Searching \ for '[PIC]: 2-tran SMPS 5v reg now 90% efficicient!' 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/microchip/devices.htm?key=pic
Search entire site for: '2-tran SMPS 5v reg now 90 efficicient!'.

Exact match. Not showing close matches.
PICList Thread
'[PIC]: 2-tran SMPS 5v reg now 90% efficicient!'
2002\09\22@175111 by Roman Black

flavicon
face
I have done some more testing on my new 2-transistor
SMPS 5v regulator, and it continues to look good. :o)

Regulation is now good enough that there is no need for
a secondary regulator like a 7805, and this simple
2-transistor regulator will work at very high efficiencies
and do the complete 5v SMPS buck conversion and regulation
in the one process. No 7805 needed.

I have made 3 "application" circuits, built and tested.
Anyone who needs a high efficiency 5v switching regulator
can simply copy my parts values and it will perform as
advertised. :o)

Already efficiency is better than a LM3485 (switcher chip
that was discussd here recently) over a decent current
range and LM3485 solution uses a FET where my "2-transistor
Black regulator" only uses a cheap BC327. I have had it to
over 91% with the BC327 and will try some high efficiency
tests later once I buy a few common P-FETs.

While i'm testing application circuits I am accepting
suggestions for applications and will build and test any
apps you might have for specific input and output voltages
and currents etc. :o)

It's here:
http://www.romanblack.com/smps.htm
-Roman

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


2002\09\22@225219 by myke predko

flavicon
face
Hey Roman,

Very nice circuit (and great page describing it as Richard said).  Just out
of curiosity (and this probably exposes my ignorance of switch mode power
supplies), but how much heating of the coil do you get?  I would think this
is the portion of the circuit where most of the power is lost, whereas in a
"normal" SMPS, most power is lost through the switching transistors - but
because they switch between hard on and hard off, very little power is
dissipated in them?

As a suggestion, you might want to submit this for publication (or even
patenting) to prevent somebody from stealing the idea from you and profiting
from it.  Submitting it to a journal like EDN would probably be sufficient
to prevent anybody else from patenting it.  At the very least, you should
print out a couple of copies of the web page and send them in separate
registered letters to yourself so you have proof that you came up with the
idea first.

Any chance of finding the 3 Transistor step up regulator circuit?  I'd love
to have something cheap that would allow me to drive a PICmicro MCU from a
"AA" battery or to create Vpp voltage in a programmer circuit.

Thanx for the circuit and explanation,

myke
{Original Message removed}

2002\09\23@034028 by Also-Antal Csaba

flavicon
face
> As a suggestion, you might want to submit this for publication (or even
> patenting) to prevent somebody from stealing the idea from you and profiting
> from it.

Good job. But... if you publish your idea, then you can't get patent
protection.

udv
Csaba

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


2002\09\23@035740 by Wouter van Ooijen

face picon face
> Good job. But... if you publish your idea, then you can't get patent
> protection.

IMHO Roman has already published his design (on the internet and on the
piclist), and granted use under his 'hippyware' license. Although not
easy to prove in court, this is enough to
- prevent him from patenting (you can't patent what you have already
published)
- prevent anyone else from patenting

Of course a publication in a magazine will make the proving part much
easier.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products

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


2002\09\23@045914 by Russell McMahon

face
flavicon
face
> > Good job. But... if you publish your idea, then you can't get patent
> > protection.
> IMHO Roman has already published his design (on the internet and on the
> piclist), and granted use under his 'hippyware' license. Although not
> easy to prove in court, this is enough to prevent him from patenting
> (you can't patent what you have already
> published)
> - prevent anyone else from patenting

In fine print at the TOP of his web page (not at the bottom where lawyers
put theirs :-) ) Roman has explicitly made the design public domain. It is
very probably worth far more as a circuit which may end up with his name on
long term than as a patent protected design. The original circuit topology
came from Richard's relay driver circuit (and I don't know where that came
from) but Roman has both re-targeted the application and refined it beyond
easy recognition. The addition of the "hysteresis" capacitor is new AFAIK
and transforms (no pun intended) the operation.

In many countries (including NZ) publication prior to patenting makes a
concept unable to be patented (and therefore public domain even when this is
not stated.) In the USA (and presumably elsewhere) you have one year AFTER
publication i which you can still file a patent. I don't know whether this
applies after you have expressly made a design public domain.

The two key points which make this circuit "special", as used in this
topology, are (IMHO)

i    the use of the transistor as a switch and comparator simultaneously (by
using the emitter as the output sense input and

ii    driving the zener reference negative with capacitor feedback when
switch-off occurs.

The first was a feature of the original circuit but the second is entirely
original. (I just checked Richard's original circuit (August 2001) to make
sure I could say that safely.)

The brilliance of Roman's circuit is that it *seems* "obvious" after the
event, but such circuits are far from obvious before they exist. Every
component has a place, operation is intuitive (unlike my 3 transistor
version which provoked protracted debate as to whether there was formal
hysteresis present) and it's not clear how one would readily improve it
without adding substantial complexity.

I can't see any reason why Roman's circuit couldn't be easily enough altered
to operate at much higher voltages. I'm about to design the next generation
of the exercise controllers which sparked my original challenge. One version
of this will still need a wide supply range high voltage buck converter.
I'll certainly try Roman's circuit to see how well it performs in this
application. (10 - 200 volts in -> 10 volts out. 0 - 500 mA out. Regulation
& efficiency not especially critical. Reliability crucial. ) I'll probably
use a PFET as the pass transistor as this greatly reduces the energy
dissipation in the drive resistor(s) at high voltages - a not insignificant
consideration.




       Russell McMahon

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


2002\09\23@051614 by Roman Black

flavicon
face
myke predko wrote:
>
> Hey Roman,
>
> Very nice circuit (and great page describing it as Richard said).  Just out
> of curiosity (and this probably exposes my ignorance of switch mode power
> supplies), but how much heating of the coil do you get?  I would think this
> is the portion of the circuit where most of the power is lost, whereas in a
> "normal" SMPS, most power is lost through the switching transistors - but
> because they switch between hard on and hard off, very little power is
> dissipated in them?

Hi Myke, the coil losses are very easily calculated as
the current in the coil remains constant DC current at
the same level as the output current, and the coil
dc ohms is known.

With a low-dc ohms coil the circuit gets very
efficient, but the lower the dc ohms the larger the
wire and larger the coil size for the same inductance.

> As a suggestion, you might want to submit this for publication (or even
> patenting) to prevent somebody from stealing the idea from you and profiting
> from it.  Submitting it to a journal like EDN would probably be sufficient

Thanks! Actually it has been "published" in a fashion
here on the Piclist. :o) I really hope people profit from
the idea, that's the idea of me releasing it as hippyware;
* use it for free if they like
* pay me a little something if they like (yeah sure! :o))

I do take your point about printed literature though.
-Roman

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


2002\09\23@055438 by Roman Black

flavicon
face
Wouter van Ooijen wrote:

> IMHO Roman has already published his design (on the internet and on the
> piclist), and granted use under his 'hippyware' license.

Yes. PLEASE use the circuit for any and all needs!

5v regulators are a mainstay of all PIC and other
logic applications. Having a VERY cheap and good
performance SMPS regulator is a cool device to use
and share.

> Of course a publication in a magazine will make the proving part much
> easier.

Hmm, that's 2 magazine suggestions. Has anyone here
submitted something of their own to a magazine and
would like to share some tips? :o)
-Roman

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


2002\09\23@070107 by Alan B. Pearce

face picon face
>Hmm, that's 2 magazine suggestions. Has anyone here
>submitted something of their own to a magazine and
>would like to share some tips? :o)

Well I know Electronics & Wireless World are always after circuits for their
design ideas section, and I believe they pay something like GBP100 (handy
number of Aus $) for circuits. It seems they are very short, even pleading
for "back of envelope" sketch circuits which they will redraw for
publication. Send it in with a bit of circuit operation explanation :)

Next step a buck-boost version ? I want to run the PicList Designer Board
from my 3V cell phone charger :))))

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


2002\09\23@072403 by Roman Black

flavicon
face
Russell McMahon wrote:
>
> > > Good job. But... if you publish your idea, then you can't get patent

> In fine print at the TOP of his web page (not at the bottom where lawyers
> put theirs :-) ) Roman has explicitly made the design public domain.

Actually "hippyware" but the lawyers will tell you
that still means you can use it for free if you choose.

> It is
> very probably worth far more as a circuit which may end up with his name on
> long term than as a patent protected design.

Exactly how I feel about it. This design works well and
NEEDS to be shared. :o)


> The original circuit topology
> came from Richard's relay driver circuit (and I don't know where that came
> from) but Roman has both re-targeted the application and refined it beyond
> easy recognition. The addition of the "hysteresis" capacitor is new AFAIK
> and transforms (no pun intended) the operation.

I think there is a blurring of the definitions of
hysteresis, phase delay, and monostable in oscillator
design. I don't like the term hysteresis here as there
is no dc hysteresis at all, only one caused by a phased
delay. As the delay period is a clearly defined RC charge
period from Rz to C1&C2, and the circuit is totally stable
during this timed OFF period, I think it is best defined
as a "monostable" type delay as used in 555 timers and
many switching regulator chips.

When the circuit hits the target output voltage it turns
OFF for a timed period.


> In many countries (including NZ) publication prior to patenting makes a
> concept unable to be patented (and therefore public domain even when this is
> not stated.) In the USA (and presumably elsewhere) you have one year AFTER
> publication i which you can still file a patent. I don't know whether this
> applies after you have expressly made a design public domain.

Good info to know. I don't mind how much money people
make from this idea provided i'm recognised as the
source. No patents needed for that.

> ii    driving the zener reference negative with capacitor feedback when
> switch-off occurs.

> The brilliance of Roman's circuit is that it *seems* "obvious" after the
> event, but such circuits are far from obvious before they exist.

That may well be the nicest compliment I have ever
received. :o)

The "monostable" effect accounts for more than half of
the efficiency gains as it markedly slows the switching
frequency by 2x or 3x. This reduces the switching losses
compared to the amount of energy in the inductor for
each pulse, and where improving switching speed itself
(from L feedback) gives about 5% better efficiency over
the original circuit, the monostable forces much longer
off and on periods and adds another 8 to 10% improvement
in efficiency.

The C2 dump back into the Zener causes it to peak and then
re-stabilise at 5.6v, this gives a very repeatable zener
voltage for each turn off point. ie great regulation.


{Quote hidden}

Cool, i'd like to know it goes. I'm going to
fiddle tonight with a 40v to 12v design as asked
by Alexandre, and then maybe try adapting that to
higher voltages.
:o)

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


2002\09\23@081729 by Russell McMahon

face
flavicon
face
> Well I know Electronics & Wireless World are always after circuits for
their
> design ideas section, and I believe they pay something like GBP100 (handy
> number of Aus $) for circuits. It seems they are very short, even pleading
> for "back of envelope" sketch circuits which they will redraw for
> publication. Send it in with a bit of circuit operation explanation :)

If sending to E&WW ENSURE that they are aware of how radical it is in its
efficacy and simplicity (while of course not APPEARING to overdo it :-) ) -
they have occasional competitions that pay far better than their regular
fare - oscilloscopes and similart as prizes. I suspect it may be worthy of
competing in such a forum. Trouble is that if delayed for that period and it
gets wide circulation others may present it as their work to E&WW (as is
almost legitimate as ccts are often improvements on existing ones).

> Next step a buck-boost version ? I want to run the PicList Designer Board
> from my 3V cell phone charger :))))

A Buck Boost (BB) could use the same general principles BUT will need some
form of driven oscillator (or at minimum an on time delay or current tripped
switch) as the on cycle of a BB shorts the inductor output to ground and
Vout dies NOT rise during that period so the simple 1 transistor and zener
"comparator" cannot be used to terminate the on cycle.


           Russell McMahon

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


2002\09\23@095351 by Dave Tweed

face
flavicon
face
Roman Black <spamBeGonefastvidspamBeGonespamEZY.NET.AU> wrote:
> > Of course a publication in a magazine will make the proving part much
> > easier.
>
> Hmm, that's 2 magazine suggestions. Has anyone here
> submitted something of their own to a magazine and
> would like to share some tips? :o)

Circuit Cellar has a new feature called "Short Solutions" that this would
be ideal for. Instead of a full-length article of 3000 (+/-500) words, you
do a shorter write-up of 500-1000 words.

To get started, just send an email with a 1-paragraph (or so) abstract
to <TakeThisOuTjennifer.huberEraseMEspamspam_OUTcircuitcellar.com>. Visit the webiste for additional
editorial guidelines:
  http://www.circuitcellar.com/authors/

Oops, I just noticed that that page still lists Rob Walker as the managing
editor. Actually, he's moved on, and the managing editor is now Jennifer
Huber, as listed under "Contact Info".

-- Dave Tweed

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


2002\09\23@175837 by Dave King

flavicon
face
>Cool, i'd like to know it goes. I'm going to
>fiddle tonight with a 40v to 12v design as asked
>by Alexandre, and then maybe try adapting that to
>higher voltages.
>:o)

Roman buddy pal best friend .. Just going to ask
if you could do the same thing. ;-]  What would it take
to run that voltage range and up to say about 3 amps?

This is a bit of serendipity. I found out I can't get any Max727
for  a few months if at all, or I need to play with a LM2576
and see how it behaves.

Cheers

Dave

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


2002\09\23@230518 by Roman Black

flavicon
face
Dave King wrote:
>
> >fiddle tonight with a 40v to 12v design as asked

> Roman buddy pal best friend .. Just going to ask
> if you could do the same thing. ;-]  What would it take
> to run that voltage range and up to say about 3 amps?
>
> This is a bit of serendipity. I found out I can't get any Max727
> for  a few months if at all, or I need to play with a LM2576
> and see how it behaves.

Hi Dave (buddy?) :o)
So you want 24v to 40v in, 12v @ 3A out?

Getting close to 40v input means the buck diode needs
to be rated for 60v, which is starting to push the
common type schottkys. I can get TO220 60v 16A diodes
for about $1.10 USD, provided you take adequate
measures for automotive spikes etc that type would do.

It will only be about 1.5A input so maybe a cheap
n-channel FET could be adapted, or even go with a
low-tech (cheap) bipolar like the BD204 with low sat
spec (and maybe a driver transistor) would work.

I'd like to give it a go in the simplest 2-tran
config I can manage, will be a bit of fun. What sort
of requirements do you have for;
* total size
* inductor type
* cost sensitivity
* quantity
* input filtering
* will it be compared with a specific chip?
:o)
-Roman

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


2002\09\24@004917 by Dave King

flavicon
face
>So you want 24v to 40v in, 12v @ 3A out?
Not quite.  I need is as low a input voltage as I can
get, and then up to about 35-40 max. Say the input range
is  7Vin to 40Vin output needs to be 5v and 2-3A.

>Getting close to 40v input means the buck diode needs
>to be rated for 60v, which is starting to push the
>common type schottkys. I can get TO220 60v 16A diodes
>for about $1.10 USD, provided you take adequate
>measures for automotive spikes etc that type would do.

I think the actual max Vin I'd see would be 30v.

>It will only be about 1.5A input so maybe a cheap
>n-channel FET could be adapted, or even go with a
>low-tech (cheap) bipolar like the BD204 with low sat
>spec (and maybe a driver transistor) would work.

1.5a input or output?

>I'd like to give it a go in the simplest 2-tran
>config I can manage, will be a bit of fun. What sort
>of requirements do you have for;


>* total size

Well if this works its replacing one TO-220-5.
So the smaller the area the better.

>* inductor type

No pref.

>* cost sensitivity

CHEAAAAAAAAAP er ah total under about $5

>* quantity

No idea.

>* input filtering

source voltage will be regulated dc ie auto alternator/reg

>* will it be compared with a specific chip?

Yup. The mark to beat is the Max727ECK

Dave

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


2002\09\24@020915 by Alan Shinn

picon face
------------------------------
I wish I could remember the name - One of the EE mags has a running
"contest" for best circuit of the month with a ($50) or so prize.
Date:    Mon, 23 Sep 2002 19:50:53 +1000
From:    Roman Black <RemoveMEfastvidEraseMEspamEraseMEEZY.NET.AU>
Subject: Re: [PIC]:  2-tran SMPS 5v reg now 90% efficicient!


> Of course a publication in a magazine will make the proving part much
> easier.

Hmm, that's 2 magazine suggestions. Has anyone here
submitted something of their own to a magazine and
would like to share some tips? :o)
-Roman
--
Looking forward:
Alan Shinn


Experience the
beginnings of microscopy.
Make your own replica
of one of Antony van Leeuwenhoek's microscopes.
visit    http://www.mindspring.com/~alshinn/

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


2002\09\24@031849 by Mike Singer

picon face
part 1 9146 bytes content-type:multipart/alternative; (decoded quoted-printable)


------ 2002\09\24@102947 by Dave Mumert
flavicon
face
Hi All

EDN publishes 'Design Ideas' each month.  Check the web site at
http://www.e-insite.net/ednmag/index.asp?layout=webzine

Look part way down the page under 'What's Up' for 'Submit a Design Idea.

Here is the direct link
http://www.e-insite.net/ednmag/index.asp?layout=siteInfo&doc_id=30988

They pay $100 per published entry plus $100 if your entry is the best of the
month, plus $1500 if it is the best of the year.

Dave

{Quote hidden}

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


2002\09\24@145646 by Dave King

flavicon
face
>EDN publishes 'Design Ideas' each month.  Check the web site at
>http://www.e-insite.net/ednmag/index.asp?layout=webzine
>
>They pay $100 per published entry plus $100 if your entry is the best of the
>month, plus $1500 if it is the best of the year.
>
>Dave

Hey if we can find enough of these for Roman he might not get rich but he
could buy the beer ;-]

Dave K

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


2002\09\24@164552 by Russell McMahon

face
flavicon
face
Nice little (literally) device but ...

WARNING: Note that in shutdown the IC draws typically 0.01 uA *BUT* the
output voltage is a diode drop less than the input !!! Unless care is taken
to ensure that the load does not draw current at this voltage the battery
will continue to be drained.

       RM

_________________________


Example 1 Li-Ion cell to whatever converter.
From http://www.national.com/ds/LM/LM2703.pdf :

The LM2703 is a micropower step-up DC/DC
(Li-Ion 12V Application - attached .gif)
- 2.2V to 7V input range
- 0.015A shutdown current


myke predko wrote:
> ...I'd love to have something cheap that would allow
> me to drive a PICmicro MCU from a "AA" battery or to
>create Vpp voltage in a programmer circuit...

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


2002\09\24@172843 by Mike Singer

picon face
Good morning, Russell.
  Thanks.
     Mike. (0:25 am).


Russell McMahon wrote:
> WARNING: Note that in shutdown the IC draws typically
> 0.01 uA *BUT* the output voltage is a diode drop less
> than the input !!! Unless care is taken to ensure that the
> load does not draw current at this voltage the battery
> will continue to be drained.

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


2002\09\25@075832 by Mike Singer

picon face
part 1 2540 bytes content-type:text/plain; (decoded 7bit)

Roman Black wrote:
> myke predko wrote:
> > Hey Roman,
> > Very nice circuit (and great page describing it as Richard
> > said). Just out of curiosity (and this probably exposes my
> > ignorance of switch mode power supplies), but how much
> > heating of the coil do you get?  I would think this is the
> > portion of the circuit where most of the power is lost,
> > whereas in a "normal" SMPS, most power is lost through
> > the switching transistors - but because they switch between
> > hard on and hard off, very little power is dissipated in them?
>
> Hi Myke, the coil losses are very easily calculated as
> the current in the coil remains constant DC current at
> the same level as the output current, and the coil
> dc ohms is known.

Roman,
"the current in the coil remains constant DC current" when
you use PWM over big inductances: motor, solenoid coils.
Here you should take a compromise, current is to pulsate
significantly. Look, please, at attached "Continuous Mode
Operation.gif" (LM3485). So some integration is needed
to calculate loses.
As well inductances core loses should be compromised.
You wrote earlier:
>  Pout      Freq
> 0.099w  97kHz
> 0.245     57
> 0.390     41
> 1.205     31
Look, please, at attached "Operating Frequency vs Output
Load Current.gif". Big DC/DC guys prefer lower frequencies
at lower loads, making converter more PFM then PWM at
lower loads shifting to "Discontinuous Mode Operation".
You have higher frequencies at lower loads, charging-
discharging parasitic capacitances even more often.

Anyway, I have to admit your converter is very useful for
EE education, much more then integrated one.

But, I think, you should be more careful when saying
something sorta:

> Already efficiency is better than a LM3485 (switcher chip
> that was discussd here recently) over a decent current
> range and LM3485 solution uses a FET where my "2-transistor
> Black regulator" only uses a cheap BC327. I have had it to
> over 91% with the BC327 and will try some high efficiency
> tests later once I buy a few common P-FETs.

Because I'm not sure you used the same labs, as National
Semi gang used.

Mike, who appreciate greatly your contribution to the List (really).
-------
"Plato is my friend, Aristotle is my friend, but my best
friend is truth." Isaac (and I hope James) Newton.




part 2 10269 bytes content-type:image/gif; (decode)


part 3 9646 bytes content-type:image/gif; (decode)


part 4 9418 bytes content-type:image/gif; (decode)


part 5 144 bytes (decode)

2002\09\25@143741 by Roman Black

flavicon
face
part 1 2596 bytes content-type:text/plain; charset=us-ascii (decoded 7bit)

Mike Singer wrote:

> Anyway, I have to admit your converter is very useful for
> EE education, much more then integrated one.
>
> But, I think, you should be more careful when saying
> something sorta:
>
> > Already efficiency is better than a LM3485 (switcher chip
> > that was discussd here recently) over a decent current
> > range and LM3485 solution uses a FET where my "2-transistor
> > Black regulator" only uses a cheap BC327. I have had it to
> > over 91% with the BC327 and will try some high efficiency
> > tests later once I buy a few common P-FETs.
>
> Because I'm not sure you used the same labs, as National
> Semi gang used.


Ok, firstly I DID use the same labs, as I got *their
efficiency curve* from *their data sheet*, which I know
you have just like I do! It came off the same page you
sent the current charts from. I noticed you didn't send
their efficiency chart when you sent the other 3. :o{

So I have attached *their* efficiency chart, with two lines
added by me.
Blue line = their switcher at 15v in 5v out.
Red line = my switcher at 15v in 5v out
(from my circuit test here):
http://www.romanblack.com/a02.htm

You can clearly see that my circuit is up to 3% to 4% more
efficient than theirs, in that important 70mA to 200mA
range, which was my goal.

Now add this:
* their switcher uses a FET, mine just a cheap BC337
* they use a low-ohms inductor, my test used a high ohms part

Now my test WAS ALREADY more efficient than theirs.

AND if I used the same expensive FET they used, it would
same me about 0.2Vce which is another 1% or so.

AND when I have used a low ohms inductor in my other
tests it gains another 1% to 2%.

AND my circuit saves the cost of an IC!

> "Plato is my friend, Aristotle is my friend, but my best
> friend is truth." Isaac (and I hope James) Newton.

This sounds like you are calling me a liar???

I know you like the LM3485 and made a post about it
earlier. Since you have it's datasheet I'm concerned
that you went out of your way to say I was wrong
when you actually have the data...

I have done a lot of work on this project and released
it for the benefit of all. I specifically downloaded
the LM3485 datasheet and checked my facts before
claiming that my circuit was "better over a decent range"
and I don't appreciate being called a liar. What
adds the final annoyance is that you wasted the list
bandwidth by attaching 3 charts but you DIDN'T post the
chart that solved the argument.
-Roman


part 2 12764 bytes content-type:image/jpeg; name="Eff3485.jpg" (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


2002\09\25@150753 by Roman Black

flavicon
face
Mike Singer wrote:

> > Hi Myke, the coil losses are very easily calculated as
> > the current in the coil remains constant DC current at
> > the same level as the output current, and the coil
> > dc ohms is known.
>
>  Roman,
> "the current in the coil remains constant DC current" when
> you use PWM over big inductances: motor, solenoid coils.
> Here you should take a compromise, current is to pulsate
> significantly. Look, please, at attached "Continuous Mode
> Operation.gif" (LM3485). So some integration is needed
> to calculate loses.
>  As well inductances core loses should be compromised.

I understand your point. But realise that their circuit
uses a VERY low inductor, 22uH. My test was with a 470uH
inductor at the same current, which you will know won't
have the same amount of current ripple that theirs does.
That means less core losses and less output ripple with
my design.

And I also stick to my statement that "the coil losses
are easily calculated from the current" as the statement
is correct (enough) in answering Myke's question.

Checking the LM3485 circuit on the datasheet front page
shows their circuit with 4 caps, 3 resistors, fet, buck
diode.

My circuit has *exactly the same parts count* with the
one change that I have replaced an entire chip with one
NPN transistor. Ok so it doesn't have current limiting
(unless you add a transistor, see my new circuit here):
http://www.romanblack.com/a02.htm

but for *much less parts cost* it CAN perform better than
a chip+fet design. :o)
-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


2002\09\25@160316 by Mike Singer

picon face
Roman Black wrote:
> > "Plato is my friend, Aristotle is my friend, but my best
> > friend is truth." Isaac (and I hope James) Newton.
>
> This sounds like you are calling me a liar???

Why? Were Plato and Aristotle called liars? If so, you
are in a good company. :-)


> I have done a lot of work on this project and released
> it for the benefit of all.

I appreciate it sincerely and hope my posting will serve
your project to be even greater!


> What adds the final annoyance is that you wasted the list
> bandwidth by attaching 3 charts but you DIDN'T post the
> chart that solved the argument.

I'm not sure I "wasted the list bandwidth" pointing out
by these 3 charts that modern DC/DC converters use mixed
PWM/PFM mode to gain efficiency with decreasing frequency
at lower loads, not increasing. (and showing that current
through the inductor should not be constant, as you wrote)
Yes, "I DIDN'T post the chart that solved the argument",
but if I did, my posting would be rejected by ListServer
for the posting's big volume.

Mike.
-------

We find comfort among those who agree with us - growth
among those who don't.   - Frank A. Clark -

(Hope this quotation on "final annoyance" theme will
not be offensive to you :-)

--
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


2002\09\25@170905 by Morgan Olsson

picon face
Regarding efficiency and other spec i use to be very careful.

It takes much calculations and verifying labs to make a design that performes good in all conditions High/low temp, High/low input volt etc, while also have margins for component variations such as min/max hFE of transistors, ESR of caps, etc.  Maybe some combinations cause stability problems.  Also remember zeners have typically 5% tolerance, add voltage change due to current change due to input voltage change, regulation of input volt, and load, and i guess simple regulators like theese have problems qualifying in a lot of areas.

For instance i guess the Black regulator would drop a little in efficiency if we were to have production quality margins for hFE for instance.  Using some combinations of worst case values is sometimes very tough for a cirquit, such problem i use best to solve with pen and paper, as i can not manufacture worst case variants of all componets... but substituting for other cap values, low hFE transstors etc is one way.

Also note that the relatively bigger inductors in the Black is making slower load pulse reponse and also cost more in component cost and board estate if talking production.

But for hobby use as well as some consumer products i find the Black regulator very nice.

Btw, new cadidates for the switch: BD438 (high gain high current pretty low cost TO126) and also check out Zetex "Matrix" range ZTXnnn IIRC for low voltage drop and very high gain.

Another application idea is to use it in battery charger, then with added precision ref and cirquit for end of charge if used for lead or Lithium.

For NiMH use a PIC as timer, and maybe a ref and A/D for -dV/dt.

See, this really is on toPIC  ;)

/Morgan

--
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


2002\09\25@190027 by Roman Black

flavicon
face
Dave King wrote:
>
> >So you want 24v to 40v in, 12v @ 3A out?
> Not quite.  I need is as low a input voltage as I can
> get, and then up to about 35-40 max. Say the input range
> is  7Vin to 40Vin output needs to be 5v and 2-3A.

Hi Dave, i've done some work on the 3A circuit and it's
looking do-able.

But i'm getting worried about clogging the piclist
with too many SMPS posts so if anyone asks this to be
taken offlist we should.

But I THINK i've got a good working 3A SMPS regulator
solution for very low parts cost. :o)

I just built a big version of my 2-tran regulator
using a cheap TO220 darlington, there are a heap of
"old" parts to buy cheap these days, say 100v 5A
types that all do under 1.0v saturated at 3A. Doesn't
sound that impressive, but they are CHEAP and in
one package.

By comparison, the Maxim Max727ECK switcher chip you
mentioned has 1.7 volts saturated at 2A! That's not real
good if you want a 2A switcher. So the cheap darlington
is about 1W to 2W more efficient, AND allows 3A or more
continuous loads as an added bonus.


> Well if this works its replacing one TO-220-5.
> So the smaller the area the better.
> CHEAAAAAAAAAP er ah total under about $5
> Yup. The mark to beat is the Max727ECK

Well the 2-tran solution is cheap! I'm using a 50 cent
darlington and a small signal transistor. The inductor and
buck diode costs you have with both circuits. It will be
more efficient than the 2A chip, and will do 3A, but the
ability to handle wide input voltages is not perfect
with it.

Now you've said its for 24v auto use that gives a
decent range (16v to 32v?) which should be tunable
with the CHEAP circuit i'm using now. If you want the
ability to go down to real low voltages as well it will
cost an extra transistor as a constant current source
for the zener monostable.

So the 5v 3A version of my regulator is now working
for very low cost, and about 83% efficiency without
needing a special FET etc, but before I write it up etc
can you confirm input voltage ranges and any other
details you might need. Do you really need that 7v
to 35v input range at full power and max efficiency?
-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


2002\09\26@033254 by Dave King

flavicon
face
>i've done some work on the 3A circuit and it's
>looking do-able.

Thats fantastic, at 2-3A it becomes very very useful.

>But i'm getting worried about clogging the piclist
>with too many SMPS posts so if anyone asks this to be
>taken offlist we should.

No problem. If anyone objects or you'd like to do any more off-line
just let me know.  This really is too good a thing not to be kept in
front of anyone who could use it (IMHO).

>But I THINK i've got a good working 3A SMPS regulator
>solution for very low parts cost. :o)

Cheap is good ;-]

>I just built a big version of my 2-tran regulator
>using a cheap TO220 darlington, there are a heap of
>"old" parts to buy cheap these days, say 100v 5A
>types that all do under 1.0v saturated at 3A. Doesn't
>sound that impressive, but they are CHEAP and in
>one package.

Too true. I bought a handful over the summer and paid about
2 bits each for new ones.

>By comparison, the Maxim Max727ECK switcher chip you
>mentioned has 1.7 volts saturated at 2A! That's not real
>good if you want a 2A switcher. So the cheap darlington
>is about 1W to 2W more efficient, AND allows 3A or more
>continuous loads as an added bonus.

That really is an improvement. I thought the Max727 or the LM2576
were a real answer to things but your switcher seems like a real
improvement.


>Well the 2-tran solution is cheap! I'm using a 50 cent
>darlington and a small signal transistor. The inductor and
>buck diode costs you have with both circuits. It will be
>more efficient than the 2A chip, and will do 3A, but the
>ability to handle wide input voltages is not perfect
>with it.

At what point ie high end or low end will there be a problem with
the input voltage? Would it be better to build a high and a low input range
or any problems really just going to show as a loss in efficiency?

>Now you've said its for 24v auto use that gives a
>decent range (16v to 32v?) which should be tunable
>with the CHEAP circuit i'm using now. If you want the
>ability to go down to real low voltages as well it will
>cost an extra transistor as a constant current source
>for the zener monostable.

Well 80% of what I'm thinking of using this for would be
auto voltages ie 12-13.8vdc, then say 10% around 30vdc
and possibly the rest being run off of a battery pack which
might be 7.2-9v which is the absolute lower limit.

I''d even like to try and adapt this to re switching regulator
on a bike alternator. That put out 40.

>So the 5v 3A version of my regulator is now working
>for very low cost, and about 83% efficiency without
>needing a special FET etc, but before I write it up etc
>can you confirm input voltage ranges and any other
>details you might need. Do you really need that 7v
>to 35v input range at full power and max efficiency?

It solves any foreseeable use I would have. What about
 setup that is optimized for a high or low range? ie both
retain the 5Vdc 3A out but say the first would work 7.2v
to 15v and the second might be optimized for 20-35v.
Not to make any work but is it easier to define it over a
smaller range to maximize the conversion efficiency?

Cheers

Dave

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


2002\09\26@104147 by Roman Black
flavicon
face
Dave King wrote:
>
> >i've done some work on the 3A circuit and it's
> >looking do-able.
>
> Thats fantastic, at 2-3A it becomes very very useful.

Great its working fine and fairly cool, 18W in and
15w out (5v @ 3A). I modified the "monostable" timer
part of my regulator to give more stable delay period,
more independant from inductor and load type. We can still
get rid of 1.5w by swapping the darlington for a low sat
device.

Now with a GOOD monostable delay it runs much more like
a chip, most of these stepper chips we've been discussing
do basically the same thing, 1. detect voltage, 2. timed
off period (hysteresis etc) but the result is the same;
longer more powerful pulses with a smaller inductor.

The regulator now oscillates slower for a given inductor
with a more stable frequency response and lower freq
at light loads than it had before. All good steps as Mike
and Morgan have pointed out. With this improvement it can
now use inductors of 50uH and 100uH etc. :o)


> >I just built a big version of my 2-tran regulator
> >using a cheap TO220 darlington, all do under 1.0v
saturated at 3A.

> >By comparison, the Maxim Max727ECK switcher chip you
> >mentioned has 1.7 volts saturated at 2A! That's not real
> >good if you want a 2A switcher.
>
> That really is an improvement. I thought the Max727 or the LM2576
> were a real answer to things but your switcher seems like a real
> improvement.

Funny how expensive some of these chips are. They are
definitely selling a "solution" as people are afraid
of building their own switchers and will pay big bucks
for a chip that makes it easy. :o)

Re the saturation voltages of these chips, pretty bad
at 1.7v @ 2A, when you can get a 50c zetex transistor
that does around 0.3v @ 3A and have a device beta of
about 300!! (I've ordered some!)


> >the
> >ability to handle wide input voltages is not perfect
> >with it.
>
> At what point ie high end or low end will there be a problem with
> the input voltage? Would it be better to build a high and a low input range
> or any problems really just going to show as a loss in efficiency?

I've solved some of the problem in anticipation that
you wanted that difficult WIDE input range. Some of
the solution was the improved monostable design.

The problem is simple, the timed off period is dependent
in some part on the current through RZ, to charge
C1 and C2 which give the timed period.

When you have input from 16v to 32v it will be good
enough. But going down to only 2 volts above your
zener voltage requires that Zener is fed from a
constant current source IF you also need the high
range. Either range can be cheaply tuned, for 12v auto
or 24v auto type inputs. But to do all ranges well means
the cost is another zener + transistor + resistor.
No big deal for 10 cents in parts if you are saving
$5 on a chip. :o)

> Well 80% of what I'm thinking of using this for would be
> auto voltages ie 12-13.8vdc, then say 10% around 30vdc
> and possibly the rest being run off of a battery pack which
> might be 7.2-9v which is the absolute lower limit.

A 7.2v battery run at 3A is going to soon be a 6v
battery, which is cutting it very close. Fortunately
the 2-tran regulator will generally go linear at very
low headrooms which is ok as the current gain with
those voltages is nil anyway.

How critical is it that it runs below 10v? That extra
voltage option causes compromises.

> I''d even like to try and adapt this to re switching regulator
> on a bike alternator. That put out 40.

You have to be careful again with the schottky diode
voltage, the 16A pack I have is 60v rated and costs
about 40 to 50 cents. For 40v schottky you can get
3 x 1A diodes with BETTER sat than the big TO220
pack for total of 25c. So a 60v input is going to cost
you in semi price.

> It solves any foreseeable use I would have. What about
>   setup that is optimized for a high or low range? ie both
> retain the 5Vdc 3A out but say the first would work 7.2v
> to 15v and the second might be optimized for 20-35v.
> Not to make any work but is it easier to define it over a
> smaller range to maximize the conversion efficiency?

What happens is the switcher is about 85%+ efficient in
it's happy range of 12v in, then maybe 80%+ at the higher
range of 30v in.

At 10v or less the efficiency is going to drop due to the
nature of buck converters. Since the design easily tolerates
30% power waste that means anything under 8v will be possibly
more efficient in linear anyway, and it will still regulate
well.

Can you confirm the load requirements, ie is this 3A all
the time or does it have some reduced current periods etc.
-Roman

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


2002\09\26@122400 by Dwayne Reid

flavicon
face
At 12:37 AM 9/27/02 +1000, Roman Black wrote:

>  you can get a 50c zetex transistor that does around 0.3v @ 3A and
>have a device beta of about 300!! (I've ordered some!)

Which Zetex part are you using?  I'll order some as well to play with for
when you publish your latest and greatest . . .

Thanks again for sharing!

dwayne


--
Dwayne Reid   <RemoveMEdwaynerTakeThisOuTspamspamplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax

Celebrating 18 years of Engineering Innovation (1984 - 2002)
 .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-
    `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'
Do NOT send unsolicited commercial email to this email address.
This message neither grants consent to receive unsolicited
commercial email nor is intended to solicit commercial email.

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


2002\09\27@093614 by Roman Black

flavicon
face
Dwayne Reid wrote:
>
> At 12:37 AM 9/27/02 +1000, Roman Black wrote:
>
> >  you can get a 50c zetex transistor that does around 0.3v @ 3A and
> >have a device beta of about 300!! (I've ordered some!)
>
> Which Zetex part are you using?  I'll order some as well to play with for
> when you publish your latest and greatest . . .


Hi Dwayne, I got these from the Farnell catalogue,

ZTX13P40DE6, PNP, 40v, 3A av, 10A pk,
25mV Vce @ 0.1A, beta=300 (SMT SuperSOT4)
$1.71 (100 0.88)

ZTX951, PNP, 60v, 4A av, 15A pk,
0.3V Vce @ 4A, beta=100 (TO92 flat)
$2.57 (100 1.32)

ZTX953, PNP, 100v, 3.5A av, 10A pk,
0.33V Vce @ 3.5A, beta=100 (TO92 flat)
$2.75 (100 1.51)

FDR858P, PFET, 30v, 8A,
Rds = 19mohm 0.057V Vce @ 3A, (SMT SuperSOT8)
$3.38 (100 1.80)

All with fantastic low sat performance and high
gain, and quite reasonable prices (prices shown
in AUD so halve them for US price).

In particular the PFET has incredible low Rds on
in a SMT pack, and will give an easy 4A buck
switcher that is not much bigger than the inductor
and buck diode. But the FET is pricier and may
be harder to drive from my simple circuit. The SMT
packs also represent some hassle in construction,
and i'll fiddle with the TO92 parts first.

The mega cheap BC337 i'm using is the /25 version
from ON semiconductor which is a lower sat 45v 0.8A
beta=160 version (not the cheaper Fairchild 45v 0.5A
beta=100 one) but these parts are both so cheap there
is no reason not to get the good one.
(about 6 cents US in 100s)

I did the calcs on paralleling two or more BC327
transistors, with a track width of 10 thou and 600mA
the track gets 22mV drop at 1.5 inches. This is about
5% of Vbe, which is enough to make effective emitter
resistors and balance current acceptably. As a minimum
cost solution 2x $0.06 is a much cheaper transistor than
1x $0.50 for the special transistor. :o)

> Thanks again for sharing!

You're welcome. :o)
-Roman

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


2002\09\27@100453 by Wagner Lipnharski

flavicon
face
Roman Black wrote:
[snip]
> I did the calcs on paralleling two or more BC327
> transistors, with a track width of 10 thou and 600mA
> the track gets 22mV drop at 1.5 inches. This is about
> 5% of Vbe, which is enough to make effective emitter
> resistors and balance current acceptably. As a minimum
> cost solution 2x $0.06 is a much cheaper transistor than
> 1x $0.50 for the special transistor. :o)
>
>> Thanks again for sharing!
>
> You're welcome. :o)
> -Roman

... paralleling is good, but never forget that paralleling has a weak
chain link too.  Not two transistors are made exactly the same, they
will have little different VCEs, for small that this difference can be
the smaller VCE transistor will drain more current than the other(s),
until it heats enough (by excess current) and increases its VCE, then
current start to be proportionally shared again with other(s).  If
this excess current and temperature is supported by the transistor,
then everything is ok, but if is not, then it will blows up, leaving
the problem for the other(s) transistor(s), and the cycle will repeat.
Of course, you already said - current limiting / balancing emitter
resistors will always help, but try that in SMD configuration, mostly
with SMD resistores with values less than one Ohm, and you'll see the
price going sky high.

/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
Wagner Lipnharski - UST Research Inc
Orlando FLorida - USA - http://www.ustr.net
/_/_/_/ Atmel AVR Consultant /_/_/_/

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


2002\09\27@135315 by Roman Black

flavicon
face
Wagner Lipnharski wrote:
>
> Roman Black wrote:
> [snip]
> > I did the calcs on paralleling two or more BC327
> > transistors, with a track width of 10 thou and 600mA
> > the track gets 22mV drop at 1.5 inches. This is about
> > 5% of Vbe,


> ... paralleling is good, but never forget that paralleling has a weak
> chain link too.  Not two transistors are made exactly the same,
> Of course, you already said - current limiting / balancing emitter
> resistors will always help, but try that in SMD configuration, mostly
> with SMD resistores with values less than one Ohm, and you'll see the
> price going sky high.


That's why I suggested using the PCB tracks,
which require a 10 thou track 1.5 inches long,
to give 600mA, 22mV or 37 milliohms. This is
about 5% of the Vbe voltage, and should be
plenty to allow paralleling with such low power
devices. Measurement of a handful of BC327s
gave Vbe within a few mV. I also don't mind hand
matching them either to save that much money.
:o)
-Roman

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


2002\09\27@143659 by Peter L. Peres

picon face
Imho only the FET qualifies (wow at what price) for a 3A supply, and I say
that without looking at max. dissipation which you did not list.
Transistors in TO92 tend to go bad when dissipating more than 0.5W or so
(even that is a lot).  According to your efficiency figures a 3A/5V supply
will have 15W out and burn 1.5W on the circuit (assuming efficiency stays
as good with the higher current, which I doubt). Even if only 1/3 goes to
the TO92 it will try to melt.

On Fri, 27 Sep 2002, Roman Black wrote:
*>
*>ZTX13P40DE6, PNP, 40v, 3A av, 10A pk,
*>25mV Vce @ 0.1A, beta=300 (SMT SuperSOT4)
*>$1.71 (100 0.88)
*>
*>ZTX951, PNP, 60v, 4A av, 15A pk,
*>0.3V Vce @ 4A, beta=100 (TO92 flat)
*>$2.57 (100 1.32)
*>
*>ZTX953, PNP, 100v, 3.5A av, 10A pk,
*>0.33V Vce @ 3.5A, beta=100 (TO92 flat)
*>$2.75 (100 1.51)
*>
*>FDR858P, PFET, 30v, 8A,
*>Rds = 19mohm 0.057V Vce @ 3A, (SMT SuperSOT8)
*>$3.38 (100 1.80)

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


2002\09\27@232613 by Russell McMahon

face
flavicon
face
> Transistors in TO92 tend to go bad when dissipating more than 0.5W or so
> (even that is a lot).

It is of course not ideal to run anything at it's maximum rated dissipation,
but note that these Zetex transistors are NOT TO92 - they just "look like"
TO92. They are Zetex's proprietary E-line package which they rate at 1.2w
unheatsunk dissipation. They seem to know what they are doing. They are a
thinner package than TO92 and I believe they do special things with the
leadframe to ensure minimum thermal resistance.

Looking at the specs of E-Line parts shows something interesting - the
Max_current x Vsat product is usually approximately 1.2 watts. ie they seem
to current rate their parts based on maximum allowable dissipation. They
also frequently publish a peak current rating which is several times the
continuous current rating. While the E-Line package is not intended to have
a heatsink added, this suggests that if it WAS heat-sunk then the current
ratings could probably be increased - probably at least to the peak current.
This of course would take the parts out of spec and anything could happen
but it would be interesting to see what results could be obtained. As
Rth-junction_case is usually lower than Rth-case_ambient for small parts
this suggests a well designed heatsink may increase allowable dissipation
several times.

Zetex produce parts virtually without equal from other sources. Typically
several amps with betas of 100+ and very low Vsats in TO92 sized packages.
The price is probably the only disincentive but even that's often acceptable
when the other factors are considered. For those who wonder wjere Zetex came
from, they used to be "Newmarket" in the very early days of the
semoconductor revolution. For those who haven't heard of Newmarket .....
:-).


       Russell McMahon

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


2002\09\28@044332 by Morgan Olsson

picon face
For both To92 and E-line, keeping very short leads, soldered to large copper areas is fundamental to high power dissipation.

This goes especially for the collector, on which the chip is mounted.

This is hard to find for hole mount packages, but for SMT devices they often specifies what power at what collector copper areas.  This will give an idea, then remember the heat resistance of the hole mount devices leads will degrade power handling in respecitve to SMT devices.

/Morgan

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


2002\09\28@145941 by Roman Black

flavicon
face
Russell McMahon wrote:

> It is of course not ideal to run anything at it's maximum rated dissipation,
> but note that these Zetex transistors are NOT TO92 - they just "look like"
> TO92. They are Zetex's proprietary E-line package which they rate at 1.2w
> unheatsunk dissipation. They seem to know what they are doing. They are a
> thinner package than TO92 and I believe they do special things with the
> leadframe to ensure minimum thermal resistance.


The Zetex parts look nice alright. :o)

The parts arrived from Farnell, I will fiddle
some this weekend.

One thing that is amusing me is the thought that
since my 2-tran switcher always has one or the
other hard saturated it *should* be possible to
use 2 low-sat FETs; one for the switcher and
the other for the BUCK DIODE.

With switcher and buck sats down to 0.05v or
less at low currents it would be possible to
get 97% efficiency from a still (relatively)
simple and cheap design. Now that's a weekend's
fiddling if I ever heard one. :o)
-Roman

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


2002\09\28@145947 by Roman Black

flavicon
face
Peter L. Peres wrote:
>
> Imho only the FET qualifies (wow at what price) for a 3A supply, and I say
> that without looking at max. dissipation which you did not list.
> Transistors in TO92 tend to go bad when dissipating more than 0.5W or so
> (even that is a lot).

Some very good points Peter, and i'm one of the
crowd who *really believe* in using parts much
tougher than they need to be, remember the years
I spent in TV and VCR repair... :o)

BUT, in this case the 3A, 4A etc ratings are the
official safe AVERAGE current, 10A or 15A etc are
the acceptable peak currents.

The 3A switcher is ON for approx 40% of the time,
at 3A. The average Pon is 1.2A, and add a safe 20% for
switching slope. Never more than about 1.5A average.
Peak current is of course 3A.

Now these transistors are NOT the normal T092 things,
they are advertised as special low-sat high performance
devices suitable for switching supplies.

So I have no qualms about using a special switcher
marked as a 3A average 10A peak device in a 1.5A
average 3A peak application. :o)

And YES I was probably going to heatsink them too,
it doesn't take much size or expense to sink a TO92
sized pack.

>  According to your efficiency figures a 3A/5V supply
> will have 15W out and burn 1.5W on the circuit (assuming efficiency stays
> as good with the higher current, which I doubt). Even if only 1/3 goes to
> the TO92 it will try to melt.

Actually my 3A version is running very well at 18W in,
15W out, 3W wasted. 83%. This is with a clunky old
DARLINGTON PNP, which I have properly tested at 0.95v
Vce sat @ 3A. At 40% on time that is 1.14W, and changing
to the Zetex part at 0.27v 3A sat will drop that 1.14W
back to 0.324W, saving almost a watt.

There's not a lot I can do with the buck diode, the one
I have now is 0.36v @ 3A x 60% which is 0.648W and
pretty good really. Shopping for super diodes shows
very few that perform better than this.

After the 1.14W lost on the darlington and 0.648W on the
buck diode there is a total 1.2W lost mainly on the
inductor, which at 3A is about 0.133 ohms and about what
my inductor measures. I haven't had much luck finding
a 3A inductor less than 0.1 ohms yet either! :o)

> *>ZTX951, PNP, 60v, 4A av, 15A pk,
> *>0.3V Vce @ 4A, beta=100 (TO92 flat)
> *>$2.57 (100 1.32) (price in AUD)

That's $1.41 (100 = 0.73) US price.
:o)
-Roman

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


2002\09\30@175606 by Peter L. Peres

picon face
On Sun, 29 Sep 2002, Roman Black wrote:

*>There's not a lot I can do with the buck diode, the one
*>I have now is 0.36v @ 3A x 60% which is 0.648W and
*>pretty good really. Shopping for super diodes shows
*>very few that perform better than this.

Of course you know that you can use a bipolar as buck 'diode' too. Try it.
You need another PNP with E at GND and negative base drive during flyback.
I'd put it in parallel with the diode for a first test.

I don't think that you can improve efficiency very much without adding
lots of parts.

Peter

--
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



'[PIC]: 2-tran SMPS 5v reg now 90% efficicient!'
2002\10\01@040135 by Morgan Olsson
picon face
Peter L. Peres wrote:
>Of course you know that you can use a bipolar as buck 'diode' too. Try it.
>You need another PNP with E at GND and negative base drive during flyback.

I think this will definitely reach over spec of reverse voltages over that transistor during the forward phase.  Unless input voltage is 5V or less.

/Morgan

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


2002\10\01@060847 by Russell McMahon

face
flavicon
face
> >Of course you know that you can use a bipolar as buck 'diode' too. Try
it.
> >You need another PNP with E at GND and negative base drive during
flyback.
>
> I think this will definitely reach over spec of reverse voltages over that
transistor during the forward phase.  Unless input voltage is 5V or less.

This is where a MOSFET as a synchronous switch has a definite advantage.

- The back body diode blocks during the "on" phase.
- The back diode starts to handle the current when the pass transistor turns
off.
- The gate is forward biased (unlike a bipolar transistor) even when the
current flows "backwards" through it during the off phase.
- Can use an N channel with good Rdson and low cost.

Driving the rectifier FET on positively during the off phase will probably
require one extra transistor. You would then be up to 4 (pass transistor,
control, rectifier transistor, rectifier controller). Still impressive.

Somewhere about here a pack of low cost comparators may be useful :-) - one
more transistor and an oscillator of sorts and we have boost buck. Then
.......   :-)




       RM

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


2002\10\01@084320 by Roman Black

flavicon
face
Peter L. Peres wrote:
>
> On Sun, 29 Sep 2002, Roman Black wrote:
>
> *>There's not a lot I can do with the buck diode, the one
> *>I have now is 0.36v @ 3A x 60% which is 0.648W and

> Of course you know that you can use a bipolar as buck 'diode' too. Try it.
> You need another PNP with E at GND and negative base drive during flyback.
> I'd put it in parallel with the diode for a first test.
>
> I don't think that you can improve efficiency very much without adding
> lots of parts.


Hi Peter, yes you are right about the efficiency.
As the simple circuit stands it is very good when
tuned for a set input voltage. Most improvements
come from making the efficient range wider and
flatter, which comes only after adding circuit
complexity.

Re the MOSFET diode issue, tests showed me that it
only saves maybe 0.15v saturation over the good and
cheap schottky diode, total efficiency change is
around half a percent which is just not worth it
for all the extra complexity.

It is what it is, for two *cheap* transistors and
a decent input range it can't be beat. Once you
start adding better comparators and better timers
then it's worth it just to buy a dedicated
switcher chip. :o)
-Roman

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


2002\10\01@140808 by Peter L. Peres

picon face
On Tue, 1 Oct 2002, Morgan Olsson wrote:

*>Peter L. Peres wrote:
*>>Of course you know that you can use a bipolar as buck 'diode' too. Try it.
*>>You need another PNP with E at GND and negative base drive during flyback.
*>
*>I think this will definitely reach over spec of reverse voltages over
*>that transistor during the forward phase.  Unless input voltage is 5V or
*>less.

Yes.

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


More... (looser matching)
- Last day of these posts
- In 2002 , 2003 only
- Today
- New search...