Searching \ for '[EE] Linear power supply blowing it's diodes' 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/actodc.htm?key=power
Search entire site for: 'Linear power supply blowing it's diodes'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Linear power supply blowing it's diodes'
2006\02\14@221724 by Peter Todd

picon face
Alright, no laughing, this is a serious linear power supply, not some
wimpy half-amp unit. Not that I'm claiming my design is any good...

I've got a device I built that has 64 stepper motors with associated pic
chip drivers in it. These are extremely cheap steppers,
www.allelectronics.com/cgi-bin/item/SMT-57/400600/STEPPER_MOTOR_.html
to be exact. Total current draw averages about 22A-25A but can peak as
high as about 40A. Not that I am totally sure about those numbers,
that's just assuming that the amp meter on my testing power supply
works. The motors are not synchronized to each other. To be exact, they
play Conways Game of Life...

The the problem is as follows. I built a linear power supply using a 28A
6.3V transformer from digikey. 237-1255-ND. 4x LM396K 10A regulators in
parallel through load-sharing resistors regulate the voltage. For my
first attempt I used a standard 50A 600V diode bridge, MP506W-BPMS-ND
That didn't work, the final voltage at the "5V" line ended up being
about 3.5V, causing the master PIC controller to crash.

My second bright idea (ha!) was to try to reduce the voltage drop by
using those fancy schottky diodes I noticed in the catelogs... So I
ordered three 115CNQ015ASL-ND 15V 110A and made them into a diode
bridge. (they are two diode packages) That worked beautifully, giving me
a usable 4.2V line, until after about 30 minutes of running I heard a
loud bang. The main AC fuse had blown, though not the 40A DC fuse, and
the after some testing the diodes were found to have developed internal
shorts. All three seem to have done this at once!

So, any ideas where I was going wrong? One possible culprit I think is
how I soldered them, I made up a circuit board and used a heatgun to
heat the whole packages up and "reflow" the heat-spreader tabs to the
PCB. All this done without much regard to temperature other than go
slowly and use as little heat as need be.

In any case, I just want to get some ideas before I go off and
potentially spent another $40 on diodes, or try to find a bigger
transformer.

--
spam_OUTpeteTakeThisOuTspampetertodd.ca http://www.petertodd.ca

2006\02\14@225156 by Chen Xiao Fan

face
flavicon
face
> -----Original Message-----
> From: .....piclist-bouncesKILLspamspam@spam@mit.edu
> [piclist-bouncesspamKILLspammit.edu] On Behalf Of Peter Todd
>
> My second bright idea (ha!) was to try to reduce the voltage drop by
> using those fancy schottky diodes I noticed in the catalogs... So I
> ordered three 115CNQ015ASL-ND 15V 110A and made them into a diode
> bridge. (they are two diode packages) That worked
> beautifully, giving me a usable 4.2V line, until after about 30 minutes
> of running I heard a loud bang. The main AC fuse had blown, though not
> the 40A DC fuse, and the after some testing the diodes were found to have
> developed internal shorts. All three seem to have done this at once!
>
> So, any ideas where I was going wrong? One possible culprit I think is
> how I soldered them, I made up a circuit board and used a heat gun to
> heat the whole packages up and "reflow" the heat-spreader tabs to the
> PCB. All this done without much regard to temperature other than go
> slowly and use as little heat as need be.

Check the specification of those Schottky diodes. Often they can not
withstand high reverse voltage. I think this could well be the
reason of failure for your diodes.

Do you have Varistors before the diode bridge? This might help as well.

Regards,
Xiaofan

2006\02\14@233318 by David VanHorn

picon face
A common killer in linear supplies is inrush current. It may have damaged
the diodes, though they usually fail at the moment of powerup.

Also too large a filter cap reduces the conduction angle, which increases
the peak current seen by the diode.

2006\02\14@234305 by Denny Esterline

picon face
Hmm... There might be better ways to go about this. Since you mentioned All
Electronics I poked through thier offerings quickly. Look at this one
www.allelectronics.com/cgi-bin/item/PS-511/480/222_WATT_SWITCHING_POWER_SUPPLY_.html
It's
a PC supply, but it's only $10 and it puts out 32A of 5.1V power. Seems like
a solution to me.

-Denny


On 2/14/06, Peter Todd <.....peteKILLspamspam.....petertodd.ca> wrote:
{Quote hidden}

> -

2006\02\15@020218 by William Chops Westfield

face picon face
On Feb 14, 2006, at 7:32 PM, Peter Todd wrote:

> I built a linear power supply using a 28A 6.3V transformer
> from digikey. 237-1255-ND. 4x LM396K 10A regulators in
> parallel through load-sharing resistors regulate the voltage.

What kind of filter caps have you got in there?  IIRC, you need
pretty big filter caps in big linear supplies, and that in turn
requires some sort of surge prevention to handle the turn-on
case where the cap acts as a dead short.

And what kind of heatsink for the bridge?  At 28A, it's going
to need to dissipate a goodly amount of power.  I've always
found it a bit mind boggling that one of the most heatsinked
elements of modern power supplies is the rectifier diode (more
so than the regulating element, frequently), but the math shows
it needs it...

BillW

2006\02\15@022455 by Chen Xiao Fan

face
flavicon
face


> -----Original Message-----
> From: piclist-bouncesspamspam_OUTmit.edu
> [@spam@piclist-bouncesKILLspamspammit.edu] On Behalf Of William "Chops" Westfield
>
> And what kind of heatsink for the bridge?  At 28A, it's going
> to need to dissipate a goodly amount of power.  I've always
> found it a bit mind boggling that one of the most heatsinked
> elements of modern power supplies is the rectifier diode (more
> so than the regulating element, frequently), but the math shows
> it needs it...

That is why there are some controlled active rectifier
(using Power MOSFET and IGBT) for the high power AC-DC
applications. I think IR has quite some device for this
kind of application but they may not be cheap.

Actually I suspect this high power rectifier will have
bad THD which is not allowed to be sold in Europe. But I
guess the OP does not really care about CE mark and other
regulatory requirement.

Regards,
Xiaofan

2006\02\15@081039 by Howard Winter

face
flavicon
picon face
Peter,

On Tue, 14 Feb 2006 22:32:48 -0500, Peter Todd wrote:
>...
> I built a linear power supply using a 28A
> 6.3V transformer from digikey. 237-1255-ND. 4x LM396K 10A regulators in
> parallel through load-sharing resistors regulate the voltage. For my
> first attempt I used a standard 50A 600V diode bridge, MP506W-BPMS-ND

You haven't mentioned capacitors - hopefully you have some huge electrolytics across the output of the
bridge-rectifier?

>...

> I ordered three 115CNQ015ASL-ND 15V 110A

15V doesn't sound like much headroom above the supply - the 6.3V is RMS, so you'll be getting peaks of about
9V.  Added to that, the 6.3V is what it will give when fully loaded - at lower loads the voltage will rise.  
Without detailed specs I don't know how high that would be, but it *could* be enough to exceed the diodes'
reverse voltage rating when the motors all happen to go quiet together.

Cheers,


Howard Winter
St.Albans, England


2006\02\15@085857 by Vasile Surducan

face picon face
On 2/15/06, Peter Todd <KILLspampeteKILLspamspampetertodd.ca> wrote:
> Alright, no laughing, this is a serious linear power supply, not some
> wimpy half-amp unit. Not that I'm claiming my design is any good...
>
> I've got a device I built that has 64 stepper motors with associated pic
> chip drivers in it. These are extremely cheap steppers,
> www.allelectronics.com/cgi-bin/item/SMT-57/400600/STEPPER_MOTOR_.html
> to be exact. Total current draw averages about 22A-25A but can peak as
> high as about 40A.

A typical situation when better split a huge supply in many smaller than
struggle with noise and spikes.



Not that I am totally sure about those numbers,
{Quote hidden}

6.3V ac is not enough to get 5V/25A dc stabilised with any schottky
diode ever built.

>
> My second bright idea (ha!) was to try to reduce the voltage drop by
> using those fancy schottky diodes I noticed in the catelogs... So I
> ordered three 115CNQ015ASL-ND 15V 110A and made them into a diode
> bridge. (they are two diode packages) That worked beautifully, giving me
> a usable 4.2V line, until after about 30 minutes of running I heard a
> loud bang. The main AC fuse had blown, though not the 40A DC fuse, and
> the after some testing the diodes were found to have developed internal
> shorts. All three seem to have done this at once!

three dual diode diode in a bridge, are you using tri-phase mains supply ?

{Quote hidden}

Wrong direction.

best regards,
Vasile


>
> --
> RemoveMEpeteTakeThisOuTspampetertodd.ca http://www.petertodd.ca
> -

2006\02\15@140633 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 11:51:53AM +0800, Chen Xiao Fan wrote:

> Check the specification of those Schottky diodes. Often they can not
> withstand high reverse voltage. I think this could well be the
> reason of failure for your diodes.

Good point, I noticed in the specs that there are two Vr's... One at
15V,  saying Tj = 100C and another one listed as Vrwm, Tj = 125C, at
5V...

So does that mean that Vr decreases to 5V when the junction temperature
is 125C?


Note that when the diodes failed the first thing I did was check how hot
they were, and they were only slightly above room temperature, a little
warm, but nothing more. That's after 30 minutes, so I don't think the
heat just hadn't reached the case or something.

> Do you have Varistors before the diode bridge? This might help as well.

No I don't. I'll look into those.

--
spamBeGonepetespamBeGonespampetertodd.ca http://www.petertodd.ca

2006\02\15@141009 by Peter Todd

picon face
On Tue, Feb 14, 2006 at 11:43:05PM -0500, Denny Esterline wrote:
> Hmm... There might be better ways to go about this. Since you mentioned All
> Electronics I poked through thier offerings quickly. Look at this one
> www.allelectronics.com/cgi-bin/item/PS-511/480/222_WATT_SWITCHING_POWER_SUPPLY_.html
> It's
> a PC supply, but it's only $10 and it puts out 32A of 5.1V power. Seems like
> a solution to me.

My thoughts too... I did do some experiments with PC power supplies,
with much aprehension I must add, last thing I want was an overvoltage
failure! That said I managed to toast two of them that claimed to output
30A and 40A respectivly at 5V... They both shutdown and wouldn't start
again. This actually happened in both cases within a minute.

I did properly load the other voltage lines BTW, measured a 1A draw on
the 12V and 0.1A on -12 and -5

--
TakeThisOuTpeteEraseMEspamspam_OUTpetertodd.ca http://www.petertodd.ca

2006\02\15@141418 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 03:24:52PM +0800, Chen Xiao Fan wrote:
> That is why there are some controlled active rectifier
> (using Power MOSFET and IGBT) for the high power AC-DC
> applications. I think IR has quite some device for this
> kind of application but they may not be cheap.
>
> Actually I suspect this high power rectifier will have
> bad THD which is not allowed to be sold in Europe. But I
> guess the OP does not really care about CE mark and other
> regulatory requirement.

You guess correctly. This is an art peice, made under a *very* tight
deadline. There's a lot of corners I cut getting it to work for my
deadline, and now I'm going back and fixing things properly.

So do try to forgive me if I made any boneheaded mistakes in it's
construction... Working 16 hour days for a month on it and a number of
other big projects is not kind on ones judgement!

--
RemoveMEpetespamTakeThisOuTpetertodd.ca http://www.petertodd.ca

2006\02\15@142605 by David VanHorn

picon face
>
>
> I did properly load the other voltage lines BTW, measured a 1A draw on
> the 12V and 0.1A on -12 and -5


Something to know about multi-output switchers.
There is one output that is used to do the regulation. All the others are
either unregulated, or regulated with linear regs.  A high load on the
regulated output causes extra energy to be delivered to all the outputs.
They usually make some assumptions about other loads being of some
proportion to the main load.

2006\02\15@142619 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 03:58:55PM +0200, Vasile Surducan wrote:
> Not that I am totally sure about those numbers,
> > that's just assuming that the amp meter on my testing power supply
> > works. The motors are not synchronized to each other. To be exact, they
> > play Conways Game of Life...
> >
> > The the problem is as follows. I built a linear power supply using a 28A
> > 6.3V transformer from digikey. 237-1255-ND. 4x LM396K 10A regulators in
> > parallel through load-sharing resistors regulate the voltage. For my
> > first attempt I used a standard 50A 600V diode bridge, MP506W-BPMS-ND
> > That didn't work, the final voltage at the "5V" line ended up being
> > about 3.5V, causing the master PIC controller to crash.
>
> 6.3V ac is not enough to get 5V/25A dc stabilised with any schottky
> diode ever built.

I know now... I screwed up my calculations somewhere, fortunately I'd
planned ahead for such screwups and designed the circuits to work with a
fairly wide voltage range, they are all LF series PIC chips for
instance.

{Quote hidden}

No, just single, the diodes are arranged as follows:

         AC
   |-->--|
   |-->--|
   |     |-->--|
- --|           |- +
   |     |-->--|
   |-->--|
   |-->--|
          AC
       
AC in the top and bottom, positive out the right, negative on the left.

The diodes are packaged as two diodes with common cathodes.

--
peteEraseMEspam.....petertodd.ca http://www.petertodd.ca

2006\02\15@142947 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 01:10:36PM +0000, Howard Winter wrote:
> On Tue, 14 Feb 2006 22:32:48 -0500, Peter Todd wrote:
> >...
> > I built a linear power supply using a 28A
> > 6.3V transformer from digikey. 237-1255-ND. 4x LM396K 10A regulators in
> > parallel through load-sharing resistors regulate the voltage. For my
> > first attempt I used a standard 50A 600V diode bridge, MP506W-BPMS-ND
>
> You haven't mentioned capacitors - hopefully you have some huge electrolytics across the output of the
> bridge-rectifier?

Relatively large electrolytes, one 100,000uF, 25V unit. I did check if
there was any ripple present, if I recall correctly it worked out to be
about 0.05Vpp ripple when everything was running.

> > I ordered three 115CNQ015ASL-ND 15V 110A
>
> 15V doesn't sound like much headroom above the supply - the 6.3V is RMS, so you'll be getting peaks of about
> 9V.  Added to that, the 6.3V is what it will give when fully loaded - at lower loads the voltage will rise.  
> Without detailed specs I don't know how high that would be, but it *could* be enough to exceed the diodes'
> reverse voltage rating when the motors all happen to go quiet together.

Hmm... Sounds quite plausable allright, and as I mentioned to antoher
poster, it looks liek the 15V can be much less under the right
circumstances.

--
EraseMEpetespampetertodd.ca http://www.petertodd.ca

2006\02\15@143246 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 02:25:04PM -0500, David VanHorn wrote:
> >
> >
> > I did properly load the other voltage lines BTW, measured a 1A draw on
> > the 12V and 0.1A on -12 and -5
>
>
> Something to know about multi-output switchers.
> There is one output that is used to do the regulation. All the others are
> either unregulated, or regulated with linear regs.  A high load on the
> regulated output causes extra energy to be delivered to all the outputs.
> They usually make some assumptions about other loads being of some
> proportion to the main load.

Oh! Well that is definetely good to know... I was carefully watching the
voltage on the 5V line, and it, until the switchers died, was within
spec, but I wasn't watching the other lines.

Last time I ever use PC power supplies... Yuck.


That said, is there anywhere where I can get cheap single output
switchers then? Like a 5V 50A power supply all nicely boxed up... A
place not named digikey which sells nice medical ones for about, oh,
$400 or more...

--
RemoveMEpeteEraseMEspamEraseMEpetertodd.ca http://www.petertodd.ca

2006\02\15@143413 by Peter Todd

picon face
On Tue, Feb 14, 2006 at 11:01:56PM -0800, William Chops Westfield wrote:
> On Feb 14, 2006, at 7:32 PM, Peter Todd wrote:
>
> > I built a linear power supply using a 28A 6.3V transformer
> > from digikey. 237-1255-ND. 4x LM396K 10A regulators in
> > parallel through load-sharing resistors regulate the voltage.
>
> What kind of filter caps have you got in there?  IIRC, you need
> pretty big filter caps in big linear supplies, and that in turn
> requires some sort of surge prevention to handle the turn-on
> case where the cap acts as a dead short.

As mentioned in another post, 100000uF, no surge prevention.

> And what kind of heatsink for the bridge?  At 28A, it's going
> to need to dissipate a goodly amount of power.  I've always
> found it a bit mind boggling that one of the most heatsinked
> elements of modern power supplies is the rectifier diode (more
> so than the regulating element, frequently), but the math shows
> it needs it...

I worried about that too, and very securely soldered the heatsink tabs
to my PCB. That said, I felt the diodes immediately after they fried,
and they were only a litle above room temp. Don't think that was the
cause.

--
RemoveMEpetespam_OUTspamKILLspampetertodd.ca http://www.petertodd.ca

2006\02\15@145235 by David VanHorn

picon face
You have a couple of problems here.

First there is the turn-on surge, which is going to be limited only by the
transformer's ability to deliver current.

Second, you're going to have a relatively low conduction angle because of
the large filter caps, which means that at N amps of output current, your
diode current will the several times N.

Third, diodes don't share current well. Putting them in parallel without a
ballasting resistor ends up with one diode having slightly lower Vf than the
other, and taking more current. The positive temp coefficient will tend to
help you here, otherwise it would go into runaway, with the hot diode taking
more current till it pops.

I'd reccomend a current probe on those diode leads, or some real power
supply design work.  Too much cap is a problem.

2006\02\15@150441 by olin piclist

face picon face
David VanHorn wrote:
> The positive temp
> coefficient will tend to help you here, otherwise it would go into
> runaway, with the hot diode taking more current till it pops.

Which is likely since diode voltage *decreases* with temperature.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products

2006\02\15@151535 by David VanHorn

picon face
On 2/15/06, Olin Lathrop <RemoveMEolin_piclistTakeThisOuTspamspamembedinc.com> wrote:
>
> David VanHorn wrote:
> > The positive temp
> > coefficient will tend to help you here, otherwise it would go into
> > runaway, with the hot diode taking more current till it pops.
>
> Which is likely since diode voltage *decreases* with temperature.


Whoops. I thought it was negative but then i corrected myself.
It's worse than that though, since only two of the four positions in the
bridge are running in parallel.

2006\02\15@153448 by Sergey Dryga

face picon face
Peter Todd <pete <at> petertodd.ca> writes:

> That said, is there anywhere where I can get cheap single output
> switchers then? Like a 5V 50A power supply all nicely boxed up... A
> place not named digikey which sells nice medical ones for about, oh,
> $400 or more...
>
Try Jameco,
www.jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-
1&storeId=10001&catalogId=10001&productId=202091&pa=202091PS

Several options over 40A@ 5V.

Sergey



2006\02\15@160425 by Gerhard Fiedler

picon face
Peter Todd wrote:

> Total current draw averages about 22A-25A but can peak as high as about
> 40A. Not that I am totally sure about those numbers, [...]

It's not quite clear what you mean by "peak", but it sounds as if that
could be peaks of more than a few milliseconds. If that's the case, a power
supply designed for 28 A may or may not help you.

I'd try to buy one. Other than that, you can go simple linear (with big
heat sinks), switched (which probably is not trivial for 40+ A), or linear
with several parallel supplies that get switched on or off as the current
requires. Or something else... :)


> As mentioned in another post, 100000uF, no surge prevention.

That could cause quite some inrush current. (BTW, there's "milli", too: 100
mF is easier to read :)

Gerhard

2006\02\15@162444 by David VanHorn

picon face
>
>
> > As mentioned in another post, 100000uF, no surge prevention.
>
> That could cause quite some inrush current. (BTW, there's "milli", too:
> 100
> mF is easier to read :)


0.1F even easier, and avoids uncommon usage.

2006\02\15@163417 by Herbert Graf

flavicon
face
On Wed, 2006-02-15 at 16:24 -0500, David VanHorn wrote:
> >
> >
> > > As mentioned in another post, 100000uF, no surge prevention.
> >
> > That could cause quite some inrush current. (BTW, there's "milli", too:
> > 100
> > mF is easier to read :)
>
>
> 0.1F even easier, and avoids uncommon usage.

Easier yes, but more prone to error IMHO. It's very easy to see 0.1F as
a "typo" of 0.1uF. OTOH 100mF is also prone to the "typo factor", so I
don't really see it as much better.

How about something like: 100 000uF? I thought about 100,000uF,
unfortunately in many parts of the world the ',' character means the
same as we use the '.' character! :)

TTYL

-----------------------------
Herbert's PIC Stuff:
http://repatch.dyndns.org:8383/pic_stuff/

2006\02\15@164704 by David VanHorn

picon face
> > 0.1F even easier, and avoids uncommon usage.
>
> Easier yes, but more prone to error IMHO. It's very easy to see 0.1F as
> a "typo" of 0.1uF. OTOH 100mF is also prone to the "typo factor", so I
> don't really see it as much better.


Nobody really uses millifarads though.
And you can use a common sense filter, 0.1uF as the filter cap for a 40A
supply at 120 Hz?  :)


How about something like: 100 000uF? I thought about 100,000uF,
> unfortunately in many parts of the world the ',' character means the
> same as we use the '.' character! :)


One fellow I worked with insisted on specifying all caps in picofarads to
avoid this.. So that nice memory super cap is :
4700000000000pF

2006\02\15@170257 by Herbert Graf

flavicon
face
On Wed, 2006-02-15 at 16:46 -0500, David VanHorn wrote:
> > > 0.1F even easier, and avoids uncommon usage.
> >
> > Easier yes, but more prone to error IMHO. It's very easy to see 0.1F as
> > a "typo" of 0.1uF. OTOH 100mF is also prone to the "typo factor", so I
> > don't really see it as much better.
>
>
> Nobody really uses millifarads though.

True, but then, it is metric, and I'm pretty sure most people who aren't
in north america would have zero problem with it.

In North America it MAY confuse some, but SO MUCH of electrical "stuff"
is metric I really don't believe it would be confusing to almost anybody
in the industry.

> And you can use a common sense filter, 0.1uF as the filter cap for a 40A
> supply at 120 Hz?  :)

In this case, yes. But not in all cases. Plus, that assumes the person
is LOOKING at a schematic, that is not always the case.

> How about something like: 100 000uF? I thought about 100,000uF,
> > unfortunately in many parts of the world the ',' character means the
> > same as we use the '.' character! :)
>
>
> One fellow I worked with insisted on specifying all caps in picofarads to
> avoid this.. So that nice memory super cap is :
> 4700000000000pF

Wow, that must have been... fun... :)

TTYL


-----------------------------
Herbert's PIC Stuff:
http://repatch.dyndns.org:8383/pic_stuff/

2006\02\15@171038 by David VanHorn

picon face
> >
> > One fellow I worked with insisted on specifying all caps in picofarads
> to
> > avoid this.. So that nice memory super cap is :
> > 4700000000000pF
>
> Wow, that must have been... fun... :)


I do agree with the convention of putting a 0 in front of any . so the dp
dosen't get lost, especially on faxes.  0.1uF becomes 01uf and you know what
happened to the point.

2006\02\15@174054 by Spehro Pefhany

picon face
At 05:10 PM 2/15/2006 -0500, you wrote:
> > >
> > > One fellow I worked with insisted on specifying all caps in picofarads
> > to
> > > avoid this.. So that nice memory super cap is :
> > > 4700000000000pF
> >
> > Wow, that must have been... fun... :)
>
>
>I do agree with the convention of putting a 0 in front of any . so the dp
>dosen't get lost, especially on faxes.  0.1uF becomes 01uf and you know what
>happened to the point.

What's wrong with 100nF?

We don't go around saying 0.01uA all the time, it's much easier to write 10nA.

Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
EraseMEspeffspamspamspamBeGoneinterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
->> Inexpensive test equipment & parts http://search.ebay.com/_W0QQsassZspeff


2006\02\15@180101 by Jinx

face picon face

> I do agree with the convention of putting a 0 in front of any . so
> the dp dosen't get lost, especially on faxes.  0.1uF becomes 01uf
> and you know what happened to the point

If I had a part that big (and let's face it, like great actors, we all want
big parts) I'd prefer 0F1. It might be unusual, as 100000uF is not a
common part, but it's not uncommon to see 0R1 for 0.1 ohms or
0u1 for 100nF. The nomenclature exists and has been in use for a
long time (eg 2k2 is seen around here more often than 2200, which
to me looks a bit "odd")

2006\02\15@180728 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 02:51:59PM -0500, David VanHorn wrote:
> You have a couple of problems here.
>
> First there is the turn-on surge, which is going to be limited only by the
> transformer's ability to deliver current.

Substantial for sure...

> Second, you're going to have a relatively low conduction angle because of
> the large filter caps, which means that at N amps of output current, your
> diode current will the several times N.

So will that repetively stress the diodes at each 60hz cycle? So at each
"charge" they get a fast burst of current through them?

> Third, diodes don't share current well. Putting them in parallel without a
> ballasting resistor ends up with one diode having slightly lower Vf than the
> other, and taking more current. The positive temp coefficient will tend to
> help you here, otherwise it would go into runaway, with the hot diode taking
> more current till it pops.

I don't think that's the problem. The diodes are rated, individually,
for 2x more current than is the steady state. They are in parallel not
because I need extra current capacity, but simply because it was
cheapest to buy them packaged like that.

--
RemoveMEpeteKILLspamspampetertodd.ca http://www.petertodd.ca

2006\02\15@182149 by David VanHorn

picon face
>
> So will that repetively stress the diodes at each 60hz cycle? So at each
> "charge" they get a fast burst of current through them?


YES.


> I don't think that's the problem. The diodes are rated, individually,
> for 2x more current than is the steady state.


Only 2x eh.. So if the conduction angle is less than 90 degrees, you're
there.
I bet if you take that cap way down, and accept some ripple, you'll have
your diodes living.

2006\02\15@183203 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 07:03:41PM -0200, Gerhard Fiedler wrote:
> Peter Todd wrote:
>
> > Total current draw averages about 22A-25A but can peak as high as about
> > 40A. Not that I am totally sure about those numbers, [...]
>
> It's not quite clear what you mean by "peak", but it sounds as if that
> could be peaks of more than a few milliseconds. If that's the case, a power
> supply designed for 28 A may or may not help you.

Ahh, I should be more clear. By peak, I mean that the highest amount of
current I can possibly get is 40A. This is achieved by turning all the
coils of the motors on. This is not peak in the sense of a momentary,
pulse, this is peak in terms of I can maintain a draw of that much
current under odd conditions. Not really a good use of the term peak
come to think of it.

> I'd try to buy one. Other than that, you can go simple linear (with big
> heat sinks), switched (which probably is not trivial for 40+ A), or linear
> with several parallel supplies that get switched on or off as the current
> requires. Or something else... :)

Probably the way to go. Another poster mentioned a switch-mode supply
from jameco that looks like it'd work, and comes with an impressive list
of things it does. Expensive though, I may just find a bigger, higher
voltage, transformer and go back to my first diode bridge. I just gotta
find a 8v, 50A transformer.

--
peteSTOPspamspamspam_OUTpetertodd.ca http://www.petertodd.ca

2006\02\15@184214 by Peter Todd

picon face
On Wed, Feb 15, 2006 at 06:21:49PM -0500, David VanHorn wrote:
> > I don't think that's the problem. The diodes are rated, individually,
> > for 2x more current than is the steady state.
>
>
> Only 2x eh.. So if the conduction angle is less than 90 degrees, you're
> there.
> I bet if you take that cap way down, and accept some ripple, you'll have
> your diodes living.

Hmm... So a 10000uF instead?

One tricky thing is I use the back emf from the stepper motors to act as
an encoders. The steppers are in a 8x8 grid and each stepper moves a
knob. The idea is that the knobs move on their own, and people can also
turn the knobs to interact. I use the back-emf as a cheap way of adding
in position encoders. It actually works very well, even though the
emf-pulse of someone slowly moving a knob is only about 5milivolts, but
the PIC's ADC can sense that just fine.

It does seem to be sensitive to noise in the power supply. The motors
are powered in a chains of 8 units connected by 22gauge wire. Stupid
yeah, but packaging constraints forced this. Each time one motor in a
chain starts moving, the rest sense it as back-emf. Once a motor is
moving it doesn't cause any further problems. A software fix was my
solution...

My worry is how well I can filter out that ripple. Of course... I use a
PIC+ADC to do the sensing, so I can probably fix that in software too.

--
spamBeGonepeteSTOPspamspamEraseMEpetertodd.ca http://www.petertodd.ca

2006\02\16@040151 by William Chops Westfield

face picon face

On Feb 15, 2006, at 11:48 AM, Peter Todd wrote:

> That said, is there anywhere where I can get cheap single output
> switchers then? Like a 5V 50A power supply all nicely boxed up... A
> place not named digikey which sells nice medical ones for about, oh,
> $400 or more...
>
Jameco has a 40A supply for about $73, and 50A for about $150...
(new, not surplus.)

BillW

2006\02\16@050847 by Gerhard Fiedler

picon face
Herbert Graf wrote:

>> Nobody really uses millifarads though.
>
> True, but then, it is metric, and I'm pretty sure most people who aren't
> in north america would have zero problem with it.

Exactly. It's not that common, but then such power supplies are not
commonly designed. If I'd design such power supplies, I'd use mF all the
time.

Come on, that's what we have these decimal multipliers for. Once you get
used to the system, stuff like kA, MV and mF starts to make sense :)

> In North America it MAY confuse some, but SO MUCH of electrical "stuff"
> is metric I really don't believe it would be confusing to almost anybody
> in the industry.

And believe me, the USA /is/ going metric. The sooner it gets there, the
easier it will be. It's just a matter of time, and the longer that twilight
zone lasts, the worse.

Gerhard

2006\02\16@051750 by Gerhard Fiedler
picon face
Jinx wrote:

> If I had a part that big (and let's face it, like great actors, we all
> want big parts) I'd prefer 0F1. It might be unusual, as 100000uF is not
> a common part, but it's not uncommon to see 0R1 for 0.1 ohms or 0u1 for
> 100nF. The nomenclature exists and has been in use for a long time (eg
> 2k2 is seen around here more often than 2200, which to me looks a bit
> "odd")

But this nomenclature is not often used with a 0 as first digit (at least
not outside the US, where it's quite common with caps). You don't see 0M22
resistors, you see 220k resistors. (BTW, not always a good idea to use only
the decimal multiplier... :)

In that sense, the 0R1 resistor would be a 100mR resistor, and the 0F1 cap
a 100mF cap.

Gerhard

2006\02\16@052041 by Gerhard Fiedler

picon face
Peter Todd wrote:

> Ahh, I should be more clear. By peak, I mean that the highest amount of
> current I can possibly get is 40A. This is achieved by turning all the
> coils of the motors on. This is not peak in the sense of a momentary,
> pulse, this is peak in terms of I can maintain a draw of that much
> current under odd conditions. Not really a good use of the term peak
> come to think of it.

Well, it /is/ peak, so that wasn't wrong. It's just that "peak" needs also
(at least) the spec of how long it lasts.

{Quote hidden}

And do something about the inrush current.

Gerhard

2006\02\16@053709 by Alan B. Pearce

face picon face
>In that sense, the 0R1 resistor would be a 100mR resistor,

I think the 0R1 is the preferable notation here, as it really emphasises
that it is LOW value.

>and the 0F1 cap a 100mF cap.

However these are at the HIGH value end of the spectrum, so to my way of
thinking using 100mF would be better. This may be part of the reason why a
220k is written that way rather than 0M22.

2006\02\16@055512 by Jinx

face picon face

> In that sense, the 0R1 resistor would be a 100mR resistor, and
> the 0F1 cap a 100mF cap

One problem with caps is that some people mistakenly use "m"
for micro. Similarly mHz instead of MHz. You should be able to
tell from the context, if you have the context available that is. It's
up to the author to be unambiguous. My preference is still to use
the multiplier instead of the DP

2006\02\16@145444 by Peter

picon face


On Wed, 15 Feb 2006, David VanHorn wrote:

{Quote hidden}

Or put a 40A choke after the bridge and some recovery diodes.

Peter

2006\02\16@145827 by Peter

picon face

On Thu, 16 Feb 2006, Gerhard Fiedler wrote:

> And believe me, the USA /is/ going metric. The sooner it gets there, the

Naw, you're kidding.

Peter

2006\02\16@153452 by David VanHorn

picon face
>
> > Only 2x eh.. So if the conduction angle is less than 90 degrees, you're
> > there.
> > I bet if you take that cap way down, and accept some ripple, you'll have
> > your diodes living.
>
> Or put a 40A choke after the bridge and some recovery diodes.


A pi-filter would be a lot better design here, than just massive caps and
looking for perfect diodes.

2006\02\16@173144 by Sergey Dryga

face picon face
Gerhard Fiedler <lists <at> connectionbrazil.com> writes:

> And believe me, the USA /is/ going metric. The sooner it gets there, the
> easier it will be. It's just a matter of time, and the longer that twilight
> zone lasts, the worse.

Totally agree, but even in England, which went metric while ago, one buys beer
by pint, not by liter (or I should say litre?).

An example of the USA going to metric, from my recent experience of purchasing
replacement glass for a window:
me: Do you have replacement glass?
store: Yes we do.  What thickness do you need?
me: 3 mm
store: Huh? Can you say it again please?
<long conversation when I was trying to explain what I need>
me: 1/8 of an inch.
store: Sure, we have that.  You can pick it up any time.

This just shows how soon US will be metric. BTW, remember Mars Lander while
ago? It crashed because commands where sent using centimeters, but lander was
thinking inches.  If NASA gets confused, is there any hope for "regular" folks?

Sergey,
learned to convert C<->F, in<->cm and L<->gal in my head.





2006\02\16@201318 by Gerhard Fiedler

picon face
Sergey Dryga wrote:

>> And believe me, the USA /is/ going metric.

> An example of the USA going to metric, from my recent experience of purchasing
> replacement glass for a window:
> me: Do you have replacement glass?
> store: Yes we do.  What thickness do you need?
> me: 3 mm
> store: Huh? Can you say it again please?

:)  I know about these, too. They will only start to sell glass measured in
mm when no imperial glass is available on the market anymore. But it seems
to me that I see a slowly growing amount of mechanical specs in metric or
metric and imperial units.

Additionally, the rest of the world is (almost) metric. (Notwithstanding
the pints in England :)  Science in the USA is metric. The automotive
industry in the USA is metric. The rest will follow, sooner or later.

> If NASA gets confused, is there any hope for "regular" folks?

That's exactly what I'm talking about -- there's no way this dual system
can work well.

> learned to convert C<->F, in<->cm and L<->gal in my head.

I went through this also... :)

Gerhard

2006\02\16@204326 by Bob Axtell

face picon face
Gerhard Fiedler wrote:

>Sergey Dryga wrote:
>
>  
>
>>>And believe me, the USA /is/ going metric.
>>>      
>>>
>
>  
>
but... I really miss "furlongs per fortnight"...

--Bob

--
Note: To protect our network,
attachments must be sent to
KILLspamattachspamBeGonespamengineer.cotse.net .
1-520-850-1673 USA/Canada
http://beam.to/azengineer

2006\02\16@205533 by Peter Todd

picon face
On Thu, Feb 16, 2006 at 01:01:50AM -0800, William Chops Westfield wrote:
>
> On Feb 15, 2006, at 11:48 AM, Peter Todd wrote:
>
> > That said, is there anywhere where I can get cheap single output
> > switchers then? Like a 5V 50A power supply all nicely boxed up... A
> > place not named digikey which sells nice medical ones for about, oh,
> > $400 or more...
> >
> Jameco has a 40A supply for about $73, and 50A for about $150...
> (new, not surplus.)

To both you and Sergy:

Thank you! That's exactly what I should have done in the first place,
and would of, had I known... Oh well.

I'll place an order in the next day or two, meanwhile, anything I should
know about these supplies? I assume with my application I should be able
to simply hook them up and be done with it. Looks like they all have
overcurrent protection, so no fuses needed. I don't see any mention of
filter caps or such in the datasheet, so I assume they are relatively
foolproof other than making sure they can still get airflow.

--
EraseMEpetespamEraseMEpetertodd.ca http://www.petertodd.ca

2006\02\17@001747 by Martin K

flavicon
face
Peter Todd wrote:

>Probably the way to go. Another poster mentioned a switch-mode supply
>from jameco that looks like it'd work, and comes with an impressive list
>of things it does. Expensive though, I may just find a bigger, higher
>voltage, transformer and go back to my first diode bridge. I just gotta
>find a 8v, 50A transformer.
>
>  
>
I've seen a 240v to 400A 5v transformer. I'm not joking.
At lower voltages you could almost take a large toroid and wrap 10AWG
wire around it to get the voltage you need.
Another option would be a microwave transformer. The primary and
secondary are usually separated so you could cut off the HV secondary
and wind on a strip of copper foil for the secondary. This is assuming
you want something less expensive. It seems like you have a mid-range
budget for this project.
I think you could get a surplus switcher to work. If not one unit then
two in parallel through some small value power resistors to share the
current.
I'd be tempted to build my own push-pull switcher with PFC, but I'm that
kind of guy.

--
Martin K

2006\02\17@025548 by William Chops Westfield

face picon face

On Feb 15, 2006, at 2:02 PM, Herbert Graf wrote:

>> Nobody really uses millifarads though.
>
> True, but then, it is metric, and I'm pretty sure most people who
> aren't in north america would have zero problem with it.
>
except that few typewriters have proper greek "mu" characters,
so mF has been used to mean "microfarads" at various times (and
mmF for picofarads...), and even if you're fully "metricized",
I'd have to look several times at the in-circuit usage to decide
if mF really meant milli or not. (becides, this was 0.1F, so
why not 1 dF or 10 cF :-)

BillW

2006\02\17@041441 by Alan B. Pearce

face picon face
>BTW, remember Mars Lander while ago? It crashed because
>commands where sent using centimeters, but lander was
>thinking inches.

I was working on a space project that involved American participation at
that time.

>If NASA gets confused, is there any hope for "regular" folks?

You should have seen the resultant double checks that went into all programs
after that. Everything was gone over with a fine tooth comb. The
embarrassment it caused was "out of this world" ...

2006\02\17@041652 by Alan B. Pearce

face picon face
>>>And believe me, the USA /is/ going metric.

>but... I really miss "furlongs per fortnight"...

I still remember the article in an HP journal describing the area of a
microprocessor chip as so many nanoacres ...

2006\02\17@143204 by Sergey Dryga

face picon face
Peter Todd <pete <at> petertodd.ca> writes:

>
> On Thu, Feb 16, 2006 at 01:01:50AM -0800, William Chops Westfield wrote:
> >
> > On Feb 15, 2006, at 11:48 AM, Peter Todd wrote:
> >
> > > That said, is there anywhere where I can get cheap single output
> > > switchers then? Like a 5V 50A power supply all nicely boxed up... A
> > > place not named digikey which sells nice medical ones for about, oh,
> > > $400 or more...
> > >
> > Jameco has a 40A supply for about $73, and 50A for about $150...
> > (new, not surplus.)
>
> To both you and Sergy:
>
> Thank you! That's exactly what I should have done in the first place,
> and would of, had I known... Oh well.

You are welcome.

>
> I'll place an order in the next day or two, meanwhile, anything I should
> know about these supplies? I assume with my application I should be able
> to simply hook them up and be done with it. Looks like they all have
> overcurrent protection, so no fuses needed. I don't see any mention of
> filter caps or such in the datasheet, so I assume they are relatively
> foolproof other than making sure they can still get airflow.
>

If I remember correctly, in current configuartion PIC voltage is supplied form
the same power bus as motors.  This might cause glitches and random "reset" of
the PIC.  If you see some random problem, consider making a separate regulator
just for PIC.

Sergey



2006\02\17@153115 by Gerhard Fiedler

picon face
Sergey Dryga wrote:

> If I remember correctly, in current configuartion PIC voltage is supplied form
> the same power bus as motors.  This might cause glitches and random "reset" of
> the PIC.  If you see some random problem, consider making a separate regulator
> just for PIC.

This is probably a good idea in any case, and if it's just a resistor (or
inductor) and a cap.

Gerhard

2006\02\18@214100 by Peter Todd

picon face
On Fri, Feb 17, 2006 at 12:17:39AM -0500, Martin K wrote:
> Peter Todd wrote:
>
> >Probably the way to go. Another poster mentioned a switch-mode supply
> >from jameco that looks like it'd work, and comes with an impressive list
> >of things it does. Expensive though, I may just find a bigger, higher
> >voltage, transformer and go back to my first diode bridge. I just gotta
> >find a 8v, 50A transformer.
> >
> >  
> >
> I've seen a 240v to 400A 5v transformer. I'm not joking.

Funny how you say that... I did a websearch to find a bigger transformer
and figure out pricing... Sure enough, I found a 220V to 600A 5V
transformer at a surplus place!

Found a 12V 60A transformer too, but I think I'll just go with a
switcher, same price.

> At lower voltages you could almost take a large toroid and wrap 10AWG
> wire around it to get the voltage you need.

Good idea, I've got a source for cheap, $10 each, 250, 500 and 1000VA
toroids. Real heavy monsters.

> Another option would be a microwave transformer. The primary and
> secondary are usually separated so you could cut off the HV secondary
> and wind on a strip of copper foil for the secondary. This is assuming
> you want something less expensive. It seems like you have a mid-range
> budget for this project.

Yeah, it's cost something like $1500 already, and I'm looking to sell it
too, so I don't really wanna be messing around with making my own
transformers, making my own power supply was pushing it.

UL listing and all...

> I think you could get a surplus switcher to work. If not one unit then
> two in parallel through some small value power resistors to share the
> current.
> I'd be tempted to build my own push-pull switcher with PFC, but I'm that
> kind of guy.

Haha, that'd be a fun project... Especially for me... I've long
forgotten my calculus, and currently know very little analog
electronics.

Anyway for this project I think I'll just buy a switcher from jameco
like some other posters suggested, but thans for the transformer tips, I
do have projects where that'd really come in handy, projects that I'm
not planning to sell to some poor unsuspecting art collector!

--
@spam@pete@spam@spamspam_OUTpetertodd.ca http://www.petertodd.ca

2006\02\18@230807 by Peter Todd

picon face
On Fri, Feb 17, 2006 at 07:31:09PM +0000, Sergey Dryga wrote:
> > I'll place an order in the next day or two, meanwhile, anything I should
> > know about these supplies? I assume with my application I should be able
> > to simply hook them up and be done with it. Looks like they all have
> > overcurrent protection, so no fuses needed. I don't see any mention of
> > filter caps or such in the datasheet, so I assume they are relatively
> > foolproof other than making sure they can still get airflow.
> >
>
> If I remember correctly, in current configuartion PIC voltage is supplied form
> the same power bus as motors.  This might cause glitches and random "reset" of
> the PIC.  If you see some random problem, consider making a separate regulator
> just for PIC.

The motors are pretty small ones, and in previous testing things seem to
be ok. That said, I did find out that a swift turn of the knobs they are
attached too generates enough back-EMF to reset the individual PIC on
the single motor control board. Unfortunately everything is connected to
an I2C bus, so this invaribly crashes the whole bus. Interestingly the
watchdog timer on the PIC fails too.

Note that I did connect the the motor leads directly to the PIC ADC
inputs, so there is a pretty direct route to the PIC. I have a ULN2000
driving the motor, and I thought the inherent protection diodes in it
would help... It worked on the breadboard, but the full circuit on a PCB
didn't. Oh well, my current plan is to turn the whole shebang of motors
on and off via FETs in series with v+ whenever the bus gets stuck.

The other bit of weirdness is related to my bootloader setup... If have
the master controller reflash the slaves on startup everything works
fine. If I tell it to skip the reflash, and rely on what's already
flashed, one or two out of the 64 slaves will fail to recieve any
commands from I2C. Very odd, and which slaves fail to work changes on
each reflash.  I've got a lot of debugging to figure out on that one...

--
spamBeGonepetespamKILLspampetertodd.ca http://www.petertodd.ca

2006\02\23@194302 by Paul Duffy

picon face
I provided electronic controls for my daughter's art project a while back -
it had 100 solenoids, powered at 5 volts.  If all energized at once, it
required 80A of current.  I used a supply from Jameco.  Part #196795
http://www.jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-1&storeId=10001&catalogId=10001&productId=196795

It costs $12.  I used 2 of them in my application.  I was able to split the
load into 2 banks of 50 coils and then powered each bank with one of the
supplies.  It worked great!

Paul Duffy


{Original Message removed}

2006\02\23@195605 by William Chops Westfield

face picon face

On Feb 23, 2006, at 4:43 PM, Paul Duffy wrote:

>> [many stepper motors.]
>
> I provided electronic controls for my daughter's art project a while
> back -  it had 100 solenoids, powered at 5 volts.

How about some pictures of all these art projects.  They sound
fascinating...

BillW

2006\02\25@200626 by Peter Todd

picon face
On Thu, Feb 23, 2006 at 07:43:27PM -0500, Paul Duffy wrote:
> I provided electronic controls for my daughter's art project a while back -
> it had 100 solenoids, powered at 5 volts.  If all energized at once, it
> required 80A of current.  I used a supply from Jameco.  Part #196795
> www.jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-1&storeId=10001&catalogId=10001&productId=196795
>
> It costs $12.  I used 2 of them in my application.  I was able to split the
> load into 2 banks of 50 coils and then powered each bank with one of the
> supplies.  It worked great!

Thanks for the datapoint! Those do look like good supplies, and that's a
very nice price.

I'm seconding the request for pictures, I can think of a lot of fun
things you could do with 100 solenoids... Where does your daughter go to
school?

--
.....petespam_OUTspampetertodd.ca http://www.petertodd.ca

More... (looser matching)
- Last day of these posts
- In 2006 , 2007 only
- Today
- New search...