flyback dumps excess energy back
to the supply. This should be very efficient, there
are no resistors in the main current path,
but it will give higher load voltage ripple.
It might need a RC or even LC pi filter on the
output, but that's easy for small loads like 150mA.

* Oscillation is (hopefully!) provided by the
correct choice of feedback winding and R3. Ideally
this should oscillate Q1 base around 5.6v giving
good hard Q1 turn on/off while not affecting the
average voltage at Q1 base by much. Playing with
R1 and R3 should give reliable start and oscillate.

I'm sure this circuit can be made to oscillate
fairly reliably. I'm also sure it would give crude
regulation, and with pretty good efficiency.
Some problems might be; range of input voltage,
and range of load current. Then what do you expect
from one transistor?? :o)
-Roman

--
http://www.piclist.com hint: To leave the PICList
KILLspampiclist-unsubscribe-requestspamBeGonespammitvma.mit.edu


2001\08\24@090601 by Olin Lathrop

face picon face
>>
I think the feedback sense is reversed from what it should be: as is,
this circuit will latch up with Q1 either on or off, and stay there.
<<

Woops!  Good catch, thanks David!  As I said, I hadn't built this yet.  At
one point there was an additional inversion in there as I was evaluating
different topologies.  You just saved me a big "Doh!" after getting boards
back.

>>
By deleting R2, connecting +5VREF to the (-) opamp input, and
connecting the junction of R5 and R6 to the (+) opamp input and R4,
the regulator works fine (at least in SPICE).  With 20V in and 10V out
at one amp, efficiency looks to be about 80%.  Output ripple should be
about 850 mVp-p at this load, with an oscillation frequency of 6
kilohertz.  With input voltage reduced to 13 volts, efficiency goes up
to nearly 90%.
<<

Thanks again.  This matches very well with what I expected, although I
arrived at the characteristics by calculating time constants, currents,
voltages, etc.  It gives me more confidence in this circuit seeing theory
and simulation agree.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, EraseMEolinspamEraseMEembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
@spam@piclist-unsubscribe-request@spam@spamspam_OUTmitvma.mit.edu


2001\08\24@090608 by Olin Lathrop

face picon face
> > > The flyback diode DBUK2 MUST be rated for a peak current capability of
> > > several times the mean output current and MUST be a high speed part
(eg
> NOT
> > > 1N400X).
>
> Olin has suggested that this is NOT the case due to circuit time constants
> etc. I'm not convinced yet but his assertion needs looking at,. The
ability
> to use a low cost general purpose diode here would be an advantage.

It's a bit amgiguous what "this" stands for in your sentence.  I did say
that in MY circuit, the flyback diode didn't need to be high speed.  That
was because I carefully designed the circuit to guarantee that the flyback
current in the inductor would have ceased (and the diode would no longer be
conducting) by the time the next current pulse starts.  However, I make no
such claims for your circuit.

And yes, the flyback diode must be rated at several times the average
current.  The whole point of a buck converter is to feed current to the
output in small but high pulses thru the inductor.  The flyback diode will
be subjected to the same maximum current the inductor is.  My circuit was
designed for a maximum average output current of 1.5A, and the inductor and
the flyback diode are rated at 6A.  Note that you need to take these higher
currents into consideration when calculating the loss in the switching
element.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, spamBeGoneolinspamKILLspamembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
.....piclist-unsubscribe-requestspam_OUTspammitvma.mit.edu


2001\08\24@101703 by Olin Lathrop

face picon face
> There is one major problem with the FET used as shown. This is an
N-Channel
> device

No it's not.  I've got the spec sheet from IR right in front of me.  I don't
know where you got this from.

> The IRF5305 is somewhat dear - Digikey charge $US6.24 in 1's.

My DigiKey catalog is a little old (March 2001), but it shows the IRF5305 in
the SMD-220 package as $1.98 single, $1.58 for quantity 10.  I picked this
FET partly because of its reasonable price.  Hopefully there hasn't been a
drastic change in price since March.

> Consider placing a small series current limiting resistor between the
drive
> transistor emitters and FET gate to limit FET switching times. You can get
> immense currents here and FET losses can be increased by the too rapid
> switching (yes - you can get too much of a good thing :-) ).

I don't think there is much danger of too high a gate dV/dt here due to the
limitations of the LM6132 op amp, but you're right in that I should look
into this more.

> > 1  -  D6 need not be a fast recovery diode.  The values of L3, C4, the
> > hysterisis, and the 0-1.5A current draw range make sure that L3 is
always
> > done conducting before Q1 turns on for the next pulse.  This allows D6
to
> be
> > an ordinary power diode, which is very useful considering the currents
> > envolved.
>
> This is an interesting claim and one I will think further about. I don't
> really believe it but I hope you are cirrect as the saving in diode cost
can
> be significant.

I don't have time to go into all the calculations right now (taking off for
a few days and have lots to do to get ready).  Basically look at the current
thru L3.  Assume Q1 is a perfect switch when it turns on, and the ripple on
the 10V output is proportionately small.  The L3 current will start when the
output gets drained to the low threshold level.  Now assume the current
builds linearly and see what it gets to by the time the high threshold level
is reached and Q1 is turned off.  Now L3 essentially sees about a 9V reverse
bias, which will linearly ramp its current to zero.  Compare the time that
takes to the fastest C4 discharge time at max output current from the high
to low switching thresholds.  Unless I messed up, the L3 current will always
reach 0 before the C4 voltage reaches the low switching limit.  There are
some more details, but this is the general gist of the argument.

> Buck convertersa re usually best run in continuous current mode ...

I see nothing inherent to buck converters to support that statement.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, TakeThisOuTolin.....spamTakeThisOuTembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
TakeThisOuTpiclist-unsubscribe-requestKILLspamspamspammitvma.mit.edu


2001\08\24@185339 by Russell McMahon

picon face
> By deleting R2, connecting +5VREF to the (-) opamp input, and
> connecting the junction of R5 and R6 to the (+) opamp input and R4,
> the regulator works fine (at least in SPICE).  With 20V in and 10V out
> at one amp, efficiency looks to be about 80%.  Output ripple should be
> about 850 mVp-p at this load, with an oscillation frequency of 6
> kilohertz.  With input voltage reduced to 13 volts, efficiency goes up
> to nearly 90%.

I have trouble believing that efficiency in a real product with the FET
drive as shown. (But stranger things have happened). This is an N Channel
FEt and the gate is not being driven above Vin when the FET is on.
Presumably SPICE models Vthreshold properly?



               Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
.....piclist-unsubscribe-requestspamRemoveMEmitvma.mit.edu


2001\08\24@195842 by Dave Dilatush

picon face
Russell McMahon wrote...

>My real results AFAIK are say 85% at Vin = 15 and 55% say at Vin = 200 v for
>a 10 volts out say. at 600 mA.

SIX HUNDRED MILLIAMPS??????  Good grief, no wonder our mileages vary!

We've got an "apples and oranges" problem here- in fact, several of
them.

In your original design challenge post, you solicited design ideas for
a switching regulator that would output "a few tens of milliamps" at
five volts for operating a PIC and its associated circuitry given a
power source whose voltage could range up to 50 volts or more, and
doing so with "reasonable" efficiency.

And it was at "several tens of milliamps" (20 mA, specifically) that I
performed the simulation of your proposed circuit which formed the
basis for the comments in my previous post.  Frankly, it didn't work
worth a darn at those levels, which are the conditions under which I
assumed you intended the circuit to operate.

I was puzzled by this, hence my request for measurement data showing
what efficiency you had actually obtained in practice.

You've provided the data, but it refers to operation WAY outside the
range which you'd earlier indicated was of interest- at an output
current of 600 mA instead of 20 mA.

Do you have any efficiency data for this circuit operating at Iout = 20 mA?

The other source of confusion is, which of your circuits are we
talking about?  The one where the MOSFET gate is turned off only by
the 100K bleed resistor?  Or the one with the active turn-off circuit
(QBUK3 and DBUK1, in diagram picbful.gif)?  My comments referred to
the former while yours, I take it, refer to the latter- or perhaps to
some other variant of one of these.  Which is it?

I was intrigued by your design challenge because that's what designing
a low-current switching regulator is: a REAL challenge.  I've designed
a few in the past, including some that had to power the innards of a
2-wire, loop-powered 4-20 mA process transmitter- an application in
which the absolute maximum available operating current is only about
3.5 mA.  It ain't easy.

An output current of 600 mA, though, is right in the middle of the
"sweet spot" when it comes to switching regulator design.  The world
is absolutely awash in application notes which describe how to design
switching regulators with outputs ranging from a few hundred milliamps
up to a few amps; and there are literally dozens of different ICs
designed to do the job and most of them do it very, very well.

The 200 volt input does, I'll admit, pose a design challenge; but it
isn't one that's of much personal interest to me.  I don't like HV; it
hurts!

Incidentally, I re-simulated your circuit with QBUK3 and DBUK1
included to improve the MOSFET turn-off speed.  After playing with
component values a little, and loading the output to 500 mA, I got
simulation results that were more or less consistent with the
efficiency measurement data you cited.

I still don't like the Zener current, though.  Someday, that's going
to come back and bite you.

Dave

--
http://www.piclist.com hint: To leave the PICList
RemoveMEpiclist-unsubscribe-requestspamspamBeGonemitvma.mit.edu


2001\08\24@195846 by Dave Dilatush

picon face
Olin Lathrop wrote...

[snip]

>Unless I messed up, the L3 current will always
>reach 0 before the C4 voltage reaches the low switching limit.  There are
>some more details, but this is the general gist of the argument.
>
The simulation results I got support that.  At maximum load, and with
input voltage varied from 11 to 20 volts, the dead time (time between
cessation of inductor current flow at the end of one cycle and switch
turn-on at the beginning of the next) looks like it should go from a
minimum of 40 microseconds (at 11V in) to more than a hundred
microseconds at 20 volts.  
In no case did I see the switch turn on while current was still
flowing in the catch diode, making diode Trr a non-issue.

>> Buck convertersa re usually best run in continuous current mode ...
>
>I see nothing inherent to buck converters to support that statement.
>
Back in the days when a switching regulator like this might be
designed with <shudder> a PNP Darlington transistor as the switch, the
relatively high peak current levels associated with intermittent-mode
operation would have caused I * Vce(sat) losses that could constitute
a significant efficiency issue; but with modern MOSFETs as switches
this is hardly a problem anymore.

Dave

--
http://www.piclist.com hint: To leave the PICList
spamBeGonepiclist-unsubscribe-request@spam@spamspam_OUTmitvma.mit.edu


2001\08\24@204804 by Dave Dilatush

picon face
Roman Black wrote...

>* L1 and D1 are slightly unconventional, load
>current is supplied only when Q1 is on, and when
>Q1 turns off the flyback dumps excess energy back
>to the supply.

Take a careful, close look, while ignoring what you intended to have
happen, at what actually transpires at the instant Q1 turns off.

Just before that instant, current is flowing through L1 from top to
bottom, dutifully recharging C1.  When Q1 turns off, where is all that
current going to go?  It can't go through Q1, which has just turned
off; and it can't go through D1 either, because D1 is pointing in the
wrong direction for current flow.

But inductors being inductors, that current is darned well going to go
somewhere.

What ends up happening at that moment in time is a brief, violent, and
probably fatal contest between Q1 and D1 to see who can best withstand
the voltages generated by the inductor's stubborn insistence on
continuing the flow of current and ridding itself of its stored
energy.

My bet is that Q1 loses, but I don't think it matters much.

Dave

--
http://www.piclist.com hint: To leave the PICList
TakeThisOuTpiclist-unsubscribe-requestspamspammitvma.mit.edu


2001\08\24@205622 by Dave Dilatush

picon face
Russell McMahon wrote...

>> By deleting R2, connecting +5VREF to the (-) opamp input, and
>> connecting the junction of R5 and R6 to the (+) opamp input and R4,
>> the regulator works fine (at least in SPICE).  With 20V in and 10V out
>> at one amp, efficiency looks to be about 80%.  Output ripple should be
>> about 850 mVp-p at this load, with an oscillation frequency of 6
>> kilohertz.  With input voltage reduced to 13 volts, efficiency goes up
>> to nearly 90%.
>
>I have trouble believing that efficiency in a real product with the FET
>drive as shown. (But stranger things have happened). This is an N Channel
>FEt and the gate is not being driven above Vin when the FET is on.
>Presumably SPICE models Vthreshold properly?

It's not an N-channel FET.  As indicated on Olin's schematic (note the
direction of the body diode indicator on the MOSFET symbol) as well as
the datasheet (at
http://www.irf.com/product-info/datasheets/data/irf5305s.pdf ), it is
indeed a P-channel device.

Dave

--
http://www.piclist.com hint: To leave the PICList
piclist-unsubscribe-requestEraseMEspammitvma.mit.edu


2001\08\25@004046 by Russell McMahon

picon face
> > There is one major problem with the FET used as shown. This is an
> N-Channel
> > device
>
> No it's not.  I've got the spec sheet from IR right in front of me.  I
don't
> know where you got this from.

Aaaagh! - you're no doubt correct - my mistake.
I thought I'd downloaded the datasheet but when I went back and looked just
now I find it's for the IRF3205 - I must have written down the wrong number
and also I suppose the price refers to that too.  Nice FET though :-). If
yours is a P Channel then all is well.


> > Buck converters are usually best run in continuous current mode ...

> I see nothing inherent to buck converters to support that statement.

Continuous and discontinuous modes are quite different in the relationship
between input and output voltages. The continuous mode can be considered as
a PWM waveform filtered by the output LC filter. The discontinuous mode adds
extra terms for frequency of switching and inductance in the Vin/Vout
relationship. Swapping between modes in one design is generally
ill advised as design criteria for both are hard to meet simultaneously.
Arguably the continuous mode is most desirable. That said, any buck
converter will be forced to run in discontinuous mode under a light enough
load and this may well be acceptable for a normally continuous mode design .

Texas Instruments report SLVA057, "Understanding Buck Power Stages in
Switchmode Power Supplies" March 1999 gives a good overview of the two
modes and their differences.
They note -

       "It should be noted that the buck power stage is rarely
       operated in discontinuous mode in normal situations,
       but discontinuous conduction mode will occur anytime
       the load current is below the critical level."

This of course doesn't mean you shouldn't design a discontinuous mode
version if there are major advantages in doing so.




regards


           Russell McMahon

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\25@060549 by Russell McMahon

picon face
Russell McMahon wrote...

>My real results AFAIK are say 85% at Vin = 15 and 55% say at Vin = 200 v
for
>a 10 volts out say. at 600 mA.

and Dave said -
/SIX HUNDRED MILLIAMPS??????  Good grief, no wonder our mileages vary!
/
/We've got an "apples and oranges" problem here- in fact, several of
/them.

Hopefully not!

No intent to decieive here - what I did was report on and publish the
version that I had done most playing with.
I am confident that it will scale down well although I have not done good
tests at that sort of power level.I will try to do so in the next few days
and post results here.
My main intention was to encourage publication of various designs and to
show one very low cost control method. I specd it at the level I did as this
is often where the PICLISt queries lie. The 50 volt plus spec eliminates
most available IC's both linear and switchers.

I'll also try and get to trying a simplifed version of the Natsemi circuit
as it gives better control of the zener current and more flexible setting of
Vout.

Results soon ...



       Russell

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\25@075638 by Byron A Jeff

face picon face
For anyone who is interested in the basic theory, including a bunch of
control circuits Maxim has an outstanding tutorial here:

http://dbserv.maxim-ic.com/tarticle/view_article.cfm?article_id=93

Personally I find all of the control schemes listed much more complicated
than the self induced VCO design we have discussed here.

BAJ

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\25@103911 by Roman Black

flavicon
face
Dave Dilatush wrote:
{Quote hidden}

Whoops! Obviously D1 needs to have its position
in the curcuit adjusted. I said it was a quickie! :o)
-Roman

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\25@110903 by Dave Dilatush

picon face
Roman Black wrote...

>Whoops! Obviously D1 needs to have its position
>in the curcuit adjusted. I said it was a quickie! :o)

I tried playing around with the idea of relocating D1, but didn't get
very far.

One way of keeping Q1 from being fried by inductive kickback would be
to place D1 right across the main inductor L1, in the same manner in
which one would place a snubber diode across a relay coil.  This would
certainly protect Q1; the energy stored in the inductor would be
dissipated harmlessly in the diode.  But alas, "harmlessly" in this
instance must necessarily also mean "wastefully".  Efficiency would be
wretched, negating the benefit of having a switching regulator.

Another way of protecting Q1, while doing something useful with L1's
stored energy (like driving the output) would be to connect D1 across
the output terminals in series with a third inductor winding phased
appropriately so that when Q1 turns off, the inductor's stored energy
is dumped harmlessly- and usefully- into the output circuit in the
manner of a flyback converter.

This would probably work fine.  But we're now up to three windings on
this inductor, and it would very likely have to be a custom part; and
custom inductors are sufficiently costly that we've probably wiped out
the initial advantage sought with this one-transistor blocking
oscillator design.

I may be wrong, but this is beginning to look like one of those
"you-can't-get-there-from-here" situations.

Dave

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\25@143152 by Dave Dilatush

picon face
Russell McMahon wrote...

>>> Buck converters are usually best run in continuous current mode ...
>
>> I see nothing inherent to buck converters to support that statement.
>
>Continuous and discontinuous modes are quite different in the relationship
>between input and output voltages. The continuous mode can be considered as
>a PWM waveform filtered by the output LC filter. The discontinuous mode adds
>extra terms for frequency of switching and inductance in the Vin/Vout
>relationship.
All this says is that if you want to model the action of these two
different operating modes mathematically, you end up with two
different sets of equations, one to describe each mode.  And if the
circuit is allowed to go back and forth from one mode of operation to
the other, juggling these two different mathematical models becomes a
real pain in the neck.

But the circuit itself cares naught about mathematics: it just sits
there and dutifully oscillates, turning on and off the flow of current
through the inductor to keep the output voltage within its prescribed
bounds.

>Swapping between modes in one design is generally
>ill advised as design criteria for both are hard to meet simultaneously.

Why is this an issue?  Olin took great care to ensure his design
ALWAYS operates in the discontinuous-conduction mode, and simulation
results suggest that's precisely what it does.

You're aware, aren't you, that your own switching regulator design (as
shown in picbful.gif) engages in this allegedly ill-advised practice
of "swapping modes"?

Yes, that's exactly what it does: run it with a high enough load
current and/or a low enough input voltage, and it operates with
continuous inductor current.  Supply it with a high enough input
voltage and/or a light enough load, and it operates in discontinuous
current mode.

And when it transitions from one mode to the other with a swept load,
what happens is... absolutely nothing.  No sudden change in output
voltage; no transients of any kind; no instability or anything else.
It just sits there and does its thing.

So what's so "hard" about this?

>Arguably the continuous mode is most desirable.
Like your earlier assertion that "buck converters are usually best run
in continuous current mode", this is simply untrue as a general
statement.  Sometimes operating a stepdown converter this way is
advantageous; sometimes, it's not.  It all depends on the performance
characteristics you're trying to attain.

To make the sweeping generalization that "continuous mode is best" is,
quite simply, WRONG.

{Quote hidden}

"Rarely operated in discontinuous mode"?  
Sometimes you have to do a bit of translation on these application
notes.  I translate the above to read, "Some of TI's competitors make
some very nice switching regulator chips that operate in discontinuous
conduction mode.  Maxim even has a trademark for their version of that
mode, as used in their ultra-low power, high-efficiency switching
regulator chips: they call it Idle Mode(tm).  We don't want you to buy
their chips, we want you to buy ours!"  
Caveat emptor.

>This of course doesn't mean you shouldn't design a discontinuous mode
>version if there are major advantages in doing so.

When designing switching regulators to operate efficiently at very low
power levels, the advantages of discontinuous mode operation are not
merely "major": they're downright compelling.

Consider the sorry state of affairs we find ourselves in when we try
to design a low-current switcher which operates in the continuous
conduction mode.  As always with inductors, V = L * (dI/dT).  Since dI
(i.e., the change in inductor current in each part of the conduction
cycle) has to be small if we're to avoid having the inductor current
drop to zero at any point in the cycle, and the supply voltage V is
fixed, we have two choices: we can either make L very large, or we can
make dT very small.  Neither of these choices is very pretty.

For examples of the latter approach, check out recent offerings from
Linear Technology and from Maxim.  Some of their newer switching
regulators operate at over a megahertz (!), which allows them to use
reasonable size inductors.  That's fine for them, because they're
integrating these things on silicon and can achieve very high speeds;
but we peons don't have that option with discrete, home-built stuff.

The other approach, increasing L, is REALLY ugly: bigger inductors are
physically large (bad), lossy (bad), and expensive (bad).  For an
example of this approach, check out an old appnote from Linear Tech at
http://www.linear-tech.com/pdf/an30.pdf and look at Figure 24 on the
bottom of page 17.  This is a micropower switching regulator using an
LT1017 dual comparator chip, a 74C907 CMOS open-drain buffer chip
(there weren't any really decent P-channel power MOSFETs back in those
days), and a one hundred millihenry inductor!  Ridiculous.

And the problem just keeps getting worse, the lower the output current
(and, by the way, the higher the input voltage).

At low power levels, discontinuous-conduction mode becomes the only
way to fly: it allows the use of small inductors (check out Olin's
circuit, which uses a 10 microhenry inductor).  And small is cheap.
And cheap is good.  This is, after all, the PICLIST...

Dave

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\25@204212 by Jim

flavicon
face
Russell, all -

I found on the 'net a design similar, but not exactly the
same as, the "flyback" supply a friend and I developed
sometime back.

I would post what was developed 19 yrs ago, but, owing
to: time constraints, all source notes/schematic are still on
paper I probably won't get a chance to do so *but* I
did run across this little gem that was designed for the
same purpose as our was:

Circuit:
http://www.vmars.org.uk/a40_pt2.jpg

Circuit description:
http://www.vmars.org.uk/a40_pt2.htm


Jim

{Original Message removed}

2001\08\26@033411 by Roman Black

flavicon
face
Dave Dilatush wrote:
>
> Roman Black wrote...
>
> >Whoops! Obviously D1 needs to have its position
> >in the curcuit adjusted. I said it was a quickie! :o)
>
> I tried playing around with the idea of relocating D1, but didn't get
> very far.
>
> One way of keeping Q1 from being fried by inductive kickback would be
> to place D1 right across the main inductor L1, in the same manner in
> which one would place a snubber diode across a relay coil.  This would
> certainly protect Q1; the energy stored in the inductor would be
> dissipated harmlessly in the diode.  But alas, "harmlessly" in this
> instance must necessarily also mean "wastefully".  Efficiency would be
> wretched, negating the benefit of having a switching regulator.

Actually this is not so. :o)
To correct the circuit I would place a schottky
diode across the coil directly. This IS actually
efficient, it's similar to the "slow decay" system
used in large stepper motor drivers, and instead
of the coil energy being dumped into the input or
output supply when Q1 turns off, the magnetic field
of the coil can only decay very slowly, so the
energy is maintained IN the magnetic field until
the next time Q1 turns on again. This should have
similar efficiencies to other SMPS flyback systems,
and is used very effectively in many commercial
switching designs, although not often in psu's.

Apart from my glaring error with the diode placement,
i'm a bit surprised no-one has commented more on my
single transistor SMPS 5v regulator idea? I expected
at least a good argument. :o)

Here it is again (flyback diode has now been moved!):
http://centauri.ezy.net.au/~fastvid/smps01.gif

-Roman

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\26@124711 by Dave Dilatush

picon face
Roman Black wrote...

>Dave Dilatush wrote:

>> One way of keeping Q1 from being fried by inductive kickback would be
>> to place D1 right across the main inductor L1, in the same manner in
>> which one would place a snubber diode across a relay coil.  This would
>> certainly protect Q1; the energy stored in the inductor would be
>> dissipated harmlessly in the diode.  But alas, "harmlessly" in this
>> instance must necessarily also mean "wastefully".  Efficiency would be
>> wretched, negating the benefit of having a switching regulator.

>Actually this is not so. :o)

Really?  Build the circuit- physically, with real components, or
virtually, with a simulator- and observe its behavior. :)

>To correct the circuit I would place a schottky
>diode across the coil directly. This IS actually
>efficient, it's similar to the "slow decay" system
>used in large stepper motor drivers,
Stepper moter drivers != switching regulators.  
Whatever similarity you're seeing here may be visually appealing, but
not particularly relevant.  Putting a catch diode directly across an
inductor is a good way to prevent inductive kickback from frying
whatever device is driving the inductor, but it doesn't do a darn
thing for efficiency.

>...and instead
>of the coil energy being dumped into the input or
>output supply when Q1 turns off, the magnetic field
>of the coil can only decay very slowly, so the
>energy is maintained IN the magnetic field until
>the next time Q1 turns on again.
If we assume the above statement is true (see next paragraph for an
explanation of why it isn't), then we are left wondering how the
inductor's stored energy ever gets transferred to the output.  The
answer is that it doesn't.  With Vin > Vout (that is, with pin 2 of L1
positive with respect to pin 1), the energy stored in L1 during the
next conduction cycle can only increase, not decrease.  What ends up
happening is that in between the periods in which Q1 is turned on, the
stored energy in L1/L2 gets dissipated as heat, partly in D1 and
partly in R1 and R3.

But the next conduction cycle isn't actually going to begin until the
energy stored in L1 is dissipated, anyway.  For until this happens, D1
is conducting and its voltage drop gets reflected at secondary winding
L2, making pin 2 of L2 slightly negative with respect to pin 1,
delaying the turn-on of Q1.  Remember, this is what causes your
circuit to self-oscillate in the first place, right?  When the output
voltage drops low enough to start Q1 conducting, positive feedback
from L2 causes it to turn on more and more until it's fully
conducting; and when the current through Q1 starts to drop as a result
of C1 becoming recharged, the voltage coming out of L2 reverses
polarity and causes Q1 to turn off abruptly.  C1 begins to discharge
into the load, and when the output voltage drops low enough to begin
pulling current through Q1 again, the cycle repeats.

>This should have
>similar efficiencies to other SMPS flyback systems,

The very essence of a flyback SMPS is that it stores energy from the
power source in an inductor during one part of its operating cycle,
then transfers the inductor's stored energy into the load during the
other part of the cycle.

Your circuit contains no physical means whatsoever for transferring
that energy: energy is certainly stored in the inductor when Q1 is on,
but the only way it gets out is as heat.

>...and is used very effectively in many commercial
>switching designs, although not often in psu's.

I've worked on a number of SMPS designs over the last 25 years, with
power outputs ranging from milliwatts up to a third of a megawatt; and
I have never seen a switching regulator design with this topology.

>Apart from my glaring error with the diode placement,
>i'm a bit surprised no-one has commented more on my
>single transistor SMPS 5v regulator idea? I expected
>at least a good argument. :o)

Up here in the States, it's the weekend before our Labor Day holiday
and I suspect many folks are away on vacation or out tending the
barbecue.

Cheers,

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\26@170913 by hard Prosser

flavicon
face
part 1 677 bytes content-type:text/plain; charset=us-ascii
How about something like this.
The circuit attached is used to drive a 5V(~15mA) relay off a supply
ranging from about 20 to 70V.
I can't guarantee all the component values but they are about right. The
coil is the relay coil and the current is sensed by R7. If R7 became the
load and D1 became a zener at about 5.6V then it should work as a buck
converter. Transistor types are BF422/BF423 or BSR19A/BSR20A for the smt
version. Similarly the IN4148s are actually BAX12 or BAV99.  D2 stabilises
the feedback so that the hysterisis is stable with supply voltage. Input
voltage is limited mainly by transistor & diode ratings.

Richard P

(See attached file: relay_~1.gif)


part 2 7431 bytes content-type:image/gif; (decode)


part 3 144 bytes
--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\26@191843 by Dave Dilatush

picon face
Richard Prosser wrote...

>How about something like this.
>The circuit attached is used to drive a 5V(~15mA) relay off a supply
>ranging from about 20 to 70V.
>I can't guarantee all the component values but they are about right. The
>coil is the relay coil and the current is sensed by R7. If R7 became the
>load and D1 became a zener at about 5.6V then it should work as a buck
>converter. Transistor types are BF422/BF423 or BSR19A/BSR20A for the smt
>version. Similarly the IN4148s are actually BAX12 or BAV99.  D2 stabilises
>the feedback so that the hysterisis is stable with supply voltage. Input
>voltage is limited mainly by transistor & diode ratings.

I think this circuit is going to end up the winner.

After setting this thing up in SPICE and making a few tweaks (see
below), I got some really good efficiency results.  At 30V in and
5V/20mA out, efficiency was over 75%.  With that high an input/output
differential, and that low an output current, that's a pretty
impressive figure.  The efficiency holds up well even at Iout = 10 mA,
which is downright amazing.

As much as I liked my own design, this one has it beat both in cost
and efficiency.  For a micropower switcher, it's going to be hard to
do any better than this.

The changes I made while fooling around with it are as follows:

1.  Added a 100uF filter cap at the output, across the load (you have
   to have one, otherwise it won't likely oscillate);

2.  Changed L1 to 2.2mH to reduce cost/size;

3.  Reduced R5 to 47K to increase Q2's base drive a little;

4.  Eliminated R1 and C2, tying Q1's emitter directly to the output;

5.  Eliminated R8 and D2, and tied the right-hand end or R9 directly
   to the top of L1.

6.  Made D1 a 1N4734 (5.6V) Zener diode and flipped it over; and

7.  Rearranged D1, R10 and R2, by connecting R10 directly to the (+)
   end of D1 and moving R2 to a position between that junction and
   the junction of R9 and Q1's base (this made a big improvement
   in line regulation).

Nice circuit.

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\26@204114 by hard Prosser

flavicon
face
Thanks for the suggestions & comments - as I noted, it works OK as a relay
driver (hence the high inductance) but I was too busy last week to try any
changes to make it a straightforward buck regulator.
Richard P




                   Dave Dilatush
                   <dilatush@HOME        To:     RemoveMEPICLISTEraseMEspamspam_OUTMITVMA.MIT.EDU
                   .COM>                 cc:
                   Sent by: pic          Subject:     Re: [EE]: Design Challenge - low power step down switching              regulator
                   microcontrolle
                   r discussion
                   list
                   <PICLIST@MITVM
                   A.MIT.EDU>


                   27/08/01 11:14
                   Please respond
                   to pic
                   microcontrolle
                   r discussion
                   list






Richard Prosser wrote...

{Quote hidden}

I think this circuit is going to end up the winner.

After setting this thing up in SPICE and making a few tweaks (see
below), I got some really good efficiency results.  At 30V in and
5V/20mA out, efficiency was over 75%.  With that high an input/output
differential, and that low an output current, that's a pretty
impressive figure.  The efficiency holds up well even at Iout = 10 mA,
which is downright amazing.

As much as I liked my own design, this one has it beat both in cost
and efficiency.  For a micropower switcher, it's going to be hard to
do any better than this.

The changes I made while fooling around with it are as follows:

1.  Added a 100uF filter cap at the output, across the load (you have
   to have one, otherwise it won't likely oscillate);

2.  Changed L1 to 2.2mH to reduce cost/size;

3.  Reduced R5 to 47K to increase Q2's base drive a little;

4.  Eliminated R1 and C2, tying Q1's emitter directly to the output;

5.  Eliminated R8 and D2, and tied the right-hand end or R9 directly
   to the top of L1.

6.  Made D1 a 1N4734 (5.6V) Zener diode and flipped it over; and

7.  Rearranged D1, R10 and R2, by connecting R10 directly to the (+)
   end of D1 and moving R2 to a position between that junction and
   the junction of R9 and Q1's base (this made a big improvement
   in line regulation).

Nice circuit.

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\26@214513 by Russell McMahon

picon face
Appears to be a very nice circuit.
Similar in concept to mine but eliminates a transistor by emitter driving
the control transistor.
And the zener is a true reference so can have its current controlled
independently of feedback.
The high side drive is no better (in fact no different) than in my circuit
and would benefit from the same optimisations if a FET was used as high side
switch. (eg a bipolar high side FET driver).

As well as Dave's suggested changes It would be worth trying removing R9
(replace with O/C) and R2 (replace with S/C).
This would remove the resistive hysteresis and end up with the same
hysteresis system as in "my" design.

Also, for higher input voltages consider driving D1 (now a zener as per
Dave's changes) from the OUTPUT via a resistor and tapping the emitter of Q1
off a resistive divider off the output. Now a high value of resistor from
Vin to D1 will initially bias the zener on and once the system starts, zener
current will be provided from Vout. This reduces the otherwise not
insignificant power taken by the zener bias current.

I realise that anyone who hasn't looked at and modified the circuit diagrams
will be unable to make any sense of this. I'll try to get to post a GIF of
this 'shortly" unless someone else beats me to it. Dave?




           Russell McMahon







Richard Prosser wrote...

{Quote hidden}

I think this circuit is going to end up the winner.

After setting this thing up in SPICE and making a few tweaks (see
below), I got some really good efficiency results.  At 30V in and
5V/20mA out, efficiency was over 75%.  With that high an input/output
differential, and that low an output current, that's a pretty
impressive figure.  The efficiency holds up well even at Iout = 10 mA,
which is downright amazing.

As much as I liked my own design, this one has it beat both in cost
and efficiency.  For a micropower switcher, it's going to be hard to
do any better than this.

The changes I made while fooling around with it are as follows:

1.  Added a 100uF filter cap at the output, across the load (you have
   to have one, otherwise it won't likely oscillate);

2.  Changed L1 to 2.2mH to reduce cost/size;

3.  Reduced R5 to 47K to increase Q2's base drive a little;

4.  Eliminated R1 and C2, tying Q1's emitter directly to the output;

5.  Eliminated R8 and D2, and tied the right-hand end or R9 directly
   to the top of L1.

6.  Made D1 a 1N4734 (5.6V) Zener diode and flipped it over; and

7.  Rearranged D1, R10 and R2, by connecting R10 directly to the (+)
   end of D1 and moving R2 to a position between that junction and
   the junction of R9 and Q1's base (this made a big improvement
   in line regulation).

Nice circuit.

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\26@223913 by hard Prosser

flavicon
face
I'm pretty sure that without the resistive hysterisis the regulator entered
a "linear" mode. Possibly with the cap. accross the load the current kick
would be enough to start it unless the supply came up slowly. Certainly as
a relay driver it needs the feedback.

Richard P




Appears to be a very nice circuit.
Similar in concept to mine but eliminates a transistor by emitter driving
the control transistor.
And the zener is a true reference so can have its current controlled
independently of feedback.
The high side drive is no better (in fact no different) than in my circuit
and would benefit from the same optimisations if a FET was used as high
side
switch. (eg a bipolar high side FET driver).

As well as Dave's suggested changes It would be worth trying removing R9
(replace with O/C) and R2 (replace with S/C).
This would remove the resistive hysteresis and end up with the same
hysteresis system as in "my" design.

Also, for higher input voltages consider driving D1 (now a zener as per
Dave's changes) from the OUTPUT via a resistor and tapping the emitter of
Q1
off a resistive divider off the output. Now a high value of resistor from
Vin to D1 will initially bias the zener on and once the system starts,
zener
current will be provided from Vout. This reduces the otherwise not
insignificant power taken by the zener bias current.

I realise that anyone who hasn't looked at and modified the circuit
diagrams
will be unable to make any sense of this. I'll try to get to post a GIF of
this 'shortly" unless someone else beats me to it. Dave?




           Russell McMahon







Richard Prosser wrote...

{Quote hidden}

I think this circuit is going to end up the winner.

After setting this thing up in SPICE and making a few tweaks (see
below), I got some really good efficiency results.  At 30V in and
5V/20mA out, efficiency was over 75%.  With that high an input/output
differential, and that low an output current, that's a pretty
impressive figure.  The efficiency holds up well even at Iout = 10 mA,
which is downright amazing.

As much as I liked my own design, this one has it beat both in cost
and efficiency.  For a micropower switcher, it's going to be hard to
do any better than this.

The changes I made while fooling around with it are as follows:

1.  Added a 100uF filter cap at the output, across the load (you have
   to have one, otherwise it won't likely oscillate);

2.  Changed L1 to 2.2mH to reduce cost/size;

3.  Reduced R5 to 47K to increase Q2's base drive a little;

4.  Eliminated R1 and C2, tying Q1's emitter directly to the output;

5.  Eliminated R8 and D2, and tied the right-hand end or R9 directly
   to the top of L1.

6.  Made D1 a 1N4734 (5.6V) Zener diode and flipped it over; and

7.  Rearranged D1, R10 and R2, by connecting R10 directly to the (+)
   end of D1 and moving R2 to a position between that junction and
   the junction of R9 and Q1's base (this made a big improvement
   in line regulation).

Nice circuit.

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\27@092845 by Roman Black

flavicon
face
Dave Dilatush wrote:
>
> Roman Black wrote...
>
> >Dave Dilatush wrote:
>
> >> One way of keeping Q1 from being fried by inductive kickback would be
> >> to place D1 right across the main inductor L1, in the same manner in
> >> which one would place a snubber diode across a relay coil.

> >Actually this is not so. :o)
>
> Really?  Build the circuit- physically, with real components, or
> virtually, with a simulator- and observe its behavior. :)

Ummm, actually I have number of stepper motor
drivers using the "standard" slow-decay system
and all showing excellent SMPS efficiency...
I credit "Jones on Steppers" :o)

> >To correct the circuit I would place a schottky
> >diode across the coil directly. This IS actually
> >efficient, it's similar to the "slow decay" system
> >used in large stepper motor drivers,
>
> Stepper moter drivers != switching regulators.

Hmm. Sounds like an argument! :o) So you're saying
the switching efficiencies shown in motor drivers
don't translate to current supplied from point A
to point B??

{Quote hidden}

Now I understand your argument better. :o)


{Quote hidden}

OK. Now we get to the crux of the argument... First,
I have the greatest of respect for your 25 years SMPS
design experience, I have similar years but probably
from a different background. :o)

Please argue the following (simplified) points: When
you apply volts to a series inductor, it takes ENERGY
to build the magnetic field in the inductor, at the same
time current does now flow until that field is built, hence
the XL=2pifL formula. So in most SMPS supplies we rely
on total (series) current being limited by XL. So with
any series inductor we get the energy we stored in the
magnetic field, that we have already paid for, which can
THEN be distributed in 1 of 3 ways when the main switcher
turns off:

1. Field collapses- energy fed back to psu input cap.
2. Field collapses- energy fed forward to psu output cap.
3. Energy is maintained within the field, by slow decay,
(ie, the coil is short-circuited) as in stepper and
most pwm DC motor drivers.

In a standard buck supply the field is encouraged to
collapse, and the energy is fed by the flyback diode to
the output cap. Cool. It's efficient. BUT, if we keep
the magnetic field WITHIN the coil, as in a motor driver,
the initial energy can't go anywhere, hence no waste.
The next time Q1 turns on, we don't need to re-establish
the field because it's still there. And WHEN Q1 turns on,
the current is fed straight to the output, because we
don't need to first establish the field. So after the
first cycle, each Q1 turnon simply transfers current to
the output, giving us current transfer with no resistive
losses. Which is the essense of SMPS efficiency, same
current in as current out, but the voltage difference
is taken care of...  :o)

Remember that once we have paid the energy cost to "charge"
an inductor with field, that energy MUST go somewhere.
If we short circuit the inductor with a Vf 0.2v schottky
diode, the inductor energy CAN'T collapse, due to simple
physics. To make this easy for argument sake, if we charge
the inductor with 20vdc, and short it at 0.2vdc, for equal
time periods, you get 100:1 efficiency, in terms of energy
flow. :o)

Please argue, I appreciate anyone who has made SMPS's for
25 years, as I know that you learned X things in those
years and I learned Y things in those years and it sure
beats arguing the football scores... :o)
-Roman

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@101224 by Dave Dilatush

picon face
Russell McMahon wrote...

>The high side drive is no better (in fact no different) than in my circuit
>and would benefit from the same optimisations if a FET was used as high side
>switch. (eg a bipolar high side FET driver).

If he were to use a MOSFET as a switch, then something would indeed
have to be done- just as in your circuit- to give a more positive
turn-off, or efficiency would suffer greatly.

However, why would he use a MOSFET?  How would the design benefit?
Would it improve the efficiency at all, at the low output currents
we're dealing with here?  Would it lower the cost?  What's the
benefit?

>As well as Dave's suggested changes It would be worth trying removing R9
>(replace with O/C) and R2 (replace with S/C).
>This would remove the resistive hysteresis and end up with the same
>hysteresis system as in "my" design.

As Richard has already pointed out, this would merely render his
design a linear regulator; under most conditions, it would not
oscillate.

R2 and R9 are important.  A properly designed switching regulator
employs hysteresis as a means of exercising explicit control over both
operating frequency and output ripple voltage.  And in the context of
switching regulators, "hysteresis" denotes a bistable circuit with two
distinct voltage trip levels: when the output voltage reaches the
upper threshold, the switch turns off; and when the output voltage
drops to the lower threshold, the switch turns on.

These two threshold levels are distinct and they are explicitly set by
component values, and these are established by calculation.  In the
case of Richard's circuit (as modified in my previous post), the upper
threshold is

  Vthupper = Vz(D1) - Vbe(Q1) + [Vin * R2 / [R2 + R9]]

and the lower threshold (assuming discontinuous conduction) is

  Vthlower = Vz(D1) - Vbe(Q1) - [[Vz(D1) - Vout] * R9 / [R2 + R9]]

which, given R9 >> R2 and Vz(D1) ~ Vout, simplifies approximately to

  Vthlower = Vz(D1) - Vbe(Q1).

In practice, the output voltage moves outside these threshold levels.


On the positive side, the output voltage will continue to increase
after the switch has been turned off until the current in the inductor
drops down to the level of the load current, whereupon the output
voltage will reach its peak and then start to fall.  On the minus
side, the output voltage continues to fall below the lower threshold
voltage until the current in the inductor builds up to a value greater
than the load current, at which point the output voltage starts to
rise again.

Neither of these effects is magic, and both can be calculated knowing
the input voltage, output voltage, load current, inductance of the
inductor, and capacitance of the output filter capacitor.  It's pretty
straightforward, if a little laborious.

Several of us have pointed out, Russell, that your circuit does not
employ hysteresis as a means of governing its operation; yet you refer
once again to the "hysteresis system" in your design.

One more time: your design does not- repeat **NOT**- employ hysteresis
as its mechanism of oscillation.  It starts out as a linear feedback
system which, because of its high open-loop gain and the large phase
shifts in its feedback path, is, shall we say, "unconditionally
unstable".  Oscillation starts small and builds up intensity until,
finally, this erstwhile linear system is now alternating, more or less
abruptly, back and forth between two more or less saturated states.
At this point, it is best characterized as a "ring oscillator": like
an oscillator constructed from three logic inverters connected in a
ring, its frequency of oscillation is governed by delays, which in
turn are determined by a variety of factors- including component
characteristics which aren't terribly well controlled.

Yes, your circuit oscillates; and yes, over a fairly wide range of
input/output conditions it appears to function fairly well as a
regulator; but its operation has nothing to do with hysteresis nor
does it exert the explicit control over frequency and output voltage
ripple which hysteresis does.  (If you continue to insist it does,
then could you tell me what the two hysteresis thresholds are, and the
equations- rough approximations will do- that govern them?)

>Also, for higher input voltages consider driving D1 (now a zener as per
>Dave's changes) from the OUTPUT via a resistor and tapping the emitter of Q1
>off a resistive divider off the output. Now a high value of resistor from
>Vin to D1 will initially bias the zener on and once the system starts, zener
>current will be provided from Vout. This reduces the otherwise not
>insignificant power taken by the zener bias current.

While in principle this can work, in practice the impedance presented
by the divider (which appears as an Re in series with the emitter of
Q1) will likely reduce the open-loop gain enough to kill oscillation.
The upper divider resistor could, of course, be bypassed with a
capacitor to reduce the effective series impedance between the output
and the emitter of Q1; or the divider resistance values could be
decreased to similarly restore the gain; but I question whether the
net decrease in Zener bias power wouldn't be offset by that consumed
in the divider.

>I realise that anyone who hasn't looked at and modified the circuit diagrams
>will be unable to make any sense of this. I'll try to get to post a GIF of
>this 'shortly" unless someone else beats me to it. Dave?

See attachment.

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@111956 by Dave Dilatush

picon face
Roman Black wrote...

[snip]

>Please argue, I appreciate anyone who has made SMPS's for
>25 years, as I know that you learned X things in those
>years and I learned Y things in those years and it sure
>beats arguing the football scores... :o)

Roman, this circuit is like one of those "Roach Motel" thingies:
energy can check into that inductor, but it can't check out.  Under no
circumstances does ANY of the energy stored in the inductor EVER find
its way to the output circuit.  Nary so much as a picojoule.

Going back over my previous post, I don't see how I could clarify my
explanation any; it seems pretty straightforward.

As a thought experiment you could try the following: assume the
inductor and the diode across it are both perfect, so that when Q1 is
off, the inductor current doesn't decay at all.  In other words,
whatever current was flowing through the inductor at the end of the
last pulse, is still flowing through it at the beginning of the next.
And each pulse will increment that current by V * dT / L, where V is
the voltage across the inductor and dT is the duration of Q1's ON
time.  Thus each pulse delivers a bigger and bigger shot of current
into the output capacitor than the one before, which recharges the
output capacitor back to the threshold level faster and faster, and
the pulses get steadily more brief- and intense- as time goes by.

Ask yourself, what is there to limit this pernicious process?

No, don't tell me.  Rather than perpetuate this debate and getting all
vehement about it or anything, why not let's agree wholeheartedly with
one another (heft a pint with me, here) that regardless of what either
of us says, the circuit itself will be the final judge.

You've proposed it, now it's time to build and test it.  I look
forward to seeing your results.

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@112356 by Dave Dilatush

picon face
part 1 167 bytes content-type:text/plain; charset=us-ascii (decoded quoted-printable)

I wrote...

>See attachment.

What attachment?  I don't see any attachment...

Oh, THIS attachment.

(sorry, all)




part 2 5042 bytes content-type:image/gif; name=RPmodified.gif (decode)


part 3 131 bytes
--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@114131 by Roman Black

flavicon
face
Dave Dilatush wrote:
>
> Roman Black wrote...
>
> [snip]
>
> >Please argue, I appreciate anyone who has made SMPS's for
> >25 years, as I know that you learned X things in those
> >years and I learned Y things in those years and it sure
> >beats arguing the football scores... :o)
>
> Roman, this circuit is like one of those "Roach Motel" thingies:
> energy can check into that inductor, but it can't check out.  Under no
> circumstances does ANY of the energy stored in the inductor EVER find
> its way to the output circuit.  Nary so much as a picojoule.

Understood. The energy doesn't have to GO anywhere,
it just needs to be NOT wasted...

> Going back over my previous post, I don't see how I could clarify my
> explanation any; it seems pretty straightforward.

OK. :o)

{Quote hidden}

Not quite. I think once the inductor is "charged",
ie, it's field is established, the current fed in is
transferred to the output. So each time Q1 turns on,
the current is fed straight to the output, and the
output is at 5vdc, and the inductor maintains its field.
Current by definition follows the water analogy, it flows
in one end and comes out somewhere else.


> Ask yourself, what is there to limit this pernicious process?
>
> No, don't tell me.  Rather than perpetuate this debate and getting all
> vehement about it or anything, why not let's agree wholeheartedly with
> one another (heft a pint with me, here) that regardless of what either
> of us says, the circuit itself will be the final judge.

Heck this weekend I had more than a pint!! :o)

> You've proposed it, now it's time to build and test it.  I look
> forward to seeing your results.

Well, I haven't tested THIS circuit, but I have a few
stepper circuits, that supply 1.0v 5.0A across the motor
(through the motor??) and only draw 24v 0.3A from the
supply. Efficiency is what we are arguing. :o)

So where does the wasted energy go??? If the inductor
is kept "charged" because there is a virtual short
circuit across the coil, the energy is kept WITHIN
the magnetic field, and every SMPS cycle the energy
is still there, it's not wasted. The waste can only
occur when the field collapses and needs to be
re-established.

Hey, this is a good argument! It would be cool if some
others added their 2c worth??
-Roman

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@125147 by Dave Dilatush

picon face
Roman Black wrote...

>Dave Dilatush wrote:

>> You've proposed it, now it's time to build and test it.  I look
>> forward to seeing your results.
>
>Well, I haven't tested THIS circuit, but I have a few
>stepper circuits...

Build THIS circuit, please, and tell us your test results.  I'm sure
you've got some very nice stepper motor drivers, but whatever they do
is NOT relevant to the matter at hand.  It's what THIS circuit does,
not what your stepper drivers do, that counts.

>Hey, this is a good argument! It would be cool if some
>others added their 2c worth??

No, this is a pointless, **silly** argument.  Let's see some actual
results and settle this.

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@133720 by Eisermann, Phil [Ridg/CO]

flavicon
face
> -----Original Message-----
> From: Roman Black [@spam@fastvidRemoveMEspamEraseMEEZY.NET.AU]

> Hey, this is a good argument! It would be cool if some
> others added their 2c worth??
> -Roman

well, ok. Here's my 2c, but please remember to adjust for inflation:

> I think once the inductor is "charged",
> ie, it's field is established, the current fed in is
> transferred to the output. So each time Q1 turns on,
> the current is fed straight to the output

But when Q1 is off, where does the energy go? Does it not get wasted as heat
in the diode and feedback windings? When Q2 is off, it certainly does not
get transferred to the output. Slow decay or not, the inductor is going to
discharge through the diode. You charge the inductor, but then don't do
anything with the energy you stored there.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@140320 by Roman Black

flavicon
face
Eisermann, Phil [Ridg/CO] wrote:
{Quote hidden}

But do we need to do anything with that energy??
Why don't we just leave that energy in the inductor?
Then when Q1 next turns on the inductor is already
charged so next energy is fed to the load?

Slow-decay seems to be a black art for people that
never work with large motor drivers or stepper motor
drivers, but it's pretty simple really. The enery
stored in the magnetic field of an inductor can be
discharged into a load, or kept within the inductor.
The rate the energy is discharged relates to the
load. With a very high load (short circuit) then the
inductor cannot release the energy, as this current
causes magnetic field in the inductor...

If you want a fast decay, dump the inductor into a
high resistance, you get a very fast very high volts
pulse. Slow decay means dumping it into a low
resistance, you get a very long, very low volts pulse.
Hence it takes longer for the field to be discharged,
so the energy is kept in the inductor. In a SMPS
circuit we can control the time the inductor has
to discharge, so we don't let it discharge. Anyone
who's made a good PWM motor driver has done this.
-Roman

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@205345 by Russell McMahon

picon face
part 1 3730 bytes content-type:text/plain; (decoded 7bit)

> > But when Q1 is off, where does the energy go? Does it not get wasted as
heat
> > in the diode and feedback windings? When Q2 is off, it certainly does
not
> > get transferred to the output. Slow decay or not, the inductor is going
to
{Quote hidden}

OK - energy storage between files can be achieved if "done properly".
Here is a circuit which DOES work and which is "reasonably" efficient.
Many different versions of this have been built and used in practical
situations.
It is similar in general concept to Roman's circuit.
It uses energy storage between power delivery cycles but in an inductor /
capacitor system.

Note that this circuit does NOT operate in quite the way that many others of
seemingly similar topology do.

The asterisks indicate coil "dottings". The dotted ends have the same
"winding sense" on the core.
The circuit is a FORWARD converter. Power is delivered when the transistor
is turned on.

C1 is an UTTERLY crucial part of the circuit. C1 / L1 resonate and the
switching frequency is largely controlled by this resonance.

R1 is used largely to provide startup bias to the transistor.
R2 limits drive current when operating.

The junction of R1/R2 operates at a DC level BELOW ground when operating
!!!!
Look at the circuit and think it through.
Clue - Vbe of Q1 is about 0.6 volts and L3 is designed to have a voltage
during turn on of several volts.

As shown the circuit provides an output voltage approx N_l2/N_l1 x Vin.
Best for fairly constant Vin as from eg car battery or similar.

This circuit can be made to produce a lower voltage than this by providing a
feedback loop which turns the oscillator off when Vout reaches Vdesign.
Typically a zener, transistor and a resistor or two. Adding a flyback diode
to ground (cathode to cathode with D1), a series filter inductor and output
capacitor would result in a fairly conventional forward converter.

Apart from the "annoyance" of several coils on the inductor this is a
remarkably simple circuit for the results it can produce.

The waveform on the collector of Q1 is educational.
It is an approximate half sinusoid ABOVE Vin during the off half cycle and a
saturated grounded  square wave during the on cycle. Switching times are
much better than might be expected for a circuit with a resonant collector
tank.
The magnitude of the collector ring above Vin can be controlled by varying
C1.

A limited amount of output above Vin may be taken from the collector during
ringing but this is during the half cycle that is normally NOT used for
power so care should be taken to not excessively perturb the waveform.

Dave may be interested in trying this circuit in Spice and telling us
whether it works in simulation as well as it does in practice.



regards

               Russell McMahon



part 2 1482 bytes content-type:image/gif; (decode)


part 3 131 bytes
--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@205357 by Russell McMahon

picon face
It seems to me that we all seem to be getting a bit too wound up over this.
I won't address every point in Dave's last few posts commenting on my
comments on his comments on my ....   :-) . I should be working so will just
make a few overall comments. I assume that our mutual aim os to advance the
useful knowledge available in this area.

I am reasonably conversant with the technical aspects that are involved in
the various circuits that have been discussed and Dave obviously is too. .
I am aware that there is no "magic" involved in any of the circuits - just
"ordinary old physics". The circuit that I provided is by no means an
optimal one if you are wanting highest efficiency, cleanest waveforms,
lowest losses, controlled frequency or even well a defined operating point.
It is however a superb solution for some applications.

RM
>The high side drive is no better (in fact no different) than in my circuit
>and would benefit from the same optimisations if a FET was used as high
side
>switch. (eg a bipolar high side FET driver).

Dave
/If he were to use a MOSFET as a switch, then something would indeed
/have to be done- just as in your circuit- to give a more positive
/turn-off, or efficiency would suffer greatly.
/However, why would he use a MOSFET?  How would the design benefit?
/Would it improve the efficiency at all, at the low output currents
/we're dealing with here?  Would it lower the cost?  What's the
/benefit?

Very little potential benefit at the power levels we are talking about (and
as per original spec). It was a general comment relating larger to higher
power applications. I also noted that my circuit could be (and has been)
used with a high side bipolar transistor. At high power and/or high voltage
levels A FET can be advantageous.

>As well as Dave's suggested changes It would be worth trying removing R9
>(replace with O/C) and R2 (replace with S/C).
>This would remove the resistive hysteresis and end up with the same
>hysteresis system as in "my" design.

/As Richard has already pointed out, this would merely render his
/design a linear regulator; under most conditions, it would not
/oscillate.

Maybe.
You may be surprised. The sdame argument applies to "my" circuit.
See below re hysteresis.

/R2 and R9 are important.  A properly designed switching regulator
/employs hysteresis as a means of exercising explicit control over both
/operating frequency and output ripple voltage.  ....
/*** big snip ***

Agree. But hysteresis does not necessarily have to be achieved by more
traditional means - as long as the means used are technically valid.
The phrase "properly designed" is a key one here. My circuit is not well
characterised. By the nature of the predominant feedback mechanisms it would
be difficult (but not impossible) to design it for a given frequency of
operation etc. This may be unacceptable in some applications but not in all
applications.

/Several of us have pointed out, Russell, that your circuit does not
/employ hysteresis as a means of governing its operation; yet you refer
/once again to the "hysteresis system" in your design.
/**snip**
/..                   (If you continue to insist it does,
/then could you tell me what the two hysteresis thresholds are, and the
/equations- rough approximations will do- that govern them?)

I think the difference is one of semantics.
Hysteresis is the condition where a state change which is triggered by an
increase (say) in a variable alters the condition under which state change
occurs so that the variable must be decreased below the original trip point
before the system will again revert to its original condition. (The same
applies for a transistion in the other direction). I assume that you accepot
this as a reasonable definition of hysteresis. It does not say anything
about HOW the effect is achieved. Importantly, it does not say whether the
state changes are static or dynamic (ie whether time plays a part in the
mechanism). A purely resistive hysteresis mechanism is static. A reactive
mechanism may be dynamic but still meet the definition in letter and spirit.

Hysteresis is most normally effected by a resistive feedback mechanism where
the reference point is shifted by negative feedback from the system output
(so that the reference level is reduced when the output goes high and
increased when the output goes low. (A typical example is the common single
opamp oscillator with resistive feedback to a reference on the inverting
input and what can be seen as negative feedback to a capacitor on the
non-inverting input).

If this is what you understand as hysteresis then my original circuit has
got it.
The mechanism is NOT the normal resistive feedback version.
I have described the mechanism in a previous post.

The circuit ALWAYS turns on due to turnon bias from Vin.
The inductor current always increases.
Vout always rises.
There will come a point where the detection threshold is reached.
Turn off will start.
Energy delivery to the output from Vin will start to decrease.
*** BUT *** energy delivery to Vout will continue to INCREASE as energy
stored in the inductor will be delivered to Vout.
The output capacitor voltage will continue to rise.due to stored energy.
The mechanism driving the turnoff will be driven harder even though the
system is turning off.
This is NOT a usually predominat  mechanism for controlling turnoff.
Using an infinite value of output capacitance will kill this loop.
As it would kill any loop that relied on output voltage evariation for its
trip mechanism.
This is hysteresis (even if not as we normally implement it, Jim) - Vout
continues to rise enhancing the turnoff even as turnoff proceeds.

A similar mechanism occurs at at turn on.

Notice that on my original circuit I explicitly addressed this means of
hysteresis by adding a capacitor which was charged when the reference zener
conducted and which continued to hold the FET drive circit on for a period
after the zener stopped conducting. Somebody (?Dave?) said that it was "in
the wrong place". If you want it to take part in the above described
mechansim it is in the correct place. In practice it is not in fact
essential to circuit operation. If you have some other role for it then it
maybe should be somewhere else. .



Re how nactively this circuit oscillates. - "try it, you'll like it". I
don't know how it works in SPICE but in practice it starts as cleanly as any
oscillator I have ever seen - immediate sharp transistion from linear mode
to full clean oscillation. Always full amplitude or no oscillation at all.
No intermediate sluggish startup states. Look at a PIC oscillator sometime
and see how poorly it starts under some conditions of voltage and
capacitance. No similarity here. It's going or it's not. I haven't looked at
it on a storage scope but I suspect it has a 1 cycle startup transition.

>Also, for higher input voltages consider driving D1 (now a zener as per
>Dave's changes) from the OUTPUT via a resistor and tapping the emitter of
Q1
>off a resistive divider off the output. Now a high value of resistor from
>Vin to D1 will initially bias the zener on and once the system starts,
zener
>current will be provided from Vout. This reduces the otherwise not
>insignificant power taken by the zener bias current.

/While in principle this can work, in practice the impedance presented
/by the divider (which appears as an Re in series with the emitter of
/Q1) will likely reduce the open-loop gain enough to kill oscillation.
/etc

Maybe. I understand what you are saying but I suspect it can be made to work
well.
I'll try it some time and report back.




regards

           Russell McMahon

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\27@231617 by Dave Dilatush

picon face
Russell McMahon wrote...

>It seems to me that we all seem to be getting a bit too wound up over this.

Oh, I don't know about that.  This is rather fun, actually.

>The circuit that I provided is by no means an
>optimal one if you are wanting highest efficiency, cleanest waveforms,
>lowest losses, controlled frequency or even well a defined operating point.

Thank you.  That, in a nutshell, was my point.

>It is however a superb solution for some applications.

I'm sure it is, particularly if you've tested it thoroughly and have
been satisfied that it will function properly in your application
under whatever conditions it needs to function.  What works, works.

As to whether or not it is a solution in anything I am likely to be
doing, that is another matter: I seldom need output currents exceeding
a few score of milliamperes, and I never have to deal with input
voltages above a few dozen volts.  Something that pumps out ten volts
at over half an ampere, with an input ranging from 50 volts to 200
volts, is a curiosity, and no more.

But if that's what you need, and this circuit works for you, that's
great.

>Re how nactively this circuit oscillates. - "try it, you'll like it". I
>don't know how it works in SPICE but in practice it starts as cleanly as any
>oscillator I have ever seen - immediate sharp transistion from linear mode
>to full clean oscillation.

Oh, it oscillates in SPICE just fine- with gleeful, wholehearted
abandon, I'd say.  :)

It is the limitations in how that oscillation is controlled, and the
consequences for output ripple and regulation, that I find
unsatisfactory.  For me, that is.

G'nite...

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\08\28@035342 by Roman Black

flavicon
face
Dave Dilatush wrote:
>
> Roman Black wrote...
>
> >Dave Dilatush wrote:
>
> >> You've proposed it, now it's time to build and test it.  I look
> >> forward to seeing your results.
> >
> >Well, I haven't tested THIS circuit, but I have a few
> >stepper circuits...
>
> Build THIS circuit, please, and tell us your test results.  I'm sure
> you've got some very nice stepper motor drivers, but whatever they do
> is NOT relevant to the matter at hand.  It's what THIS circuit does,
> not what your stepper drivers do, that counts.


I apologise Dave! You are completely correct
and I got my efficiencies reversed. Now i've had
a bit of time to spend thinking about my
one-transistor circut it was fatally flawed in
another area too, it would oscillate, and around
the target voltage, but Q1 could never saturate
so efficiency was doomed even without the inductor
error. :o}

Ok so first attempt at the one-transistor SMPS
5v regulator is a complete failure... Back to the
drawing board. I've got some designs that oscillate
well and are efficient, that's fairly easy with one
transistor and 2 or 3 windings, but no output voltage
regulation...
-Roman

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email @spam@listservspam_OUTspam.....mitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@071345 by Dave Dilatush

picon face
Roman Black wrote...

>I apologise Dave! You are completely correct
>and I got my efficiencies reversed...

Absolutely no need to apologize, Roman; none whatsoever.

>Ok so first attempt at the one-transistor SMPS
>5v regulator is a complete failure... Back to the
>drawing board. I've got some designs that oscillate
>well and are efficient, that's fairly easy with one
>transistor and 2 or 3 windings, but no output voltage
>regulation...

Yes, it's tricky.  My opinion is that you either "pay over here" or
"pay over there"- that is, either a single inductor and several
transistors, or else a single transistor and a multi-winding inductor.
The former arrangement tends to be somewhat less expensive.

There is a school of thought that says, "the best design is the one
with the fewest parts" and I've gathered that philosophy is common
here on the PICLIST.  I've been in the "will do circuits for food"
racket, in one capacity or another, since 1966 and I abandoned that
philosopy long ago: in my experience, the best design is one which
includes EVERY component that is necessary and has NO components which
are not; every one has an important function, and that function can be
clearly described AND quantified.  If a part performs an important
function, removing it is folly; and if it doesn't perform an important
function, it didn't belong there in the first place.

A good design will be nearly incapable of doing anything other than
what its creator intended, and it is constrained to doing what it's
supposed to do by design- not by luck, or magic, or accident.

Ahem.  I'll stop ranting, now...

Cheers,

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email spamBeGonelistservEraseMEspammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@081658 by Alan B. Pearce

face picon face
>Roman Black wrote...

>[snip]

>>Please argue, I appreciate anyone who has made SMPS's for
>>25 years, as I know that you learned X things in those
>>years and I learned Y things in those years and it sure
>>beats arguing the football scores... :o)

>Roman, this circuit is like one of those "Roach Motel" thingies:
>energy can check into that inductor, but it can't check out.  Under no
>circumstances does ANY of the energy stored in the inductor EVER find
>its way to the output circuit.  Nary so much as a picojoule.

>Going back over my previous post, I don't see how I could clarify my
>explanation any; it seems pretty straightforward.

I suspect that it is not so much a case of apples and oranges, as apples and
pears that are being compared :)))

In the case of the stepper motor, energy is being checked out while the
diode shorts the inductor, because the magnetic field is the exit route
(check out desk) for the energy. Remember that in a motor there is
significant air gap in this path which will "release" a certain amount of
the energy until the next current replenishment cycle. This energy will be
used to hold the rotor stable, or continue pulling it into the next
rotational position (which ever was in progress when the current drive was
turned off).

In a SMPS the diode will cause a loss of efficiency as the stored energy is
now being dissipated in the resistive portion of the inductor, instead of
being transferred to the output load.

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservspamBeGonespammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@085045 by Russell McMahon

picon face
part 1 5347 bytes content-type:text/plain; (decoded 7bit)

OK - here are some actual test results from the divinely inspired super low
cost switching regulator.
I have included results below in an approximately unreadable table plus
attached an Excel spreadsheet. This is in Excel Version 2.1 format which
keeps the size small and should be readable by most other spreadsheet
programs.

Pasting the table below into a document with fixed width (eg Courier) font
makes it reasonably readable. (I used notepad in Windoze 95 as a trial and
it worked fine.)

Tests were carried out from about 8 volts to 30 volts in with loads of open
circuit, 1000 ohms and 100 ohms.
Vout was about 11.3 volts nominal giving output currents of about 11 mA and
110 mA for 1k and 100 r respectively..

These test were made on a circuit intended to operate at 600 mA out working,
1 amp max at 11 volts with 10 < Vin < 200 volts.
NO changes were made to the circuit for low power or lower voltage
operation.

The main switch is a P channel FET with RDSon = 1 ohm !!!
There is also additional current limiting circuitry present which does not
affect performance at all but may add slightly to current drain

The FET is driven by a high side driver (BC327) - removing the FET and using
this transistor as the main switch should produce superior results but I
haven't tried this yet - another night.

The Inductor is relatively immense as it is designed to allow reasonably low
switching times at Vin = 200 volts. This uses over 200 turns on a Micrometal
I-really-can't-remember-and-its-downstairs-and-its-late-and-nobody-will-care
toroid. I'll dig up the size and spec if anyone is interested. (Just wind
some turns on a toroid that will fit over a pencil and try it. Buck
converters can be pretty tolerant  :-) .)

The only "plus" for this circuit over one purpose designed for low power
levels is that this uses a BYV28 flyback diode which is liable to give less
losses than a diode liable to be used at the lower power levels. For very
low current applications a super cheap diode that would work OK would  be a
1N9148.(!)(Probably the world's most common and cheapest semiconductor).


Results:
**********

NO LOAD
++++++++

Input current ranged from 0.22 mA at Vin = 8 volts to 0.65 mA at Vin = 30v.
Note the slight drip in Iin at Vin = 11.5 volts as the regulator switches
from linear to buck modes.

This idle current is a significant part of the efficiency losses when 100r
load is used and optimisation for low current levels should produce superior
results.

1000 r LOAD
+++++++++

Efficiency ranged from

   over 90% at Vin <= 13 volts
   with about 3% of losses being due to idle current
to
   55% at Vin = 30 volts
   with idle current contributing about 8% of losses.

A linear regulator at Vin = 30 volts would be about 38% efficient giving
losses higher by a factor of (100-38)/(100-55) = 1.38 times.
The regulator is more efficient than a linear regulator for all Vin >= 13v
with best ratio at Vin = 25 volts.

100r LOAD
++++++++++

Efficiency ranged from

   over 90% at Vin <= 13 volts
to
   about 76% at Vin = 30 volts

Idle current is an insignificant part of losses at this load.

A linear regulator at Vin = 30 volts would be about 38% efficient giving
losses higher by a factor of (100-38)/(100-55) = 1.38 times.
The regulator is more efficient than a linear regulator for all Vin >= 13v
with best ratio at Vin = 25 volts.

Efficiency compared to a linear regulator increased with increasing Vin and
hadn't topped out by Vin = 30 volts.
Ratio of Eff-buck/Eff-linear at 30 volts = ~~ 2:1 at 76% versus 38%
Relative dissipations would be (100-38)/(100-76) = 2.6 times as high in the
linear regulator.

COMMENTS
***************

Improvements will be obtained by

- Optimising the design for the target load range

- ditto target Vin range

- Changing nasty FET to a small bipolar.
(The FET is used because of its suitability at high voltages and higher
power levels - this is not relevant here).

- Changing inductor



I'll try to get to testing a more low power optimise design shortly.


regards


       Russell McMahon




_____________________________________

*** VIEW WITH FIXED WIDTH FONT (eg COURIER) ***


 Low cost buck converter
 Non optimised design


   No Load    1 K load     100 r load

 Vin  Vout Iin mA  Vout Iin mA Pout mW Effic %  Vout Iin Pout mW Effic %
       = Iout mA      = Iout/10
 8  8 0.22           mA
 9  9 0.25
 10  10 0.28
 11  11 0.32
 11.5  11.31 0.28
 12  11.31 0.28  11.30 11.5 128 92.9%  11.10 111 1232 92.5%
 13  11.31 0.3  11.31 10.8 128 91.1%  11.29 110 1275 89.1%
 14  11.31 0.32
 15  11.32 0.34  11.31 10.1 128 84.3%
 16  11.32 0.36  11.31 9.8 128 81.7%  11.22 99 1259 79.5%
 18  11.32 0.4  11.32 9.3 128 77.0%
 20  11.32 0.44  11.32 8.8 128 73.1%  11.29 79.5 1275 80.2%
 25  11.33 0.54  11.34 7.1 129 72.1%  11.30 66.3 1277 77.0%
 30  11.34 0.65  11.33 7.8 128 54.6%  11.31 56.3 1279 75.7%

 These test were made on a circuit intended to operate at 600 mA out
working, 1 amp max at 11 volts
 with 10 < Vin < 200 volts
 The main switch is a P channel FET with RDSon = 1 ohm !!!
 No changes were made to any components.
 There is also additional current limiting circuitry present which does not
affect performance at all but may add slightly
 to current drain




part 2 4966 bytes content-type:application/vnd.ms-excel; (decode)

part 3 136 bytes
--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistserv@spam@spamspamBeGonemitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@171436 by Russell McMahon

picon face
part 1 1931 bytes content-type:text/plain; (decoded 7bit)

> From: "Dave Dilatush" <.....dilatush@spam@spamEraseMEHOME.COM>
> I wrote...
>See attachment.
> What attachment?  I don't see any attachment...
> Oh, THIS attachment.

Just a note about this circuit (derived by Dave from the original relay
driver cct).
It looks really excellent but I can see one potentially significant problem.

R9 provides positive hysteresis (that word again) by raising the effective
reference voltage on the base of Q1 when the pass transistor Q2 is on and
lowering the reference when Q2 is off.

In the original circuit the extent of this effect was limited by splitting
R9 into two resistors and providing a clamp zener to ground in the middle.
This limited the maximum amount that the reference voltage could be
increased by when the pass transistor Q2 was turned on.

It is traditional in step down converters to "feed forward" the effect of
the
input voltage to reduce the on time as input voltage rises. (This effect
acts in addition to the voltage feedback loop). Unfortunately, this circuit
does the opposite - as Vin rises the positive increase in reference when Q2
is turned on INCREASES - the faster you go the faster you go :-).

Presumably the split resistor and zener in the original were placed there as
a result of practical experience by the original designers. The problem (if
there is a problem) would increase as the range of Vin decreased. For modest
ranges of Vin (eg say 20 to 28 volts in an automotive application) the
effect may be tolerable.

What is ideally wanted to drive is a voltage which increases at a decreasing
rate as Vin increases. In such a simple circuit this is probably a bit much
to ask for. The divided resistor plus zener (add 1 + ?5 cost units) is a
simple compromise. Whether the result is bad enough to be important in
practice depends on specific circumstances.


regards


           Russell McMahon



part 2 5042 bytes content-type:image/gif; (decode)


part 3 136 bytes
--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email .....listservRemoveMEspammitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@173119 by hard Prosser

flavicon
face
part 1 3896 bytes content-type:text/plain; charset=us-ascii
Russel
Yes, that was exactly the reason the zener & additional resistor were
positioned there. I did also get compensation by adding a resistor from the
emitter of the first transistor to the incoming rail - this provided quite
good compensation also but prooved to be somewhat "select on test". The
actual optimal resistor value was different for different transistors  -
and different again from the spice simulations.
Since I require an input voltage range of at least 18-70V, and good
immunity from component variations, I elected to go for the zener option to
limit the +ve hysterisis effect. I did try a standard diode here also -
can't remember what I am using in the latest version but I think a BAV99
double diode worked - probably with modified resistor values.

There is a certain amount of RF filtering added as well - but not shown on
my diagram.

It may be important to note that the circuit will always exhibit a
reasonable amount of ripple on the output. This is required as it is this
ripple that is used to determine the condusct/block operation of the main
switch. Reducing the ripple will increase the operating frequency and also
the losses. The addition of an LC filter on the output (or an LDO
regulator!) may assist here.

Richard P




                   Russell
                   McMahon               To:     .....PICLISTSTOPspamspam@spam@MITVMA.MIT.EDU
                   <apptech@CLEAR        cc:
                   .NET.NZ>              Subject:     Re: [EE]: Design Challenge - low power step down switching              regulator
                   Sent by: pic
                   microcontrolle
                   r discussion
                   list
                   <PICLIST@MITVM
                   A.MIT.EDU>


                   29/08/01 09:15
                   Please respond
                   to pic
                   microcontrolle
                   r discussion
                   list






> From: "Dave Dilatush" <dilatushEraseMEspam@spam@HOME.COM>
> I wrote...
>See attachment.
> What attachment?  I don't see any attachment...
> Oh, THIS attachment.

Just a note about this circuit (derived by Dave from the original relay
driver cct).
It looks really excellent but I can see one potentially significant
problem.

R9 provides positive hysteresis (that word again) by raising the effective
reference voltage on the base of Q1 when the pass transistor Q2 is on and
lowering the reference when Q2 is off.

In the original circuit the extent of this effect was limited by splitting
R9 into two resistors and providing a clamp zener to ground in the middle.
This limited the maximum amount that the reference voltage could be
increased by when the pass transistor Q2 was turned on.

It is traditional in step down converters to "feed forward" the effect of
the
input voltage to reduce the on time as input voltage rises. (This effect
acts in addition to the voltage feedback loop). Unfortunately, this circuit
does the opposite - as Vin rises the positive increase in reference when Q2
is turned on INCREASES - the faster you go the faster you go :-).

Presumably the split resistor and zener in the original were placed there
as
a result of practical experience by the original designers. The problem (if
there is a problem) would increase as the range of Vin decreased. For
modest
ranges of Vin (eg say 20 to 28 volts in an automotive application) the
effect may be tolerable.

What is ideally wanted to drive is a voltage which increases at a
decreasing
rate as Vin increases. In such a simple circuit this is probably a bit much
to ask for. The divided resistor plus zener (add 1 + ?5 cost units) is a
simple compromise. Whether the result is bad enough to be important in
practice depends on specific circumstances.


regards


           Russell McMahon


(See attached file: RPmodified.gif)



part 2 5042 bytes content-type:image/gif; (decode)


part 3 136 bytes
--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservspamspamBeGonemitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@174842 by Dave Dilatush

picon face
Russell McMahon wrote...

>For very
>low current applications a super cheap diode that would work OK would  be a
>1N9148.(!)(Probably the world's most common and cheapest semiconductor).

1N9148?

>I'll try to get to testing a more low power optimise design shortly.

What would be most interesting is- as was originally proposed in the
challenge- a version with 5.0 volt output suitable for PICs.

>*** VIEW WITH FIXED WIDTH FONT (eg COURIER) ***

I did.  The table was a jumble.

Cheers,

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email spamBeGonelistservKILLspamspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body


2001\08\28@191340 by Dave Dilatush

picon face
Russell McMahon wrote...

>It is traditional in step down converters to "feed forward" the
>effect of the input voltage to reduce the on time as input
>voltage rises.

In the context of switching regulators, feedforward is a technique
used (when it is needed and appropriate) for a very specific and
narrow purpose: to improve a regulator's line transient response.  
The essence of this technique is to provide a means whereby sudden
voltage changes at the input are quickly, but only approximately,
cancelled out by the feedforward mechanism, while allowing the
feedBACK mechanism (notice the emphasis) to take care of the
longer-term input variations and maintain the output voltage against
changes in load.  
Appropriate use of feedforward techniques can make the design of a
feedBACK (emphasis, again) system easier by partially relieving it of
the burden of having to make large adjustments very rapidly in
response to sudden process changes.  
In loose, intuitive terms, the feedforward mechanism allows the
feedback mechanism to concentrate more on "fine-tuning" the output and
less on rapidly compensating for sudden input changes.
Other than for this particular purpose, I have never seen feedforward
used in a switching regulator.  
And if I had, I would have considered it a Band-Aid applied by the
designer to cover up some shortcoming in his feedback system.  As
Richard pointed out, these appliques tend to be a "select on test"
sort of thing.

>Presumably the split resistor and zener in the original were placed there as
>a result of practical experience by the original designers.

In Richard's application, as in yours, the circuit has to work with an
extremely wide range of input voltages- 4:1 or even more.

In my own work I seldom have to deal with more than a +/- 30% input
variation, so I eliminated those components in my quest to see just
how simple I could make Richard's circuit.  
Obviously, these components perform an important function (that is,
stabilizing the hysteresis feedback voltage) when huge input
variations are expected and should, when needed, be included.

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservspam_OUTspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body


2001\08\29@030416 by Roman Black

flavicon
face
Russell McMahon wrote:
>
> > From: "Dave Dilatush" <spamBeGonedilatush@spam@spamHOME.COM>
> > I wrote...
> >See attachment.
> > What attachment?  I don't see any attachment...
> > Oh, THIS attachment.


That circuit is looking very cheap. Is 2.2mH enough?
What speed does it operate at? It would be nice to be
able to use one of those tiny 2.5mH inductors, the
resistor sized green ones that are available for
20c each etc.

I think a circuit similar to this could be adapted for
low power use, ie 9v battery to 5v PIC applications,
or 12v or 5v low power use.

With that type inductor the thing would fit in about
a 1.5cm x 1.5cm footprint and be very handy for some
PIC applications.

What would be great if someone could build and refine
some, ready for any newbie to build, with a standard
parts list etc. Like pre-made ones for:

9v to 5v, 0 to 10mA
9v to 5v, 0 to 25mA
12v to 5v, 0 to 10mA
12v to 5v, 0 to 30mA.

I think within these small ranges the price, size and
efficiency of each design could be optimised. I might
have a bash at it this weekend.
-Roman



>     ---------------------------------------------------------------
>
>                               Name: RPmodified.gif
>                Part 1.2       Type: GIF Image (image/gif)
>                           Encoding: base64

--
http://www.piclist.com hint: To leave the PICList
RemoveMEpiclist-unsubscribe-requestEraseMEspamKILLspammitvma.mit.edu


2001\08\29@061002 by Dave Dilatush

picon face
Roman Black wrote...

>That circuit is looking very cheap.
Cheap + simple + efficient + good control = elegant.  Richard's
circuit is definitely a winner.

>Is 2.2mH enough?

That, or even less, should do fine.  The deciding factors will be how
high a peak current the transistor can deliver (Ipeak increases as
inductance decreases) and how fast can it turn off.  I haven't
simulated this circuit with a large range of inductance values, but I
will before this weekend and give some results.

>What speed does it operate at?
The higher the output current, the higher the oscillation frequency.
At Iout = 20 mA, it looks like a couple of kilohertz.

>It would be nice to be
>able to use one of those tiny 2.5mH inductors, the
>resistor sized green ones that are available for
>20c each etc.

They should work OK, provided they don't have too high a series
resistance (I'll check this out, too) and they don't saturate at the
currents involved.

>I think a circuit similar to this could be adapted for
>low power use, ie 9v battery to 5v PIC applications,
>or 12v or 5v low power use.

In doing my simulations, I've tried input supply voltages down to 9V
with this circuit (using different resistor values, of course) and it
should work just fine.

>What would be great if someone could build and refine
>some, ready for any newbie to build, with a standard
>parts list etc. Like pre-made ones for:
>
>9v to 5v, 0 to 10mA
>9v to 5v, 0 to 25mA
>12v to 5v, 0 to 10mA
>12v to 5v, 0 to 30mA.

I can give us a starting point on that; I might get at it this evening
after I get home from work.

>I think within these small ranges the price, size and
>efficiency of each design could be optimised. I might
>have a bash at it this weekend.

As might I.  This weekend is our long end-of-summer holiday up here in
the States but alas, I'm grounded: my youngest son is in college and
with $$$out = $$$in, I'm not going anywhere.  Might as well fiddle
with hobby stuff. :(

One thing people should keep in mind about ALL of the switching
regulators we've talked about in this thread: they're going to have
output voltage tolerances which are pretty wide.  The Zener diodes
themselves will have a 5% tolerance, with other factors (transistor
Vbe's, for example) making the error band even wider.  These are NOT
precision devices.

The other limitation to keep in mind is that these things are, after
all, switching regulators: they WILL have ripple in the output
voltage.  Powering an A/D converter off the output of one of these
things might not give very good results.

Both these limitations can be overcome by setting the output a little
high and then following the switcher with a good, low-dropout linear
regulator chip as a post-regulator to clean things up.  I like
National's LP2951 which, in the 8-pin version, also gives me a
reliable undervoltage-reset output.

Dave

--
http://www.piclist.com hint: To leave the PICList
spamBeGonepiclist-unsubscribe-requestspam_OUTspamRemoveMEmitvma.mit.edu


2001\08\29@120827 by Olin Lathrop

face picon face
>>
Roman, this circuit is like one of those "Roach Motel" thingies:
energy can check into that inductor, but it can't check out.  Under no
circumstances does ANY of the energy stored in the inductor EVER find
its way to the output circuit.  Nary so much as a picojoule.

Going back over my previous post, I don't see how I could clarify my
explanation any; it seems pretty straightforward.
<<

Roman, I agree with Dave's analisys too.  Here's another way to look at it:

One characteristic that differs between linear and switching regulators is
where the output current comes from.  In a linear regulator, all the output
current comes from the input supply.  In fact, except for the small current
used to power the regulator itself, the output current equals the input
current.

A switching regulator achieves higher efficiency because part of the output
current comes from elsewhere other than the input supply.  This extra
current usually comes thru the inductor via the flyback diode while the
switching element is off.  A theoretical perfect regulator has the input
current inversely proportional to its input voltage.  In your design, ALL of
the output current ultimately comes from the input supply.  Regardless of
what else goes on or why, it simply can't beat a linear regulator for that
reason alone.  All the rest of the circuit just shuffles around where the
energy gets dissipated in dropping the same current from the input to the
output voltage.  In a normal linear regulator, the energy gets dissipated in
the pass element.  In your circuit it mostly gets dissipated in the flyback
diode accross the inductor and in the resistance of the inductor itself.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, .....olinspamRemoveMEembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
piclist-unsubscribe-requestspam@spam@mitvma.mit.edu


2001\08\29@152830 by jamesnewton

face picon face
This (excellent) thread with all its circuit GIF, etc.. prompted me to
finally get the darn mime decoder and image content display working on the
piclist.com archive.
www.piclist.com/techref/postbot.asp?by=time&id=piclist\2001\08\26\170
913a

I don't know what to do with application/octet-stream so the PDF and XLS
files are a wash... anybody know?
Please reply offlist if you do...

---
James Newton, Admin #3 EraseMEjamesnewtonRemoveMEspamSTOPspampiclist.com
1-619-652-0593 VM 1-208-279-8767 FAX
PIC/PICList FAQ: http://www.piclist.com or .org

{Original Message removed}

2001\08\29@163443 by Russell McMahon

picon face
> That circuit is looking very cheap. Is 2.2mH enough?
> What speed does it operate at? It would be nice to be
> able to use one of those tiny 2.5mH inductors, the
> resistor sized green ones that are available for
> 20c each etc.
.......................

> I think within these small ranges the price, size and
> efficiency of each design could be optimised. I might
> have a bash at it this weekend.
> -Roman


Try building a version of my circuit as well for comparison. Both are very
simple and cheap.
The regulation on my circuit is extremely good and I suspect the ripple may
be lower.

For very small power levels small potted inductors may suffice but these
tend to have very low saturation currents. You can get quite small open form
ferrite bobbins and small powdered iron toroids which likely to be better
suited. Core volume is roughly related to power handling capability.




       Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
RemoveMEpiclist-unsubscribe-requestKILLspamspamTakeThisOuTmitvma.mit.edu


2001\08\30@044521 by Russell McMahon

picon face
>It is traditional in step down converters to "feed forward" the
>effect of the input voltage to reduce the on time as input
>voltage rises.

/In the context of switching regulators, feedforward is a technique
/used (when it is needed and appropriate) for a very specific and
/narrow purpose: to improve a regulator's line transient response.
/The essence of this technique is to provide a means whereby sudden
/voltage changes at the input are quickly, but only approximately,
/cancelled out by the feedforward mechanism, while allowing the
/feedBACK mechanism (notice the emphasis) to take care of the
/longer-term input variations and maintain the output voltage against
/changes in load.

PRESCRIPT: I'm sure that you are aware of the details of the circuit's
operation and its fine points so I think that the difference is mainly one
of terminology between us. However, the conversation looks interesting so
here we go ...

Indeed. Dynamic response is indeed the major target of fedforward. This is
not to say that it isn't, cannot be, or even shouldn't be used to respond to
more static conditions. As Saint Albert is quoted as saying (I'm always
suspicious of the vast number of quotes attributed to Einstein)

    "Everything should be made as simple as possible,
       but not simpler."   --Albert Einstein

As I imagine you will agree, there is no holy writ about what techniques
should or must be used in electronics. As long as the pertinent laws of
physics are understood and obeyed all will be well. Many designs incorporate
"loops within loops" for various reasons such as a current feedback and
voltage feedback loop within the same design.

/Appropriate use of feedforward techniques can make the design of a
/feedBACK (emphasis, again) system easier by partially relieving it of
/the burden of having to make large adjustments very rapidly in
/response to sudden process changes.

Agree.
The point I was making was that, in the present application, the hysteresis
mechanism is applying POSITIVE feedback to the control loop when NEGATIVE
feedback is what the main control loop is trying to provide. ie here a
rising input voltage tends to produce a rising output voltage and the
regulator will produce a shorter duty cycle to compensate. However, the
hysteresis resistor from the hot side of the buck coil helps to increase the
duty cycle with increasing input voltage at a time when we want it to be
decreasing. I was concerned with the relatively static case but In a dynamic
situation this feed forward would also be of the opposite sense to that
desired.

/Other than for this particular purpose, I have never seen feedforward
/used in a switching regulator.

I have. Mayhaps it is less usual in modern designs as ICs become more able
to integrate all desired functions.

/And if I had, I would have considered it a Band-Aid applied by the
/designer to cover up some shortcoming in his feedback system.

As above.
If you can simplify a design and/or improve its performance overall and to
do so you use a mechanism such as static feed forward then I sugesst that
Einstein's adage firmly applies.

/ As Richard pointed out, these appliques tend to be a "select on test"
/ sort of thing.

I think the select on test nature in this case is due to the fact that , as
noted above,  this resistor applies the opposite sense of "feed forward" to
that desired. This is entirely acceptable in such a simple and elegant
design if it meets the user's needs in all cases - such a circuit makes some
compromises to achieve its simplicity. One way of preventing this
undesirable increasing of the reference voltage would be to provide a
voltage which toggles to a fixed high value when the switch is on and to
ground when it is off. This would add complexity and cost - the zener
clamped resistor is a reasonable compromise and your zener-less
simplification is OK if transient response and restricted input range are
OK - as they may be in many cases. It will be interesting to examine
transient response of the various designs proposed so far. I will happily
admit that I do not (yet) know the formal transient performance of "my"
design - what I do know is that it is entirely adequate in the application I
have primmarily put it to so far. I also suspect that it will prove adequate
in many other cases but when I build the low power version I have been
threatening I will try some step load and input variations and report the
results.  I'll also compare the simulated and actual results and comment.


regards


               Russell McMahon

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\30@051548 by Roman Black

flavicon
face
Russell McMahon wrote:

> > I think within these small ranges the price, size and
> > efficiency of each design could be optimised. I might
> > have a bash at it this weekend.
> > -Roman
>
> Try building a version of my circuit as well for comparison. Both are very
> simple and cheap.
> The regulation on my circuit is extremely good and I suspect the ripple may
> be lower.

Hi Russell, your circuit has much more sophisticated
looking feedback but the added complexity is probably
going to make it less attractive for the newbie.
2 transistors + a fet, 2 zeners, etc. Fets and zeners
scare newbies. :o)

We've seen a few newbie's making PIC projects with a
9v battery, and I think a super-simple SMPS regulator
just for that purpose would be very useful, even though
it won't perform nearly as well as your pickbukmin2
circuit. Efficiency is probably more important than
regulation for many of these circuits. I've reduced
Richard's circuit to 4R, 2C, 2Q, 2 diodes, and that's
including an RC filter on its output.

For simplicity I tried to build a single transistor
circuit, but the only thing that looked decent needed
3 separately wound coils on the inductor, and you can
imagine the newbie problems there with coil polarities
etc. No prizes for one transistor simplicity if the
thing is a real pain to build. :o)

> For very small power levels small potted inductors may suffice but these
> tend to have very low saturation currents. You can get quite small open form
> ferrite bobbins and small powdered iron toroids which likely to be better
> suited. Core volume is roughly related to power handling capability.

Yeah, I have a heap of toroids and formers here, but
again I think for total newbie proofing a cheap
pre-made inductor, and a variation of Richard's
2-transistor circuit might be best. Especially
appealing to me would be a 9v to 5v 10mA version
using one of those tiny RF style inductors that look
like resistors. I could have used a circuit like
that many times!!

Most "hobby" style shops sell small potted inductors.
Catalogue shows:
<0.8mH at 100mA max, 65 cents 6x8mm
>0.8mH at 50mA max, 65 cents 6x8mm
18mH to 120mH 100mA max, 65 cents, 10x14mm
and of course the tiny RF style ones, maybe 10mA??
The less exotic the inductor the better. :o)

Another option might be to use a very simple air-wound
coil of X turns, this is newbie proof and cheap. It
might push the switching speed through the roof, with
all the associated problems, but for low power stuff
the little transistors are very fast switchers and maybe
efficiencies could be maintained.

I'm looking forward to having a hardware fiddle, and
it will be nice to see what Dave D. finds with the
simulator too.
-Roman

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\30@052631 by Roman Black

flavicon
face
Dave Dilatush wrote:

> >It would be nice to be
> >able to use one of those tiny 2.5mH inductors, the
> >resistor sized green ones that are available for
> >20c each etc.
>
> They should work OK, provided they don't have too high a series
> resistance (I'll check this out, too) and they don't saturate at the
> currents involved.

If you found 2kHz with the other circuit,
it should be possible to decrease the inductor
down to 0.1mH or maybe much less than that,
with the tiny RF inductors the wire size is
larger with the lesser values, and increasing
the speed should help with saturation issues.

A lot of PIC projects only need a few mA, and
a nice alternative to a 7805 regulator chip
or resistor/zener would be great for 9v battery
operated PIC gizmos.


> >9v to 5v, 0 to 10mA
> >9v to 5v, 0 to 25mA
> >12v to 5v, 0 to 10mA
> >12v to 5v, 0 to 30mA.
>
> I can give us a starting point on that; I might get at it this evening
> after I get home from work.

I'm excited to see what you come up with.
It's also pretty interesting that no piclister
seems to have done this before?? We're always
seeing posts about sleep mode and battery
operated PIC stuff, and these SMPS circuits
should be nicely compatible with sleeping
PICs, giving the best of both worlds.

> One thing people should keep in mind about ALL of the switching
> regulators we've talked about in this thread: they're going to have
> output voltage tolerances which are pretty wide. These are NOT
> precision devices.

Ha ha! Neither are batteries! ;o)
-Roman

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\30@053350 by Roman Black

flavicon
face
Olin Lathrop wrote:

> Roman, I agree with Dave's analisys too.  Here's another way to look at it:
>
> One characteristic that differs between linear and switching regulators is
> where the output current comes from.  In a linear regulator, all the output
> current comes from the input supply.  In fact, except for the small current
> used to power the regulator itself, the output current equals the input
> current.

Ha ha! Yeah I posted that about 3am and realised
the next morning I got my efficiencies backwards.
Slow decay is great for keeping the field and current
high in a motor winding, where the circuit being
discussed needed the exact opposite effect.
:o)
-Roman

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\30@141426 by Peter L. Peres

picon face
> That, or even less, should do fine.  The deciding factors will be how
> high a peak current the transistor can deliver (Ipeak increases as
> inductance decreases) and how fast can it turn off.  I haven't

More exactly Wl = L*I^2/2 and at constant frequency I increases with the
sqrt of the L decrease, after neglecting other factors.

The problem is to get small chokes that are rated the specced inductance
at the DC current that a buck regulator passes through them. This usually
rules out normal small inductors, unless the current is very low. A
saturated inductor core can have 1/10 or less of the nominal inductance
and cause all sorts of trouble.

I once quickly put a 2W buck regulator together using parts from the junk
box and I remember that I tried at least 5 inductors that looked suitable
until I found one that was well behaved, and I had to rewind it with more
wire to be happy. It was an 'open' choke and it radiated like mad. The
strange part is that some of these seem to be used in office equipment
(unshielded plastic case) by some manufacturers, and the units bear FCC
class A or B compliance stamps (real ones, I checked).

Peter

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\30@181649 by Dave Dilatush

picon face
Roman Black wrote...

>If you found 2kHz with the other circuit,
>it should be possible to decrease the inductor
>down to 0.1mH or maybe much less than that,
>with the tiny RF inductors the wire size is
>larger with the lesser values, and increasing
>the speed should help with saturation issues.

With this thing arranged for operation off a 9V battery, expect about
20 KHz with a 2.2 mH inductor and a 20 mA load.  Note that in a buck
regulator like this, the relationship between inductance and frequency
isn't simple: the frequency depends on the amount of hysteresis
designed into the switch, the load current, and the output capacitance
as well.

If you'd like to try this circuit with 9V input, try making the
following adjustments to RPModified.gif that I posted on 8/27:

R2 = 1.0K
R9 = 100K
R5 = 3.3K
R10 = 6.8K

Experiment as suits you, and see what results you get.

>A lot of PIC projects only need a few mA, and
>a nice alternative to a 7805 regulator chip
>or resistor/zener would be great for 9v battery
>operated PIC gizmos.

If you want a good alternative to a 7805 for operating a PIC off a 9V
battery, use an LDO regulator chip like National's LP2951.  At 20 mA
it will work down to an input voltage of only 5.25 volts, at which
point your 9V battery is pretty much sucked dry.  The LP2951 only
takes a hundred microamps or so for its own use, compared to a
milliamp or two (IIRC) for a 7805.

The LP2951 also provides a very solid low-voltage inhibit output that
can be used as a PIC reset signal.

If you want the maximum possible battery life from a 9V unit at low
currents, I'd suggest using an IC switching regulator like Maxim's
MAX639.  It will be a lot more efficient than any of these circuits
we're playing around with: about 90% efficiency at 1 mA output, AFAIR.

Dave

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\30@195003 by Dave Dilatush

picon face
Russell McMahon wrote...

>I will happily
>admit that I do not (yet) know the formal transient performance of "my"
>design...

Somehow, after going 'round and 'round in circles with you on this
design for over a week now, this doesn't surprise me a bit.

Since you've decided- now that we've mentioned it in passing- to do a
bit of testing re transient response, here are two other things to
look at while you're at it:

First, I strongly suggest you do something about that Zener diode and
its low operating current.  At the very least, test a large number of
units (with various manufacturer's diodes) at Tmax and verify your
regulation doesn't go all to pot.  Operating Zener diodes at only a
few score microamperes reverse current is asking for BIG trouble if
you're expecting them to maintain their specified breakdown voltage,
even at room temperature.

And second, take a close look (I'd suggest a VERY close look) at the
effects of output capacitor ESR.  Because of the unique manner in
which this circuit operates, I think you're going to have major
problems keeping your operating frequency within tolerable bounds if
output capacitor ESR goes over a few milliohms.

Dave

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\08\31@010055 by Russell McMahon

picon face
Russell McMahon wrote...
>I will happily
>admit that I do not (yet) know the formal transient performance of "my"
>design...

/Somehow, after going 'round and 'round in circles with you on this
/design for over a week now, this doesn't surprise me a bit.
/Since you've decided- now that we've mentioned it in passing- to do a
/bit of testing re transient response, here are two other things to
/look at while you're at it:

All statements of "fact" below may optionally be prefixed with "IMHO :-) "
if desired.

Someone with a thinner skin than mine might just suspect you were trying to
be rude to me but fortunately my skin is getting thick and wizened. I have
noticed us going in a few circles lately but had the distinct impression
that I was following someone else's footprints at the time :-).

Re transient response - what I referred to was "formal transient
performance". I do not have theoretical numerical predictions of the
response - what I can tell you is how it has performed in practice. The
converter forms a small but important part of a larger product. Initially I
built numerous prototypes. These were subjected to tests aimed at exceeding
real world conditions as thoroughly as reasonably possible. While the real
environment would not reasonably conceivably allow large input transients
the converter has been subjected to anything you can reasonably do with a
lab power supply and various pieces of wire, string, duct tape and the other
contents of a well equipped electrical workshop. It has also been through
extensive operation within the target equipment in real conditions both here
and in Taiwan. Then it was trialled in about 50 advance test models
(supplied for initial evaluation and to allow torture testing by the systems
integrator). It is now in limited production and so far 2000+ of these have
been built, production tested and accepted by the Customer. So far none of
the converters has misperformed in any way. There have been no failures of
the converter (or, I'm pleased to report,  of the other electronics that I
designed), Due to the relatively "gentle nature" of  buck converters
generally * this design has operated up to the specified voltage rating of
the power FET - not something I would ever wish to see done in real world.
(On such occasions I kept waiting to see what colour the escaping smoke
would be but so far none has escaped captivity).

Ideal design? - No.
Met objectives so far? - Yes.
Useful elsewhere? -  Certainly!
Best design in all cases? - Certainly not!!!


/First, I strongly suggest you do something about that Zener diode and
/its low operating current.  At the very least, test a large number of
/units (with various manufacturer's diodes) at Tmax and verify your
/regulation doesn't go all to pot.  Operating Zener diodes at only a
/few score microamperes reverse current is asking for BIG trouble if
/you're expecting them to maintain their specified breakdown voltage,
/even at room temperature.

"Doing something" about the zener would conceivably somewhat improve the
regulation but the spirit of the original design requires that any such
change is a low cost one. Adding a resistor to ground from the anode of
ZBUK1 (to increase zener current) would assist. Moving PBUK1 from the lh
side to the rh side of RBUK1 and reducing its value substantially would
achieve the same result. Neither change alters the core concept.

DISCUSSION ONLY:    You made this point several times before - then as now
your point is taken.
This is not using a zener in its more normal reference manner. As already
discussed, there will be some 2nd order effects due to operating it further
down the "knee" than usual. BUT - and I'm sure you know this - the rated
voltage for a given zener diode is in fact an arbitrary value set by the
desire to push it as high up its exponential operating curve as possible
(steeper = better here as you note) and the desire to keep the current low
compared to eg device maximum ratings and to minimise self heating. If you
plotted the zener voltage/current curve with say 1mA or 100 uA as full scale
voltage rather than the more normal 10's or 100's of mA as full scale then
the shape of the curve would be remarkably familiar - this is what
exponential curves are about - the rate of increase of the value is
proportional to the value so the curve has the same intrinsic shape at all
scales.


/And second, take a close look (I'd suggest a VERY close look) at the
/effects of output capacitor ESR.  Because of the unique manner in
/which this circuit operates, I think you're going to have major
/problems keeping your operating frequency within tolerable bounds if
/output capacitor ESR goes over a few milliohms.

I believe this is much less critical than you suggest. (But see rough
calculations at end).
You didn't say whether you had determined this from your own consideration
of the circuit or from a simulation - if so a commentv on results obtained
would be interesting. This point can be demonstrated (or not) in practice in
due course if desired. I have used a wide range of capacitor values in this
location including devices obtained from a range of sources (here and in
Taiwan). Operating frequency is not critical in this design and varies over
a very wide range with load (unless a minimum load is set). This tends to be
true of all buck converters unless they have a "sleep" mode where they turn
off completely in low current operation. (This circuit does in fact "sleep"
but less formally than in a circuit with an IC based controller where such a
function can be explicitly implemented by a purpose designated "building
block" - you can only do so much with 3 transistors.)  A full t_off
derivation would be remarkably complex for such a simple circuit (function
of Vin, Vout, load current, Inductance, Cout, R_effective for various
components (inductor, switch, output capacitor, more ...), prior cycle
history and where in the last cycle the converter was when Vout_design was
reached and turnoff started. The latter is important as the amount of
voltage rise after turnoff depends on the portion of energy stored which is
delivered AFTER turnoff starts which depends substantially on where in the
cycle this occurs. This factor alone will swamp quite substantial changes in
capacitor ESR. Certainly, output capacitor capacitance alone has a
substantial affect on toff and the ESR of Cbuk2, which affects dVout sensed
due to I_Cbuk2 dropping across the ESR during charge and discharge.
A rough theoretical calculation suggests that these effects are of a similar
order (unless I lost a few powers of 10 along the way).
. [[[ With a few unstated assumptions: If ALL ring energy was saved in C
then L * Ipk^2/2 = C(Vend^2-Vstart^2) = C(Ve-Vs)(Ve+Vs) ~~ 2VavgC  * dV
or dV = ~~~ (L * Ipk^2) / (4 * Vavg * C)
dV due to ESR is in the order of 2 * Ipk x ESR. Equating
ESR =~~ Li/*8VC) for similar effects.
No doubt someone will point out where/if I went astray - not something i
want to spend too much time on just now. Maybe later after some more
practical results demonstrated.  ]]]


Summary:

Low cost.
Simple design.
Breaks several "rules"
Wide frequency range.
Reasonable performance.
Works well in practice (so far).
Docile and forgiving behaviour (so far).

Core design could be improved by better circuit design without changing core
concept.
Core design could be improved by changing core concepts, almost certainly
with increased cost.

regards


               Russell McMahon

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\31@091213 by Russell McMahon

picon face
Decided to have a quick trial of a lower power version of the simple
regulator, taking note of some of Dave Dilatush's comments.

I decided to try a nominal 12 volt in, 5 volt out 5 mA unit as suggested by
Roman and Dave. I don't see too much point to such a low power limited drop
design EXCEPT for battery powered applications where long battery life is
desirable. In such cases I suspect that a "proper" controller would be
desirable due to the extra efficiency probably achieved by the better
attention to proper design which can be given by using such an IC. Still
interesting though.

I also tried the circuit at 50 ma output and ran Vin over 5 to 30 volts. .

The circuit did NOT scale simply and easily down to a very low power low
voltage design. With some playing (and design :-) ) it will work but certain
care needs to be taken. Used BC337 & BC327 transistors as required.

1.    Dave notes the low zener current is undesirable.
I decided to try eliminating the zener. I instead divided the output voltage
down to 0.6v and used this to turn on the "turn off" transistor, I used a
diode to drive these two resistors so that I could also add a hold off
capacitor at this point.



____________________Vout__________
                          |
                         _
                         V Diode
                         --
                          |-----------
                         R            |
   -----               R 1      ===  C
            \|           R            |
      Q   /|--------- |             |
          V            R            |
           |             R2          |
           |             R            |
           |              |             |
-----------------------------------

Vout rises above Vdesign.
R1/R2 divide Vout-Vdiode so

   (Vout-0.6) * R1/ (R1+R2) = 0.6v to turn on transistor Q.

Capacitor C retains Vout when Vout falls with time constant approx = C * R1

This works!
It eliminates the zener and replaces it with the transistor Vbe - arguably
out of the frying pan and into the fire.

C gives the ability to increase turn off time and therefore circuit
"hysteresis" / ripple voltage which as Dave notes is unusually derived in
this circuit.

I used C = 0.1 uF, R1 = 100K, R2 = 470K,
Q = BC337.



2.    Ripple voltage is affected by Cout. Making this large will reduce
ripple and the circuit may not oscillated unless ripple is introduced by
other means eg C in a above diagram.
A more formal hysteresis method will, as dave noted, make the designers life
easier but it is certainly possible to make this work well. here.


3.      The "speedup" capacitor used by Dave in the two transistor circuit
makes a substantial difference here. (across drive resistor to high side
transistor base). I again used a 0.1 uF here.

4.     I used a medium sized toroid of "forgotten" inductance. I also tried
a small 2 mH potted inductor rated at 30 mA as suggested  by Roman. This
worked but produced substantially inferior efficiencies (as expected)

5    Swapping from a 1N4148 catch diode to a 40 volt 1A Schottky (used
because to hand) produced a very significant improvement in efficiency. A
BYV28 was worse than the Schottky but better than the 1N4148.

I didn't end up with a good enough efficiency to be worth writing up in
detail. Will play some more in due course although it may be a
while.Typically was getting 65% odd at 20 volts at 50 mA. Best efficiencies
were BELOW those achieved using the FET circuit yesterday.

Turning the circuit into a linear regulator was not hard :-) - eg use
largish Cout, no or very small speedup capacitor or feedback capacitor.
However, making a design which oscillated across the full range was not
hard. The circuit is certainly "pickier" when scaling than some more
standard designs but it will be possible to obtain a good result for any
desired operating conditions.

More anon but probably not soon.



regards



               Russell McMahon




.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\08\31@182558 by Dave Dilatush

picon face
Russell McMahon wrote...

>...It is now in limited production and so far 2000+ of these have
>been built, production tested and accepted by the Customer...

This makes everything MUCH more clear and I can now understand a
little better why every suggestion made to you so far has been met by
obdurate resistance and long-winded self-justification.

I've pointed out a number of things in your design that warrant close
attention.  Whether or not you attend to them is your business.

It is certainly no concern of mine and, so long as it appears that
whatever I say might just as productively be said to a brick wall,
neither is any continuation of this discussion.

Best of luck with your product, and have a nice day.

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics



'[EE]: Design Challenge - low power step down switc'
2001\09\02@023903 by Roman Black
flavicon
face
Hi everyone, just a quick note to say i've been testing
a stripped down version of Richards 2-transistor SMPS
buck 5v regulator, I removed a couple of parts and have
been testing it in its simplest possible configuration.

The good news is that the tiny 1.5mH RF choke i mentioned
before is working pretty well, efficiencies over the
current range from 10mA out to 70mA out(!!) seem pretty
constant, so I think the saturation issue is not as bad
as Dave D. and myself thought it would be! Excellent
for a resistor-sized inductor you can buy for a few cents.

I really haven't tried to get the efficiency right up
yet, i'm only using a 1N4148 for the flyback diode,
I will try a low Vf diode next. Also the main buck
transistor is only a BC327, so I will try a low Vce
sat transistor there too. The little cheap inductor
has a 9.6 ohm DC resistance, so I'm going to try
a lower H and lower R inductor, and play with it to
find the upper speed limits of the circuit. Since
these BC3?7 transistors are good for 100MHz I think
we could get good efficiencies right up in the 100's
of kHz range.

Even with the parts I have in there now, current gain
is about 2:1, at 12v in 10.3mA : 5v out 20mA. Not bad.

All in all, for a few cents worth of parts it's looking
like it could be refined to be very impressive!
-Roman

PS. Russell, I tried a variation using NO zener like
you were suggesting, using a R:R voltage divider on
the output rail and the Vbe of a transistor as the
regulator. It regulates ok, and saves one part, but
efficiency is not as good as the transistor must be ON
to turn Q1 off, where Richards original circuit is
very efficient as nothing turns on until the output
voltage drops, which is better for low current
situations.

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email spamBeGonelistservspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body


2001\09\02@083155 by Dave Dilatush

picon face
Roman,

You're getting some very encouraging results!

>...I'm going to try
>a lower H and lower R inductor, and play with it to
>find the upper speed limits of the circuit. Since
>these BC3?7 transistors are good for 100MHz I think
>we could get good efficiencies right up in the 100's
>of kHz range.

If you do try to raise the operating frequency, don't be too
disappointed if it doesn't work as well as expected; even using a
transistor with a high Ft you could end up losing significant power
due to transistor turn-off time.

The speedup capacitor (C3) in the "Modified Prosser" circuit I posted
improves the turn-ON time of Q2, but it has no role in helping turn Q2
off any faster; it can't, because there's no pull-up drive from Q1.

When I was playing around with this circuit I did some
back-of-the-envelope reckoning to see if a pullup resistor between Vin
and the collector of Q1 would improve efficiency by allowing C3 to
shorten Q2's turn-off time.  But at a low operating frequency, it
looked like any improvement from doing this would have been offset by
the power consumed in the pull-up resistor itself so I omitted it.

If you intend trying to operate this circuit up around a hundred
kilohertz, however, such a pullup resistor might help.  I'd suggest,
as a starting point, using the same resistance value as R5.

Cheers,

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservspam_OUTspammitvma.mit.edu with SET PICList DIGEST in the body


2001\09\02@135905 by Roman Black

flavicon
face
Dave Dilatush wrote:
>
> Roman,
>
> You're getting some very encouraging results!

Thanks for the input Dave. Being spice deficient
i'm doing it the old fashioned protobaord and cro
method, which has done me for 20 years so far. :o)

{Quote hidden}

Actually I should explain, I reduced Richard's circuit
to the basics, which was mainly simplifying all the
base drive circuits to a simple resistor. Then I decided
the weak point was that he relied on large enough ripple
at the output to make it oscillate, but not too much
ripple to make it worthless...

The first thing I did was re-design it to drive BACKWARDS
so I reduced the filter cap to a small 0.68uF cap and
then decoupled it from the final filter with a 10 ohm
resistor. This gives a deliberate high ripple of about
120mV on the cap, which slams the transistors hard
into a squarewave, beautiful to see on the cro, and
giving it good efficiency even at 55kHz which my first
circuit ran at.

Here is the circuit i'm testing now;
http://www.ezy.net.au/~fastvid/smps02.gif

Point A slams a very nice square wave, giving good
efficiency from about 20mA out to 80mA out. Point B
is a beautiful smooth sinewave, about 120mV P/P.
Freq is about 40kHz to 45kHz through the whole range,
and output voltage is totally ripple free.

Regulation suffers a bit from the 10 ohm resistor
in the output filter, but even as shown this circuit
is VERY usable for driving a 5v PIC from a 12v style
battery, giving 2:1 current gain with output currents
from 15mA to 100mA. Considering this is a cheap RF
"choke" inductor with 10 ohms dc resistance, and the
entire circuit is only 50 cents in parts or so, this
2:1 current gain might be very handy with battery
operated devices.

I'm charting stuff now, and will put it up on my web
page soon. Compared to a 5v regulator IC, even a pricy
low-dropout one, this circuit is at least 2x more
efficient, and maybe 3x more efficient than a 7805
regulator. And we're not even started yet!

Please comment, Dave and Russell, and I wish Richard
would add input as well. In my version of the circuit
i've sacrificed a bit of regulation to give a very
cheap, very squarewave, SMPS regulator with excellent
oscillation. The tiny RF choke is good to 100mA!!!
:o)
-Roman


> The speedup capacitor (C3) in the "Modified Prosser" circuit I posted
> improves the turn-ON time of Q2, but it has no role in helping turn Q2
> off any faster; it can't, because there's no pull-up drive from Q1.

With my mod, the transistors turn on/off very
well, even at 80kHz I tested. The trick was
ENCOURAGING a higher ripple at Point B, giving
much harder biasing of the transistors. The loss
is now shifted to regulation, which is a non-problem
if the device uses current in a set range. It still
compares ok to a resistor/zener regulator for
regulation, at about 3x more efficient. :o)


> If you intend trying to operate this circuit up around a hundred
> kilohertz, however, such a pullup resistor might help.  I'd suggest,
> as a starting point, using the same resistance value as R5.

I've eliminated the transistor biasing issues,
from what I can see on the cro, the next issue is
testing a toroid inductor that has 0 ohms dc
resistance, as the cheap RF choke i'm using
costs a lot in efficiency. Then the 10 ohm
output resistor can be reduced, and C1 "tuned"
to the match the inductor depending on what
current range is required.

The circuit shown gives:
* <1mA out = works as a linear regulator.
* 1mA to 12mA out = increasingly oscillates, gets more
 efficient. Totally reliable oscillation.
* >12mA out, slams a nice square wave, max efficiency
about 30mA out, good results to 100mA out, even at
120mA out with the cheap inductor gives 1.75 current
gain, MUCH better than any linear regulator.

Regulation is poor;
4.75v @ 10mA
4.60v @ 20mA
4.00v @ 80mA
These figures are easily adjustable based on the
zener voltage and the desired current range.
But it's usable! :o)
-Roman

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservspamspammitvma.mit.edu with SET PICList DIGEST in the body


2001\09\02@154407 by Olin Lathrop

face picon face
> Here is the circuit i'm testing now;
> http://www.ezy.net.au/~fastvid/smps02.gif

Nice job, Roman.  This is the first circuit I've seen in this discussion
which I might actually be willing to use in a low current, low cost
situation.

As you mentioned, the 10ohm output resistor limits application somewhat.
Here's an idea that increases cost slightly but broadens the useful range of
applications:

You've basically got a somewhat "dirty" switching regulator that you want to
clean up a bit before driving the load.  The RC filter is one approach.
Another is a cheap linear post regulator.  In keeping with the low cost
philosophy of this circuit, a simple emitter follower from a voltage source
should be suitable for circuits with 5V logic chips.  Replace R1 with a NPN
transistor, collector to C1, emitter to C3, and base to the 5.6V reference
created by D2.  This forms a "good enough" linear regulator for many
purposes, but the voltage at B needs to be a little higher to give it
something to work with.  This is easily accomplished with a forward biased
diode between D2 and R2, with the base of Q2 tapping off above the diode.
That way the switcher tries to keep B one junction drop above the desired
output voltage, which is then dropped by the emitter follower linear
regulator.  I might also want to put a small cap accross D2.  This scheme
will cost a few pennies more, descrease efficiency slightly (the extra 700mV
linear drop), but should provide significantly better load regulation and
lower output impedance.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, spam_OUTolinspam_OUTspamspam_OUTembedinc.com, http://www.embedinc.com

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservspam_OUTspammitvma.mit.edu with SET PICList DIGEST in the body


2001\09\02@170600 by Dave Dilatush

picon face
Roman,

>The first thing I did was re-design it to drive BACKWARDS
>so I reduced the filter cap to a small 0.68uF cap and
>then decoupled it from the final filter with a 10 ohm
>resistor. This gives a deliberate high ripple of about
>120mV on the cap, which slams the transistors hard
>into a squarewave, beautiful to see on the cro, and
>giving it good efficiency even at 55kHz which my first
>circuit ran at.

What you've done is help the circuit oscillate by increasing the loop
gain at high frequencies.  If you look at this circuit in a
small-signal sense, what you've got is a common-base amplifier stage
(Q2) whose input is at point B; this common-base stage then feeds a
common-emitter stage (Q1) which feeds back to the input (emitter of
Q2) through L1.  C1 shunts some of this signal to ground, and reducing
C1 lessens that effect giving you stronger oscillations.

>
>Here is the circuit i'm testing now;
>http://www.ezy.net.au/~fastvid/smps02.gif
>

[snip]

>...and the
>entire circuit is only 50 cents in parts or so, this
>2:1 current gain might be very handy with battery
>operated devices.

Agreed wholeheartedly, PROVIDED the user keeps in mind the limitations
inherent in this circuit: lots of high-frequency ripple (like all
switching regulators) and rather so-so regulation- a consequence of
this particular, "CheapMode(tm)" design.

>The circuit shown gives:
>* <1mA out = works as a linear regulator.
>* 1mA to 12mA out = increasingly oscillates, gets more
>  efficient. Totally reliable oscillation.
>* >12mA out, slams a nice square wave, max efficiency
>about 30mA out, good results to 100mA out, even at
>120mA out with the cheap inductor gives 1.75 current
>gain, MUCH better than any linear regulator.

The fact that oscillation ceases below about a milliamp or so is a
consequence of the basic design: there has to be enough current
flowing through Q1 and Q2 that they have enough gain to sustain
oscillation.

The varying "enthusiasm" of its oscillations between 1 mA and 12 mA is
a result of not using any hysteresis (i.e., positive feedback) in the
circuit.  If you put enough hysteresis in, like in Richard's original
circuit, it'll oscillate like a politician at all current levels
except the very lowest.

>Regulation is poor;
>4.75v @ 10mA
>4.60v @ 20mA
>4.00v @ 80mA

Yeah, well waddya want, it's CHEAP!

Good work, Roman.

Dave

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email RemoveMElistservKILLspamspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body


2001\09\02@185041 by Russell McMahon

picon face
> Actually I should explain, I reduced Richard's circuit
> to the basics, which was mainly simplifying all the
> base drive circuits to a simple resistor. Then I decided
> the weak point was that he relied on large enough ripple
> at the output to make it oscillate, but not too much
> ripple to make it worthless...
>
> The first thing I did was re-design it to drive BACKWARDS
> so I reduced the filter cap to a small 0.68uF cap and
> then decoupled it from the final filter with a 10 ohm
> resistor. This gives a deliberate high ripple of about
> 120mV on the cap, which slams the transistors hard
> into a squarewave, beautiful to see on the cro, and
> giving it good efficiency even at 55kHz which my first
> circuit ran at.
>
> Here is the circuit i'm testing now;
> http://www.ezy.net.au/~fastvid/smps02.gif


I like it!
A beautifully simple circuit.

The series resistor does hurt efficiency but only at the ratio of "extra"
ripple to output voltage regardless of power. ie if it is designed to allow
0.2v of ripple to ensure proper operation, and the output is 5 volts then
the efficiency loss in this resistor is about 0.2 / 5 = 4%. If the output
current range is not too wide this resistor can be designed to allow the
desired amount of ripple on C1. Olin's suggestion of an emitter follower
regulator works at constant voltage drop over a wider range of load currents
but is potentially less efficient than a simple resistor as it introduces an
additional 0.6 volt increase in output voltage by adding a diiode in series
with the zener to give the regulator "headroom". Using the transistor it
would be easier to produce low ripple output and the regulation would be
significantly improved (at the expense of efficiency.) The transistor is
probably "free" as the output capacitor required to achieve acceptable
output ripple is very much smaller with the active regulator.

As may be expected, I find the method of operation acceptable but it is
interesting to note that this circuit actually eliminates the static
feedback component used in Richard's original circuit (and Dave's modified
version) so that it is left only with the dynamic method used in my original
design. Unclocked or untimed buck regulators generally rely on output
voltage variations to initiate oscillation. While the dynamic feedback from
output ripple indeed helps oscillation in his circuit and can even be the
main feedback mechanism, Richard also added "static" feedback by using a
resistive divider from the input side of the inductor and the zener to
provide a reference which varied depending on whether the the circuit was
operating in the forward or flyback part of its cycle. You have eliminated
static feedback by making the reference voltage "stiff" by eliminating both
the resistive feed to Q2 base and the "state" feedback  from the input end
of the inductor. Now the only feedback is the ripple voltage on C1 relative
to the voltage on zener D2. This is the "dynamic hysteresis" mode which I
described in my original circuit. This would ***theoretically*** allow the
circuit to assume a linear operation mode as was noted by others for my
original circuit. Should this theoretical loss be of concern to anyone, any
desired amount of static feedback can be added by reinserting a resistor
between the R2/D2 junction and Q2 base and another between Q2 base and Q1
collector. Whether this is felt necessary depends on how happy one is that
the dynamic feedback method will start unconditionally in practice. (The
"wrong sense feed-forward" that this reintroduces is not significant for
designs operating over a limited range of input voltages.)

It would be interesting to see how much delivered energy gain such a circuit
would achieve when providing a 5 volt or 3 volt output from a 9 volt battery
over the battery's lifetime compared to the use of a linear regulator. For a
battery endpoint of 6 volts and if a linear voltage decay is assumed the
battery life at 5 volts output would be 150% at 100% efficiency (ie 50%
longer)  [  (9-6)/2 - 5 ) / 5 x 100% = 150%]  and 250% at 3 volts and 100 %
efficiency. Real results would of course be rather less.



regards

               Russell McMahon

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservspamBeGonespam.....mitvma.mit.edu with SET PICList DIGEST in the body


2001\09\03@055036 by Roman Black

flavicon
face
Olin Lathrop wrote:
>
> > Here is the circuit i'm testing now;
> > http://www.ezy.net.au/~fastvid/smps02.gif
>
> Nice job, Roman.  This is the first circuit I've seen in this discussion
> which I might actually be willing to use in a low current, low cost
> situation.

Thank you Olin, and thanks to Richard whose circuit I
butchered... :o)


{Quote hidden}

I know where you're coming from, I had a quick fiddle
with post-regulators on the circuit including something
a lot like what you're suggesting, and also with a
simple zener shunt reg.

However, the best effciencies in the circuit come from
keeping R1 high, as this causes the best saturation
(squarewave) of the transistors which gives better
efficiency. Lowering R1 to 5 ohms saves losses at
R1 but overall efficiency was down.

Output regulation is not really that bad;
4.50v @ 25mA
4.25v @ 50mA
And this is over the peak efficiency range, which
makes it quite usable, regulating better than a lot
of RZ zener regulators i've used. For many PIC
projects is regulation really that critical?

Because the RF choke I used has a 10 ohm dc
resistance, I think regulation will be tightened a
lot by using a better inductor (toroid?) as
Russell suggested. And efficiency will improve
also of course.

One of my favorite tools is a 2.7v 25mA high
brightness LED, attached to 7v NiCd. At the moment
it has a series resistor, but this new circuit
will be IDEAL for something like that. :o)
-Roman

--
http://www.piclist.com hint: To leave the PICList
KILLspampiclist-unsubscribe-requestspam.....mitvma.mit.edu


2001\09\03@083255 by Russell McMahon

picon face
> > You've basically got a somewhat "dirty" switching regulator that you
want to
> > clean up a bit before driving the load.  The RC filter is one approach.
> > Another is a cheap linear post regulator.

> I know where you're coming from, I had a quick fiddle
> with post-regulators on the circuit including something
> a lot like what you're suggesting, and also with a
> simple zener shunt reg.
>
> However, the best efficiencies in the circuit come from
> keeping R1 high, as this causes the best saturation
> (squarewave) of the transistors which gives better
> efficiency. Lowering R1 to 5 ohms saves losses at
> R1 but overall efficiency was down.

If you don't mind the extra cost you might consider trying a second of your
mini-inductors in the output in place of R1. You would want to choose one
with lower or no more resistance than the present R1. This then gives both a
smoother output due to the inductance, and the ripple on C1 is also  further
buffered by the inductance from being reduced by output load.

It would be nice if C1 could be reduced to a 100 NF monolithic ceramic as
these are common and rather cheaper than the present 680 NF.


> Output regulation is not really that bad;
> 4.50v @ 25 ma
> 4.25v @ 50 ma
> And this is over the peak efficiency range, which
> makes it quite useable, regulating better than a lot
> of RZ zener regulators i've used. For many PIC
> projects is regulation really that critical?

Even a standard spec PIC will certainly work across a wide range of voltages
and as long as support circuitry also tolerated this variation it would be
OK in many cases. (eg 16F84-04 commercial & industrial XT mode 4 <= Vdd <=
6v. LF = 2 to 6V !)
[ I produce a small number of devices for a special application which run on
a 3 cell NiMH button pack with a nominal voltage of 3.6V and actual voltage
of as low as 2.5V. While this is out of spec for a standard part I have so
far had a limited quantity run perfectly for my purposes at all voltages
including eerom writing and reading. Your mileage may vary but for hobbyist
applications this is worth noting. ]

> Because the RF choke I used has a 10 ohm dc
> resistance, I think regulation will be tightened a
> lot by using a better inductor (toroid?) as
> Russell suggested. And efficiency will improve
> also of course.

I am using "Micrometal brand" powdered iron toroids in a wide range of
sizes. These generally are substantially superior to ferrite and their cost
is impressively low (10's of cents up for typical sizes in the 1 to 10 watt
range). In Australia Jaycar sell a limited range of sizes (but not as cheap
as they should be) and you can order them directly from Micrometals in USA.
They supply a very useful design and data book. (The yellow core with white
side band (colour indicates grade) is reasonably well suited for low to mid
frequency use - for over say 100 kHz use you may wish to select another
grade. (catalog not to hand as I write or I'd put in details).

The greatest problem with toroids for amateurs is winding them. For
prototypes with many turns I make a bobbin which all the wire is wound on
and then repeatedly threaded through the core. (failing something better two
flat headed nails that fit through the core with wire in place can be cut to
desired length and soldered to make a dumbbell. When wound professionally
they often use a machine which does the same thing  but a zillion times
faster. The cost of commercially wound smaller toroidal coils is typically
dominated by winding costs. If radiation is not a major concern then
reasonable results can be obtained from small open gap dumbbell shaped
ferrite
cores which have the great benefit of being trivially easy to wind.

> One of my favourite tools is a 2.7v 25 ma high
> brightness LED, attached to 7v NiCd. At the moment
> it has a series resistor, but this new circuit
> will be IDEAL for something like that. :o)

Lots of people would find that application useful. In this application, if
you stuck to 10 r for R1 you would drop a mean voltage of 250 mV or about
10% efficiency loss in the resistor. Bearable, but with such a well defined
load characteristic some fine tuning should be easy enough and beneficial. I
intend to do a single cell step up design at some stage. Made relatively
easy by the nature of the load - feed it controlled packets of energy in an
inductor and let it sort out what voltage it wants to conduct at.



regards

           Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
spam_OUTpiclist-unsubscribe-requestspamKILLspammitvma.mit.edu


2001\09\03@085542 by Olin Lathrop

face picon face
> Output regulation is not really that bad;
> 4.50v @ 25mA
> 4.25v @ 50mA
> And this is over the peak efficiency range, which
> makes it quite usable, regulating better than a lot
> of RZ zener regulators i've used. For many PIC
> projects is regulation really that critical?

Note that the acceptable supply range for many PICs at 20MHz is 4.5 to 5.5
volts.  I'm not trying to fault your circuit since it works within the
limitations you have stated.  For low current loads, an RC filter like your
R1 and C3 can be fine and is certainly cheap and simple.  Your main
objection to an emitter follower post regulator is that it reduces
efficiency because of the extra 700mV drop at full current.  Agreed, but
note that you've got 500mV accross R1 at 50mA, and the supply has drooped a
total of 750mV.  In other words, at higher currents you are going to have
some losses anyway.  You might as well "spend" those losses to get better
regulation and lower output impedance.  It looks like your circuit is
probably better for currents up to 25mA.  I bet that somewhere around 30mA
to 50mA the emitter follower post regulater will start to look better,
although I haven't worked this out carefully.

> Because the RF choke I used has a 10 ohm dc
> resistance, I think regulation will be tightened a
> lot by using a better inductor (toroid?) as
> Russell suggested. And efficiency will improve
> also of course.

Your oscillation depends on having high enough gain and the interaction of
L1 and C1 to define the frequency.  This is the one part of this circuit
that does make me a bit nervous, but not like Russel's because at least the
oscillations are reasonably predictable as long as the Q1-Q2 gain is high
enough.  If you start messing with the inductor, I would consider a little
hysteresis.  One easy way to do this is a little feedback from the collector
of Q1 to the base of Q2.  Only two extra resistors (yeah I know, the parts
count is starting to creep up).


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, RemoveMEolinRemoveMEspamEraseMEembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
KILLspampiclist-unsubscribe-requestspamspamBeGonemitvma.mit.edu


2001\09\03@115824 by Roman Black

flavicon
face
Dave Dilatush wrote:
>
> Roman,
>
> >The first thing I did was re-design it to drive BACKWARDS
> >so I reduced the filter cap to a small 0.68uF cap and
> >then decoupled it from the final filter

> What you've done is help the circuit oscillate by increasing the loop
> gain at high frequencies.  If you look at this circuit in a
> small-signal sense, what you've got is a common-base amplifier stage
> (Q2) whose input is at point B; this common-base stage then feeds a
> common-emitter stage (Q1) which feeds back to the input (emitter of
> Q2) through L1.  C1 shunts some of this signal to ground, and reducing
> C1 lessens that effect giving you stronger oscillations.

Absolutely agreed, the brutality of my transistor
biasing was done deliberately, and the main "tuning"
was shifted to L1,C1,R1 to make it easy to play
with on the protoboard. Those 3 parts can be easily
adjusted to give a fairly efficient regulator at
a set current.
:o)

{Quote hidden}

Yeah, the regulation is pretty poor, but over
the last year I have developed a number of PIC
projects driven from batteries and regulated
by a simple RZ zener shunt. That has poor
regulation too, AND 3x worse efficiency. I have
a few tools and things that will get this cheap
SMPS regulator stuck in them soon. :o)

I should mention that the output ripple of the
circuit above is actually VERY good!! Less than
10mV ripple, and the ripple is sine shaped. This
is excellent by switchmode standards. I even
downgraded the final filter cap from 100uF to
10uF as the ripple was so good. The 40kHz freq
and 10 ohm/10uF network see to that. :o)

{Quote hidden}

Oscillation is reliable enough to satisfy me,
under all loads and start conditions I could test.
There was no possible way I could get it to NOT
oscillate. This circuit really is best suited to
apps with a narrow current requirement, like
running a LED from a 9v battery, or any device
that draws a fairly regular current. The goal was
to strip it to the basics, and maybe someone
else will be happy to mod it, or compete with
their own circuit, I think this is a REALLY
exciting PIC thread, as so many newbies are always
asking HOW to save battery power with their
projects...
:o)
-Roman

--
http://www.piclist.com hint: To leave the PICList
piclist-unsubscribe-requestspamspammitvma.mit.edu


2001\09\03@121705 by Roman Black

flavicon
face
Russell McMahon wrote:
{Quote hidden}

Really excellent suggestion, thanks Russell!!
The main problem I found is that you need about 150mV
ripple at C1 to give good squarewave switching.
Reducing R1 made that worse as the Q1 didn't saturate
as well, even though R1 losses were reduced and
regulation improved, the overall efficiency was worse.

Now using an inductor here, or series RL, would
give some good gains. Am i the only one that thinks
with a home-wound toroid to optimise L1 poperties,
and ANOTHER winding to replace R1, you sould achieve
the goals of higher ripple at C1, giving good switching,
and less ripple at the output with no R losses??
Opinions?? :o)

> It would be nice if C1 could be reduced to a 100 NF monolithic ceramic as
> these are common and rather cheaper than the present 680 NF.

Very good point, a super cheap circuit should
use the cheapest and most common parts. This
change will mean increasing the freq a bit...

{Quote hidden}

Sure. Obviously the voltages mentioned were with
the 5.6v zener, if you used a 6.2v zener that would
become:
5.10v @ 25 ma
4.85v @ 50 ma


{Quote hidden}

Excellent toroid info! I will try the circuit with some
better inductors, I have some of the powdered iron
ones you mentioned. It's a pain winding 100 turns on
a toroid, and unless the efficiency is critical, that
cheap RF choke is looking better by the minute.

{Quote hidden}

I look forward to seeing what you come up with.
People have been turning towards these LEDs for
bicycle headlights, point lights like mine,
even emergency lighting from a D cell NiCd would
be very useful project. In a totally dark room
a white LED is pretty impressive.
-Roman

--
http://www.piclist.com hint: To leave the PICList
RemoveMEpiclist-unsubscribe-requestspamBeGonespamRemoveMEmitvma.mit.edu


2001\09\04@034611 by Vasile Surducan

flavicon
face
On Tue, 4 Sep 2001, Russell McMahon wrote:

> If you don't mind the extra cost you might consider trying a second of your
> mini-inductors in the output in place of R1. You would want to choose one
> with lower or no more resistance than the present R1. This then gives both a
> smoother output due to the inductance, and the ripple on C1 is also  further
> buffered by the inductance from being reduced by output load.
>
> It would be nice if C1 could be reduced to a 100 NF monolithic ceramic as
> these are common and rather cheaper than the present 680 NF.

 I often use a combination at the output of the switching regulator with
100nF, a toroidal inductance with two coils on both active and gnd line
and another 100nF to output. This usual decrease the output ripple below
10 mV on supplies which have more than 100mV ripple ( and more than 3A on
loads )
Vasile

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\09\04@092830 by MATTHEWS, DEAN (D.)

picon face
TO ALL,

i WOULD LIKE TO USE THE ONBOARD A/D CONVERTERS OF THE PIC, CAN ANYONE RECOMMEND THE MOST USER FRIENDLY PIC WHICH WOULD BE THE EASIEST TO USE. i WOULD LIKE TO INPUT A VARYING 0-5V SIGNAL INTO THE PIC AND THEN DISPLAY SOME VALUES ON AN LCD SCREEN.

Dean Matthews
Reliability Engineering
Ford Engine Plant Bridgend
South Wales, U.K
Tel No:(01656)672597
Fax No:(01656)672558
Email: KILLspamdmatth14spamBeGonespamford.com
` Please consider your environmental responsibility before printing this e-mail

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\09\04@102828 by uter van ooijen & floortje hanneman

picon face
> i WOULD LIKE TO USE THE ONBOARD A/D CONVERTERS OF THE PIC, CAN ANYONE
RECOMMEND THE MOST USER FRIENDLY PIC WHICH WOULD BE THE EASIEST TO USE. i
WOULD LIKE TO INPUT A VARYING 0-5V SIGNAL INTO THE PIC AND THEN DISPLAY SOME
VALUES ON AN LCD SCREEN.

You sound like a
- beginner
- hobbyist
If this is both true (and don't be offended!) the 16F877 is probably the
best 'general pupose' start. Take your choice of
- almost 'zero components' LVP programming
- HVP programming using a somewhat larger programming circuit
- a bootloader (but you must program the bootloader in the PIC first!)
- the (uChip or self-build) debugger
And take some time to select a language!
- MPASM (free, assembler)
- my Jal (free, but not a commercial product)
- is a free C compiler limited to 1K code (kbd?)
- ....

Wouter van Ooijen

Van Ooijen Technische Informatica: http://www.voti.nl
Jal compiler for PIC uC's:  http://www.xs4all.nl/~wf/wouter/pic/jal

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\09\04@151034 by Olin Lathrop

face picon face
> i WOULD LIKE TO USE THE ONBOARD A/D CONVERTERS OF THE PIC, CAN ANYONE
> RECOMMEND THE MOST USER FRIENDLY PIC WHICH WOULD BE THE EASIEST TO USE.

Someone probably can, but it's more likely after you stop shouting.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, @spam@olinSTOPspamspam@spam@embedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\09\05@061508 by dr. Imre Bartfai

flavicon
face
See comment below.

+-----------------------------------------------------------------------+
| The information transmitted is intended only for the person or entity |
| to which it is addressed and may contain confidential and/or          |
| privileged material.  Any review, retransmission, dissemination or    |
| other use of, or taking of any action in reliance upon, this          |
| information by persons or entities other than the intended recipient  |
| is prohibited. If you received this in error, please contact the      |
| sender and delete the material from any computer.                     |
+-----------------------------------------------------------------------+

On Tue, 4 Sep 2001, wouter van ooijen & floortje hanneman wrote:

{Quote hidden}

                                            ^^^

http://www.bknd.com


Regards,
Imre

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2001\09\08@110222 by Russell McMahon

picon face
part 1 7318 bytes content-type:text/plain; (decoded 7bit)

Encouraged by Roman's input and results with his version of Richards two
transistor buck converter (not to mention Dave and Olin's input  (-
encouragement takes many forms :-)) ) I decided to try a purpose built 12 to
5v design and look at addressing some of the concerns raised.

Results were "pleasing". This is by no means a fully optimised design but
indicates what is able to be easily achieved.
The circuit is slightly more complex than Roman's. Whether the extra
complexity or the circuit itself are justifiable is up to potential users to
decide.

In the following discussion and data figures are quoted to RESOLUTIONS in
excess of their probable PRECISIONS . eg Iout = 10.78 mA. The use of
extended precision is to allow similar results to be compared and trends to
be seen as other  variables are changed. The number of significant figures
used are consistent with the observed resolutions in each case.

Efficiencies are measured including losses in R10 and also in a 10r
decoupling resistor in the Vin lead.
At 6 mA in and 10 mA out these add a few percent to the overall loss of
efficiency.

TARGET SPECIFICATION:
Target Vin / Vout was 12 v / 5v as per recent discussion.
Target nominal Iout was 10 mA.

I tried Vin from 7.9 to 30 volts and Iout at nominal 10 mA and 20 mA.

Best efficiency at 10 mA was 76% at 9v in.
At 12v in efficiency was about 72% .

"Current multiplier ratios" over that achievable with a linear supply
increased with voltage in giving 1.7 times at 12 volts, 2.4 times at 20
volts, 3.1 times at 30 volts. This figure is identical to the ratio of
efficiency of this circuit to efficiency of a linear power supply under the
same conditions.

This circuit was designed with the aim of operating at Vin = 12v and only
then was tested at other Vins.
Operation in the range 5 < Vin < 9 v could be improved to provide a
converter intended to optimise power output from a "PP3" 9v transistor radio
battery


Core component values were as shown on the diagram.

CATCH DIODE                Catch diode was BYV26 for most measurements with
results for a 1N4148 being compared for a few data points. The BYV26 was
superior but not by as much as I had expected. Say 1N4148 is ~~ 97% as
efficient.

INDUCTOR:                Inductor was either 120 turns or 30 turns on a
Micrometal 20 x 13 x 8 (OD ID height) core.
The 120 turns produced measurably better results (eg Iin = 6.27 mA versus
Iin = 7.83 mA at Vin = 12 V)
This was probably due to the very short switching period with the smaller
inductor.
This core is MUCH larger than that used by Roman. Results with miniature off
the shelf prewound will follow "in due course".



C1 was usually a 1 uF mylar but a 0.1 uF monolithic ceramic was tried for a
few data points.
The 0.1uF was about 95% as efficient - probably mainly due to the increase
in switching frequency which resulted.

CAPACITOR ESR        Dave raised the issue of output capacitor ESR which he
said was crucial to frequency of operation. I produced a pessimistic numeric
expression for affect of ESR on frequency of operation.
A 1 ohm resistor was placed in series with C1 to simulate a capacitor with
significant ESR. At 30v frequency of operation shifted from 20 kHz to about
38 kHz when  the series resistor was added. This shift is expected and is a
natural consequence of the method of feedback and is entirely acceptable in
envisaged applications. An ESR of this magnitude would be "unusual".

ZENER CURRENT        Dave and Olin have noted the low zener current in my
original design which would lead to somewhat reduced voltage regulation and
potentially extra noise due to operation on the flatter part of the zeners'
V/I knee. I increased zener current by adding R1 to ground from Zener anode.
The R1/R2 junction must now be driven to about 0.6 volts plus by the zener
before Q1 is driven on. At this stage zener current will be about 600/R1 mA.
This current can be made arbitrarily large by selecting appropriate R1. In
this case with R1 = 600r Izener is about 0.6 mA. In practice the regulation
achieved suggests the zener is being operated well enough up its knee for
typical applications. This "wasted" zener current represents an efficiency
loss of about 5% at 10 mA out!

A 5v1 zener was used. Output voltage was 5V +/- 0.05 volt over most Vin.
The fact that Vout < Vzener + 0.6 v indicates the zener is still operating
further down the knee than per spec sheet. Given Iz is about 0.6 mA this
is expected.


STATIC FEEDBACK / HYSTERESIS        I considered adding explicit static
hysteresis as suggested by Dave & Olin. While this would meet the
theoretical concerns raised as to the possibility of the circuit failing to
oscillate in some cases, in practice the circuit is so utterly convinced
that it wants to oscillate that I did not pursue this at this stage. I will
examine this in due course. Roman's latest design removed the static
feedback which had been present in Richard's initial design and relies
entirely on dynamic hysteresis as in this design. He also reports
enthusiastic oscillation in practice in all cases to date. Adding static
hysteresis should help switching waveforms and I will examine this aspect in
due course.

TURN ON/OFF TIMES                                 Comments were made about
the potentially long switching times of the main switching transistor. Drive
values were semi-optimised by tuning to minimise input current at a fixed
Vin
and Pout. The very small value of R5 (1k) was chosen for best efficiency.
With R4 = 10 k as shown, this has the effect of limiting circuit operation
to Vin > about 7 volts as R4/R5 form a divide and Q3 cannot be turned on
when Vin * R5/(R4+R5) is less than about 0.6 volts. For operation at lower
Vins R4 & R5 would need to be reoptimised. Use of a high side driver
transistor (current amplifier) for Q3 as per my previously posted FET based
design would allow good tune off drive without  increasing dissipation in R4
excessively. Turn on is generally less of a problem than turn off.

CIRCUIT OPERATION:

All transistors off.
R6 turns Q2 on.
Q2 on turns Q3 on via R4.
Q3 on supplies current via Q3, L, R10 to Rload and Cout.
V at  L1/R10 junction rises until Z1conducts and pulls R1/R2 junction to 0.6
volts.
Now Q1 on so Q2 off so Q3 off.
Q3 off causes L to 'ring" so left hand end of L goes to ground (actually
diode drop BELOW ground) due to D1.
V on C1 CONTINUES TO RISE die to stored energy in L.This is a primary part
of the feedback mechanism.
When energy drops Vc1 drops again and cycle repeats.

RESULTS

See attached Excel V2 spreadsheet.

Output voltage regulation at Io = 10mA nominal was 4.96v <= Vout <= 5.13v
for 8 <= Vin <= 30 v or better than +/- 2%.
Regulation was about +/-2.2% at 20 mA load.

Efficiency was slightly better at 20 mA out at all values of Vin.

FREQUENCY / PERIOD OF OSCILLATION

At 10 mA the period of 1 cycle for Vin = 10, 20, 25, 30 was 25 uS, 35uS, 42
uS, and 50 uS respectively.
(40 kHz to 20 kHz).

At Io = 20 mA frequency was about 33 kHZ at 12v in and 24 kHz at Vin = 30
volt.


**** ANYONE INTERESTED ADVISE ME OFFLIST AND I WILL SEND EXCEL RESULTS FILE
(10 KB)









_____________________________________





part 2 4789 bytes content-type:image/gif; (decode)


part 3 105 bytes
--
http://www.piclist.com hint: To leave the PICList
piclist-unsubscribe-requestspamBeGonespamspamBeGonemitvma.mit.edu


2001\09\09@051901 by Roman Black

flavicon
face
Russell McMahon wrote:
{Quote hidden}

Are these the wrong way around?? :o)

> "Current multiplier ratios" over that achievable with a linear supply
> increased with voltage in giving 1.7 times at 12 volts,
>     ---------------------------------------------------------------
>
>                               Name: gsr12v.gif
>                Part 1.2       Type: GIF Image (image/gif)
>                           Encoding: base64


Great info Russell! :o)
I think maybe a toroid is not the best way
to go for such low currents. My circuit was getting
1.8x current multiplying over 20mA to 70mA output
range, from 12v input, and with 13.0v input it was
getting over 2.05x current gain from 20mA out to
80mA out, and still over 1.75x current gain
at 120mA out!!

This was all with the little RF inductor, 1.5mH
from WES components, they use them a lot in Philips
TVs so I had a few here. A 1.0mH may be even better.

I tried MANY pre-wound inductors, i'll put some
pictures up on a page soon, I have a couple thousand
small inductors collected over the years, I tried most
shapes and types, and not many performed better than
the RF inductor! The larger core mass ones were more
efficient at very low currents, but often saturated
at 60mA or less. The RF inductor with basically no
core, was less efficient at currents under 15mA,
but performed admirably out to 200mA which really
surprised me for a tiny cheap package.

I tried some toroids that I had wound before, these
weren't as good! I think they require much larger
currents to get efficient. :o)

I'll put the pictures and charts up soon.

How did you go with the regulation? That was the
main problem with the RF inductor, as it had 10 ohms
dc resistance and R1 also is 10 ohms, regulation
suffered. I found with inductors under 3 ohms I could
drop R1 a lot, even to 4.7 ohms, and regulation
was 3x better but efficiency wasn't always as good
due to slower Q1 switching.

I'll show my charts if you show yours?? ;o)
-Roman

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2001\09\09@071720 by Russell McMahon

picon face
A short addenda to yesterday's post re my recent measurements.

C1 could be removed completely or changed to eg a 10uF electrolytic without
quenching oscillation.

The 10r resistor R10 was added as used by Roman to isolate the feedback
mechanism from load variations.
R10 can be replaced by a short circuit without quenching oscillation.

R3 makes no difference to circuit operation and can be removed. (R2 & R1
swamp R3)



RM

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


More... (looser matching)
- Last day of these posts
- In 2001 , 2002 only
- Today
- New search...