Searching \ for '[EE]: Design challenge - switching regulator' 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/priswitch.htm?key=switching
Search entire site for: 'Design challenge - switching regulator'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: Design challenge - switching regulator'
2001\01\18@075308 by Russell McMahon

picon face
A "design challenge" seems to be the best way to get input - hopefully :-)

I need to implement a switching regulator of modest power output but
somewhat outside the scope of the normal "off the shelf" simple-switchers or
their ilk.

At present I'm prototyping using the time honoured and capable UC3845 or
equivalent (3842/3/4/5 family) but it's minimum startup voltage of around 9
volts is annoying.
I'd be interested in anyone's suggestions on a design which at least meets
the following specification.
When I ask this sort of question here the answers are usually interesting
and often enough add to my general body of knowledge BUT they often also
fail to address the actual problem as proposed. To be useful to me the
design MUST meet all the vital specs and as many of the "useful" specs as
possible.

The 100 volt  (occasionally 130 volt) input requirement eliminates most
integrated solutions.

VITAL REQUIREMENTS

1.    Operates from an input of typically 10v to 100v DC but input MAY rise
to 130 volt DC for up to 2 seconds (and ideally will operate on a much lower
voltage than 10 volts).

2.    Mean maximum output 600 mA DC at 9 volts (5.5 watts) BUT must be able
to supply up to 1.5A at 7.5 volts for 1 second.

       OR

       6.5v at 600 mA plus 1.5A as above AND a separate 9 volt 75 mA output


3.    Must start with worst case load applied with an input voltage of 10
volts or more.

4.    As cheap as reasonably possible.

5.    Acceptable EMC performance in a stand alone battery operated
equipment.


DESIRABLE REQUIREMENTS


5    Very low cost of manufacture including components, magnetics and
assembly.

6.    Isolated or semi-isolated design so that failure of the switching
supply will (probably) not destroy the powered circuitry
    ie a buck regulator would probably not be a nice solution as failure of
the pass transistor exposes the output circuitry to 100v plus.

7    Start and run from 4 volt input



=================================================================

The UC3845 which I am presently using provides current limiting and voltage
regulation.
High voltage operation is possible due to use of an external pass FET.
A flyback configuration provides semi-isolation (ground common but not
direct path from 100 volts input to the output).
The IC is cheap (well under $US1).
The power FET is cheap (under $US1).
Magnetics are not certain yet - powdered iron toroid or small ferrite U core
both have advantages.
Startup voltage is 10 volts plus but once started it will operate down to a
few volts input.


Your ideas?


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\01\18@091832 by M. Adam Davis

flavicon
face
That's a tough call.  I'd say stick with your current thought, and if you
absolutely need it to start at lower voltages add a boost regulator to
provide the starting voltage, set up so it is shut down when the main
regulator is going.

That input voltage range really is huge though, and the requirement of
operating with inputs above and below your output limits you as well.
Unless you change your requirements I doubt you'll get any off the
shelf/easily implemented/cheap solutions.

A new triangle, pehaps?
Wide input range, high current output (greater than .1A), and low cost.
Pick any two...

-Adam

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\01\18@112236 by Roman Black

flavicon
face
Ok Russell, got the lateral bit of my brain working
now, your input range of 10v - 130v is pretty big for
a 9v supply. Will need huge range of PWM width, which
shoots you in the foot for designing inductors
and getting good filtering. Seems more suited to
a fixed pulse width variable freq system, with some
type of wacky filtering...

How about run your load from a NiCd pack, at about
9v. Then you can use a cruder switchmode system
to dump enough power into it to keep it's voltage
reas constant. I noted your 9v - 7.5v output range
and i'm assuming regulation is not critical.
Now your inductor doesn't have to "filter", really
it's just there for switchmode efficiency, so
it is cheaper and less critical design.

Now you go and ask for it to run from a 4v input?
So you want step-up too? I'd go for a fixed pulse
width system, (i've seen this in OLD switchers,
it allows optimum design for filtering as the
pulse width is stable over a huge input voltage
range) into a small toroid transformer, (gives
isolation) and simple half wave rect on output.
Decent filtering on output, or maybe NiCd as
mentioned.

Regulation is on/off, it dumps pulses until output
voltage is reached, then stops pulsing. This will
give pretty good regulation while being safe and
cheap for isolation.

Think of it more as a way of dumping pulses of
energy from the primary, until the output is full.
Could use simple self excited primary, (with some
mechanism of turning it off when output is "full")
would probably work from 2v up with the right primary
semi. Whew. My lateral brain cell is tired. ;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\01\19@010721 by Russell McMahon

picon face
>A new triangle, pehaps?
>Wide input range, high current output (greater than .1A), and low cost.
>Pick any two...


The irony is that it is only an inbuilt feature of the regulator IC that
limits the lower starting voltage in my current design. It starts at 9 volts
or so but there doesn't seem any good reason why it could not go quite a bit
lower. Once it starts I self power the IC from its own output (standard
practice) so the input voltage can drop to 2 or 3 volts as long as the input
supply can source the current peaks.

There is a battery involved but I can't rely on it being there and it's
voltage is almost always less than 9 volts.


RM

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


2001\01\19@040230 by Vasile Surducan

flavicon
face
Uc3842/45 is a standard driver known in most PC power supplies and
monitors. It has all requirements you need except is not working with
power supply less then 8...9V. The other requirements can be easily
achieved. So, I propose to keep your actual design and to built a cheap
( two transistors and a few other components ) oscillator who may run from
under 9V and to supply the UC3842 with minimum 11V all the time.
Vasile

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


2001\01\19@051232 by Vasile Surducan

flavicon
face
 So, again my ideea is good then ! You need a circuit just for power-up
the driver at the begining of the regulation process. A small oscillator
to keep the driver supply until the feedback regulation voltage may
supply continuously the driver.
 I don't understand how you manage one decade input voltage variation
(10...100V) but this is your business...
Vasile
{Quote hidden}

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


2001\01\19@082624 by Russell McMahon

picon face
>  So, again my ideea is good then ! You need a circuit just for power-up
>the driver at the begining of the regulation process. A small oscillator
>to keep the driver supply until the feedback regulation voltage may
>supply continuously the driver.

Yes, that is one approach I'm looking at BUT it is unfortunately not quite
that easy in this application.
While there IS nominally a battery present with a 9v rating (dropping to
more like 5 volts in some cases) the presence of a battery may not be relied
on.
The alternative source is the wide ranging 0 to 130 volt supply.
If I voltage double off this supply I will get operation at about 5 or 6
volts which is acceptable BUT I will then need to disable the voltage
multiplier as I do NOT want it to multiply the 100+ volts to 200+ :-)

An alternative is ICs like the UCC3802 family (upgradeds alternatives to the
3842) some of which have 5 volt operation. Unfortunately they are several
times the price of the industry very-standard UC3842 series and much less
available.

>  I don't understand how you manage one decade input voltage variation
>(10...100V) but this is your business...


Alternator powered by a person exercising :-).
User speed varies and load varies which alters the terninal voltage seen by
the equipment.
Voltage is typically 10 to 50 volts at light loads and 5 to 15 at very heavy
loads but can rise to 130 volts with VERY enthusiastic user at zero load.
Worst case user MUST be designed for :-)
Under heavy loads 10 volts plus is not reached until an unacceptably high
user speed.


RM

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


2001\01\19@101815 by Roman Black

flavicon
face
Russell, what did you think of my self-excited oscillator
idea? This can be made to startup at 2v and will work
fine to 130v with a simple on/off voltage regulation.

I see so many self-ex switchmode primaries lately when
fixing VCRs. They use one transistor/fet, one optocoupler
and a few passives for the whole primary! You don't need
the expense of a controller chip, unless you need really
tight output voltage regulation.
-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\01\19@102647 by mike

flavicon
face
On Sat, 20 Jan 2001 01:05:19 +1300, you wrote:

{Quote hidden}

..couldn't you just clamp the upper voltage with a simple shunt
regulator, to make your input range more sensible - doesn't seem like
you'd be wasting much power, and there are plenty of SM regs for <50V
input applications.
--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2001\01\19@104551 by Alan B. Pearce

face picon face
>..couldn't you just clamp the upper voltage with a simple shunt
>regulator, to make your input range more sensible - doesn't seem like
>you'd be wasting much power, and there are plenty of SM regs for <50V
>input applications.

As in a bit like motor cycles used to do to limit the alternator output voltage
so they did not need to put in a regulator - just a big hefty zener diode on a
heatsink. I had been thinking along the same lines - might mean those peddling
will ease off a bit. If you could also blink a led when this happens then they
will know they are producing excess output energy that is going to waste - or is
this a fitness machine?

BTW I believe that this is effectively the method used to regulate the output
voltage of the Bayliss windup radio's and flashlights. There is some sort of
beefy transistor across the output of the generator, and when the voltage gets
too high an electronic circuit turns on the transistor and loads the output of
the generator so the voltage drops sufficiently. Ugh!

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


2001\01\19@172102 by Peter L. Peres

picon face
Add a small self-oscillating converter to the auxiliary supply of the 384x
to supply its 12V supply until the circuit starts properly. The converter
should have double switchoff, once by the function of the main converter,
and once by the primary voltage being higher than say 12V.

I have done similar things, though lower powered, and discrete (no IC).

imho the transformer is very poorly used if it is to span the entire
voltage scale you indicate. Usually in these cases there are two or more
primaries and primary drive circuits, one for each range, with separate
inputs.

Peter

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


2001\01\20@084009 by Russell McMahon

picon face
>Russell, what did you think of my self-excited oscillator
>idea? This can be made to startup at 2v and will work
>fine to 130v with a simple on/off voltage regulation.
>
>I see so many self-ex switchmode primaries lately when
>fixing VCRs. They use one transistor/fet, one optocoupler
>and a few passives for the whole primary! You don't need
>the expense of a controller chip, unless you need really
>tight output voltage regulation.
>-Roman


Well, I'm afraid that that seems to work quite well! :-)
I was not too keen to go to a discrete design as parts can tend to
proliferate and with the large input voltage range I was quite keen to have
a closed cycle by cycle current loop as well as the voltage control loop.
You can of course do that with discretes but again parts count rises. I
think an LM339 quad comparator (which has the virtue of being about the
cheapest IC out) would do everything and have 1 or 2 sections left over.

However, I decided to try a "cheap and nasty" design using a 40106 hex
Schmitt trigger.
One gate oscillates and is followed by 2 more as buffers with a resistor
from gate two out to gate 3 in.
If you ground gate 3 input  (with a transistor or whatever) the oscillation
output from gate 3 stops (this is a wired OR style AND gate) .

Drive a FET with gate 3 using a 2 transistor push sort-of-pull driver.

Two coils on core. One in FET drain and other is ringing choke output.
Rectify, smooth.
Connect zener from this output via a resistor to a common emitter npn
transistor.
When  Vout exceeds Vzener + Vbe (very roughly) the transistor turns on.
Connect transistor collector to gate 3 input.
Voila - voltage regulation.
Very crude.

At present I am driving 40106 from a fixed supply but it will self power off
own derived supply easily.
Should start with under 3 volts or so on Vcc and self boot itself to full
output very quickly.

With a 50% max duty cycle (oscillator was 50:50 square wave) Vout van not
exceed Vin (but can be less).
Adjust oscillator to 66:33 square wave as seen at   FET gate

Output will now be "more" than input before regulation cuts in.
In practice this makes Vout about 1.5x Vin which is not quite enough in very
very very worst case but a little more playing (design?) will see it right.

Works extremely well considering.
Loop gain?, Phase margin?, .... ??? - you've got to be kidding :-) :-(.
Now I've got to have a good look at this approach and see how it compares to
the other contenders.

1 x 40106
3 x BC337 or equivalent.
1 x zener
1 x osc cap
smoothing & decoupling caps
diodes
resistors

Maybe I need to look at your entirely self oscillating design too.
Back to the good old days of the self resonant ringing choke converter :-).



regards




     Russell McMahon
_____________________________

What can one man* do?
Donate food daily free !!! -  http://www.thehungersite.com/
Donate Vitamin A!  http://www.thechildsurvivalsite.com/
http://www.rawa.com  - one perspective on Afghanistan
http://www.changingourworld.com    http://www.easttimor.com   http://www.sudan.com

(* - or woman, child or internet enabled intelligent entity :-))

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


2001\01\20@112308 by Roman Black

flavicon
face
Glad to see you got a result from that idea! I know the
on/off regulation is a bit rough but what the hey. :o)

Just a quick note, I would get rid of the chip, you
don't need it. Run a one-transistor self excited osc
for the primary, and keep your zener/transistor to
kill it for regulation. In many of the new VCRs I see,
the osc is a high gain tran or fet, in a typical
self-ex configuration, and the feedback opto affects
its turn on period to give regulation. Very simple.

For the self-ex primary, try a high gain tran like
a cheap darlington, with base to + through 56k resistor.
This will give a fairly guaranteed turn on at most
voltages. It's emitter to gnd (obviously) and its
collector to + through the main primary. Then use
a feedback winding that turns the thing off. ie;
start resistor turns it on, feedback winding turns
it off, it oscillates. Now all you need to do is put
the other transistor in the feedback circuit somehow
and you can get anything from on/off to a reasonable
linear regulation.

Shouldn't be that hard, and takes two transistors, NO
chips and a few discretes. You can probably even get
rid of the zener and use the second transistor's
base voltage and a resistor divider, you will get
a few % voltage variance between units but this is
probably no big deal.

Sometimes simple is best?? :o)
-Roman

PS. One potential problem with self-ex primaries is
the worst case deal where the feedback winding can't
quite turn the transistor off. Can you say "exploded
fuse"?? Many circuits include a driver transistor
that switches the main chopper on, and if it doesn't
oscillate the driver tran can't keep the main tran
biased on. So it takes 3 transistors total. Such
is life! ;o)



Russell McMahon wrote:
{Quote hidden}

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


More... (looser matching)
- Last day of these posts
- In 2001 , 2002 only
- Today
- New search...