Sciencemadness Discussion Board

PC PSU to laboratory PSU

 Pages:  1  2  

Rosco Bodine - 16-1-2008 at 06:59

The latest variant which has been examined for its possible usefulness as a parallelable , high current capable Mosfet pass element output module is attached .

This is *not* a complete system but merely one 25 Amp output module "block" , showing the architecture needed for the control circuitry , which IMO has distilled from other
arrangements which have been examined as possible ways
of driving a Mosfet as a linear device , and keeping it stable .

The control loop is a highly damped three stage active low pass filter which has something similar to a PID control signal processing inherent to the drive signal applied to the power Mosfet . This is the one scheme which seems most likely to
achieve a stable control loop with a minimum of phase shift
which did appear much more potentially problematic when using simpler control loop schemes . Many simpler drive schemes were considered and some may be workable with
much higher resistance current sensing shunts , where the
voltage drop may be tolerable there . But for sensing shunts
of a milliohm or less , as is desirable on low voltage platforms
where every tenth volt counts .....things get a lot more complicated given the amplifications involved for the sensing to control loop , which leads to something like the attached schematic as a probable solution .

BTW I have the asc file for LT Spice from which the attached schematic was printed and will link it here if anyone wants to look at the spice sim displays for the performance of the attached circuit .

The values and components shown are not necessarily finalized or optimized , but this is simply meant to be illustrative using some components which should work .

[Edited on 16-1-2008 by Rosco Bodine]

Attachment: LTC1152 25 Amp VCCS modeled stable.pdf (19kB)
This file has been downloaded 991 times

12AX7 - 16-1-2008 at 10:50

So why does it have two ICs and numerous resistors, when mine has less? And "per module" he says!

Tim

Rosco Bodine - 16-1-2008 at 11:40

There's nothing really wrong with the one op amp configuration if you want to use a lot of them , maybe a half dozen or eight of them , ( maybe ten or twelve :P ) instead of just two . But you will have to spread out the task to a larger array using higher resistance shunts to maintain stability . And you will have to work out some scheme there for limiting the frequency response or the loop will "ring out" past the frequency where it goes into self-reenforced feedback induced oscillations . Anyway that's what the sims indicate . I "fought the good fight" trying to make that simpler configuration behave , but it has too much high frequency gain for the loop , the response doesn't roll off steeply enough and early enough for stability with low resistance shunts . And trying various schemes of using just an integrator or simple filtering didn't tame the loop response
either . Seeing that what was needed most for damping was some sort of increasing gain error amplifier signal dependant upon the slew rate of the signal from the sensing shunt , and an assymmetrical amplification/attenuation output from the error amplifier
(slower on the accellerator / faster on the brakes )
I concluded that a stacked pair of op amps in lead lag with their outputs ganged to provide a derivative , and then
using the Mosfet driver amp as the integrator , would
produce the desired result .

The stacked op amps which are the shunt voltage monitor and amplification stage function as a lead-lag network
and a low pass filter , whose effective gain is frequency variable . The frequency variable gain is also asymmetrical . What happens is that if a constant current
load is producing a constant shunt voltage reference , the
stacked op amps respond with an effective DC gain that is fixed . But that effective gain changes , increases in response to a changing shunt voltage , not just proportionately but exponentially , which provides the
desired damping and low pass filtering . And because
the response is steeper in attenuation than in amplification , that makes the circuit settle quickly in
response to any transients ......a full power pulse spike
settles to DC with under 3 ms of ringing at the full 25A
and maybe 5 or 6% overshoot . That's a worst case scenario response .

IIRC the total phase shift at rolloff to 0 dB is something around 100 degrees which is grrrrreat :D

[Edited on 16-1-2008 by Rosco Bodine]

bio2 - 16-1-2008 at 14:03

........ you can get 7 volts from the +5 V and +12 V outputs -- the +5 V is considered the negative (GND) and +12 the positive.....

I found this in an article and tried it and indeed it worked on my ATX 2.

This 480W power supply delivers 36A@5V and 16A@12V.

When connected as above the voltage drooped from 6.96V to 6.30V with a load of only 1.6A. Measurement was at the PS not the end of the load wire.No higher load was tried for fear of damaging my brand new PS.

My question is what would be the current output when connected in this arrangement? Also how would the current limiting be affected?

7V suits my cell better than 12V if I could get a high current output but seems doubtful given the drop at such a low current.

Also I am wondering if the 3.3V and 5V outputs could be used in a similar fashion?

bio2 - 17-1-2008 at 09:46

A quick up-date for anyone that might try this.

I applied a 3.6 ohm load to the 7V configuration and the unit tripped on
overcurrent (I assume). No damage ensued.

So it seems this hook-up is only suitable for "pilot" loads.

-jeffB - 17-1-2008 at 11:07

If you simply hook up a load between the +12V and +5V outputs, there's no way it's going to work. The PS is designed to source current from the positive outputs, not sink it. I imagine your PS tripped on overvoltage on the +5V supply.

The only way you could get regulated 7V power out of the configuration you describe is if you put a fixed load between 5V and ground that drew at least as much current as the load you're putting between 12V and 5V. In other words, if you want to draw 10A at 7V, you'd need to put a .5-ohm load across the 5V supply first, drawing a constant 10A. Then, when you add your .7-ohm load between 12V and 5V, the current through both loads would be 10A, but the current out of the 5V output would be zero. I can't make any assertions about the stability of an arrangement like this, though.

(Actually, you'd need to draw even more than that, because IIRC most PC-type switching power supplies require a minimum load on the 5V line of something like 10% of capacity. So you'd need to draw at least 13.6A through your dummy 5V load.)

Disclaimer: it's been a while since I've worked with power-supply designs, and I never quite got a 100% intuitive grasp of switching-regulator function even when I was working with them.

bio2 - 17-1-2008 at 12:08

......If you simply hook up a load between the +12V and +5V outputs, there's no way it's going to work. .........

Good thing you put in your disclaimer, lol. Read my post again!

The 1.6 amp load was applied for a couple hours with the droop as stated.
I did not imagine this or make it up so maybe think a little before stating absolutes.

The quoted statement was taken from a website detailing CPS units modifications for other uses. So I thought it's worth a try. I asked for input from our local electronics experts in this thread because I could think of no immediate explanation.

BTW There is a continuous 1 amp load applied between 5V and ground for minimum load. A 5 to 10 watt load is adequate. The dummy load can be applied to any winding, IIRC.

Funny, these last couple of days I keep being told that what I have experienced first
hand is impossible. I saw tungsten melt with my own eyes yet was told that my
flame temperature couldn't achieve this.

Guess there is no escaping the know- it- alls!

If I had a schematic of this PS then I could probably explain it, though it seems the
7V comes from subtracting 5V from 12V with the large drop being through the windings in series.

-jeffB - 17-1-2008 at 13:02

Quote:
Originally posted by bio2
Good thing you put in your disclaimer, lol. Read my post again!


I've read it several times, along with this one. Keep filling in more details, and maybe we can figure out what's going on. :)
Quote:
The 1.6 amp load was applied for a couple hours with the droop as stated.
I did not imagine this or make it up so maybe think a little before stating absolutes.


My apologies if I came off as arrogant or dismissive. If I prefaced every statement with a thorough qualification describing all my uncertainties, my posts would get too long for anyone to wade through. (There are probably some who say this has already happened.)

A "droop" from 6.97V to 6.30V corresponds to a 10-15% rise on the 5V line (from, say, 5.03V to 5.7V, assuming the 12V line stays stable). This is WAY outside tolerance, and indicates that you're driving the PS well outside its normal operating regime. It's also not surprising, given that you're apparently trying to sink a net 0.6A into a supply line that's supposed to source tens of amps.
Quote:
The quoted statement was taken from a website detailing CPS units modifications for other uses.


Link, please? I'd just as soon knock out as many of my own misconceptions as possible on my own time, rather than airing them here one by one.
Quote:
So I thought it's worth a try. I asked for input from our local electronics experts in this thread because I could think of no immediate explanation.

BTW There is a continuous 1 amp load applied between 5V and ground for minimum load. A 5 to 10 watt load is adequate. The dummy load can be applied to any winding, IIRC.


That emphatically was NOT the case for supplies I was looking at -- IIRC, I was looking at the Jameco catalog, back around fifteen years ago. At that time, again only as best I remember, most of the multi-output switchers specified a minimum load on the +5V line to maintain regulation.
Quote:
Funny, these last couple of days I keep being told that what I have experienced first
hand is impossible. I saw tungsten melt with my own eyes yet was told that my
flame temperature couldn't achieve this.


What you think you see isn't always what's really there. And sometimes, what you're looking at isn't really what you should be watching. For example, in the application you're describing, you mentioned only the voltage you observed across your load. Unless your load is known to be very stable -- and I didn't see any description of what the load was -- you might also want to monitor the current through it, to see whether you're getting what you expect. Most importantly, though, you should monitor the voltages of the 12V and 5V supplies with respect to the PS ground line, because that's what the PS is trying to regulate. I think you may find that the 5V line is going even further out of regulation than my simple-minded arithmetic above indicates.
Quote:
Guess there is no escaping the know- it- alls!

If I had a schematic of this PS then I could probably explain it, though it seems the
7V comes from subtracting 5V from 12V with the large drop being through the windings in series.


If you were dealing with a simple multi-tap transformer, maybe. But I think if you had a schematic of the PS you'd find that things are a lot more complicated than this.

bio2 - 17-1-2008 at 13:46

Here is the link and the complete sentence which is found about 3/4 of the way down the page just above the heading NEW.

http://web2.murraystate.edu/andy.batts/ps/powersupply.htm

.............As an aside, you can get 7 volts from the +5 V and +12 V outputs -- the +5 V is considered the negative (GND) and +12 the positive -- some geeks will use this combination to run their fans at a lower speed to reduce noise............

So it seems that this is an unregulated derived output intended for very small loads.

............If you were dealing with a simple multi-tap transformer, maybe. But I think if you had a schematic of the PS you'd find that things are a lot more complicated than this..............

Yea, wishful thinking on my part as I expected as much. I've built a couple switchers
but am more at home with 500KVA UPS and Static switches. The ones with the SCR's
that are really the size of bricks. Not the "bricks" in the vernacular that are 100 amp.
The biggest one we ever built was 2K amp 480V/3Phase, special order custom unit.

The load I used is my little test, water electrolysis, cell which is very stable given
a stable temperature which needless to say will gradually rise too equilibrium with ambient.

Interestingly with a 4.78V input the cell functions (4 cells series) at 8amps+/-
(11% NaOH electrolyte) which is right at the theoretical minimum voltage. This measured with an el cheapo chinese meter so it's not 1/4% accuracy, lol.

My venerable Fluke 87 won't boot up these last few days and I haven't got around to taking it apart yet. I'm hoping it's the rotary switch because it's been giving trouble for some time in the same way.

What happens is when the full display test is initiated by turning the rotary switch on it then slowly blinks the full display on and off along with the backlight accompanied by the relay clicking in about a one second interval. This is as far as
it gets, refusing to go to the proper input regardless of the rotary switch setting.
The previous symptoms for a few months prior were the same except after 2 or 3
tries returning the switch to off it would finally work normally.

Anybody have a clue what could cause this? Previous examination of the switch mechanism revealed no obvious defects like corrosion or dirt etc. in the sliding contacts.

-jeffB - 17-1-2008 at 16:07

Quote:
Originally posted by bio2
Here is the link and the complete sentence which is found about 3/4 of the way down the page just above the heading NEW.

http://web2.murraystate.edu/andy.batts/ps/powersupply.htm

Good link, thanks!
Quote:
.............As an aside, you can get 7 volts from the +5 V and +12 V outputs -- the +5 V is considered the negative (GND) and +12 the positive -- some geeks will use this combination to run their fans at a lower speed to reduce noise............

So it seems that this is an unregulated derived output intended for very small loads.


I think the more important point is that the "geeks" using this combination to run their fans at a lower speed are probably also powering the rest of their PC with the PS, which means there's always more than enough load on the 5V line to accept the extra current being shunted in from the 12V-5V load. It's still "regulated", in the sense that it won't fluctuate any more than the difference in the regulated 12V and 5V lines, but you've got to make sure that you're drawing enough current OUT of the 5V line to maintain regulation.

I'm also vaguely remembering that many multi-output switching PSes, especially cheaper ones, base their other voltages off the 5V reference instead of setting each independently. If I'm remembering correctly, this means anything that causes the 5V supply baseline voltage to rise or fall will also cause the other voltages to rise or fall. I don't know if this is still true.

bio2 - 17-1-2008 at 16:28

An interesting tidbit on my cheapo chinese 480W PS (but everything seems to be made in China these days). They used to burn these things in for a few hours in the days of quality electronics.

Anyway, I had adjusted the electrolyte concentration for 9 amps at 5V and was doing
some fiddling around taking measurements. I turned the unit back on when much to my surprise it was drawing 27 amps. Wow I thought, what is causing this, must be a short in the cell somewhere.

So a couple seconds later smoke started pouring out thru the fan so I quickly switched it off only to discover I had hooked it up to the 12V wire (rated 16A) Damn, so much for integral current limiting and temperature protection.

I took it apart and found no melted components only some discoloration and warping of the insulating sheet under the board. Much to my amazement when I tried it again
the little POS still worked.

It had previously tripped a couple times at a little over 16A but 27A and it kept running, go figure. Yea, yea, I know should have had a fuse in line but the best I could do without a trip to the store was 10A which was too small.

Wonder how many years were taken from it's miserable life?

12AX7 - 17-1-2008 at 16:36

The current limiting is usually done with a current transformer at the output transistors. So, you were drawing about 12V * 16A = 192W, which is well under the 480W limit (probably around 6A peak through the transistors, for what it's worth), but well over the rating of the 12V rectifier and filter circuit. Lucky for you, the rectifier did not fail (or at least fail shorted).

They put dinky ass wire in the coils in those power supplies. Try drawing a full load at 5V some time....then don't try it again. :rolleyes:

Tim

P.S. We still don't have a :rolleyes: smiley! :rolleyes:

dann2 - 17-1-2008 at 16:42

Hello,

Pity the '7v' outputs do not work. I had always presumed they would.
Some Links:

http://www.webx.dk/oz2cpu/radios/psu-pc1.htm
http://www.qrp4u.de/index_en.html

@ 12AX7.Will the constant current moule of the circuit that you give a page or two back work on its own.
If you put a simply voltage divider (a pot) at the input (In) will the CCM work from as a simple circuit for constant current out from a PC power supply.
What MOSFET would you recomment for output MOSFET.

Cheers,

Dann2
Dann2

Rosco Bodine - 17-1-2008 at 18:06

Quote:
Originally posted by 12AX7
P.S. We still don't have a :rolleyes: smiley! :rolleyes:


Hey are you a whining and complaining....
argumentative person ???? It's soooooo
not politically correct
to use evil symbology having connotations of sarcasm ......
like the roll eyes icon !

Didn't they beat that attitude problem out of you during
conditioning and retraining ??? Are you reverting to
your former condition as a cretan and miscreant having
little antisocial toad like attitude problems in need of adjustment !!!! :P;)

You even dare to mention the censored and banned
infamous roll eyes icon and think you can summon it
at will ???

This symbol has been reserved for use by only "specially priviliged and esteemed" members :P

To your having the idea ......
that you would even imagine yourself worthy ,
I can only say


:D;):D;):D;):D:D:D:D:D:D:D

http://uk.youtube.com/watch?v=UW32D_Y4Bhg

Now just watch the shiny object , swinging back and forth ,
and listen to the music , .........back and forth , your eyes are getting heavy , you have only thoughts of peace and love ,
there is no roll eyes icon , there is no roll eyes icon ......
there is no roll eyes icon .....you feel well and your thoughts are free .....you feel well and your thoughts are free :D

[Edited on 17-1-2008 by Rosco Bodine]

12AX7 - 17-1-2008 at 19:42

Quote:
Originally posted by dann2
@ 12AX7.Will the constant current moule of the circuit that you give a page or two back work on its own.
If you put a simply voltage divider (a pot) at the input (In) will the CCM work from as a simple circuit for constant current out from a PC power supply.
What MOSFET would you recomment for output MOSFET.


For computer PSU scales, a few IRFZ44 or so (commonly found in automotive apps, e.g., car amps, 12-120V inverters), or something in an IRFP, like an IRFP064 or so (same idea, but bigger package (TO-247) --> more dissipation (about 100W, vs. 50W for a TO-220 like the IRFZ44, et al) in a single unit). Rosco seems to be fond of the IRF3703, which is along the same lines.

It sure will work alone. The input voltage (above common) sets current according to the relation shown.

Tim

Rosco Bodine - 17-1-2008 at 21:39

Textbook circuit , top left corner of page .

Look at the shunt resistance value .

It could have something to do with how well
this circuit works .....or doesn't work .

Attachment: Pages from AN106 Op Amp circuit collection.pdf (51kB)
This file has been downloaded 766 times

and here's another one

Rosco Bodine - 17-1-2008 at 22:05

Bottom right on the page

Attachment: Pages from AN105 Current Sense Circuit Collection.pdf (89kB)
This file has been downloaded 1786 times

bio2 - 18-1-2008 at 08:47

I still not sure I understand why the 12V X 27A = 324W on my 480W PSU didn't trip on overcurrent. Were you implying that the drop thru the too small CT coil wires is the reason.? The toroids didn't have burnt windings and the source of the copious smoke
, strangely, could not be determined.

As I mentioned it had previously tripped OK at slightly over it's 16A rating.

Thanks for the 5V at full load advice as I was intending to pull the full 36A from it.
Probably would have set the load and walked away as the coil wires enamel smoked off, lol.

Don't you love this cheap shit? What ever happened to name plate ratings being meaningful?

Edit; OK I see now the effect of the 1 ohm resistor value.

But, shouldn't the circuit be capable of protecting itself from
a dead short condition or is this a separate circuit using a
crowbar type arrangement?

Also I would have expected the AC input fuse to have blown
from a near double rated load applied for a few seconds.
It didn't.

[Edited on by bio2]

-jeffB - 18-1-2008 at 10:25

Quote:
Originally posted by bio2
I still not sure I understand why the 12V X 27A = 324W on my 480W PSU didn't trip on overcurrent. Were you implying that the drop thru the too small CT coil wires is the reason.? The toroids didn't have burnt windings and the source of the copious smoke
, strangely, could not be determined.


Again, I am not an EE, I do not have a schematic, etc, etc. But if a cheap PSU bases its regulation on the 5V line, I wonder if it might also base its overcurrent cutoff on that line alone? It would give me the shivers to think so, but I can also imagine that you could shave quite a bit off the cost that way.

The most "copious smoke" I've seen come off any type of electronic equipment has come from blown electrolytic capacitors. You can easily blow them with overvoltage, but if you hit them with too much ripple current, I'd think you could overheat them as well. If that's what happened, you might well be left with an open-circuit capacitor and an unfiltered 12V line. A DC meter would still show it producing the expected voltage and current, but it would have a nasty AC component superimposed. Any chance you could put a scope on it while it's "still working" with a significant load?

12AX7 - 18-1-2008 at 10:31

Circuits I've seen add the voltages together and regulate them jointly.

What I mean by the CT is, all the power goes through the output transistors and CT. How it's split up after the transformer doesn't matter, be it 12V at 50A, 5V at 100A or the correct loading on each output.

Nameplate ratings are just that, ratings; the nameplate didn't specify what, if any, the current limits are!

The coils may not be burnt, but they may be a bit "carmelized". You probably saw volatiles from the varnish on the coils, or maybe the goop sometimes applied around them.

Capacitors are easy to check, they split open on top.

Tim

[Edited on 1-18-2008 by 12AX7]

Xenoid - 18-1-2008 at 13:15

Quote:
Originally posted by -jeffB
Again, I am not an EE, I do not have a schematic, etc, etc.


@ -jeffB

There is a circuit for an AT supply near the bottom of page 3 of this thread. There is a link to a circuit for an ATX supply about halfway down page 4.

bio2 - 18-1-2008 at 14:06

........Nameplate ratings are just that, ratings; the nameplate didn't specify what, if any, the current limits are!.............

OK, I hear you but in the electrical world I came from the nameplate rating was
the maximum allowable continuous input or output.

This, is of course, notwithstanding SF (service factor) etc. And for motors or transformers anyway as long as temp rise was not exceeded the nameplate could be trusted. A long time ago NEMA changed FLA to RLA for their own reasons.

Having built power electronic equipment from 20 amp up to many thousands of amps the ones we built were tested at 110% of rated load (nameplate) for a minimum
of 8 hours before leaving the shop.

Needless to say this was not cheap consumer apparatus. I would hate to be the owner or specifier of a 15KV motor knowing that I could not absolutely trust the nameplate ratings.

Then again, it seems the world has changed as, the now, global manufacturers struggle to compete with junk made with the cheapest materials assembled by unskilled labor that is designed to last only as long as the warranties duration.

Remember years back when you put a NEMA starter alongside an IEC starter.
What's wrong with this picture, lol?

dann2 - 18-1-2008 at 19:48

Hello,

@Aqua_Fortis_100% (quite a handle)
The circuit that you posted some time ago (page 8 of the KClO3 via water.......) has some Portuguese written towards the bottom:
Obs: Os dois.......................isolades.
What does this mean?
Sorry my Portuguese is not too hot.
I will guess it says that you need to isolate the trannies?

Cheers,
Dann2

Xenoid - 18-1-2008 at 20:03

@ Dann2

I guess it just means the two transistors are attached using plastic insulators around the screws and mica washers between the transistor and heatsink.

I used a 5K pot. instead of the 4.7K pot, I found the TIP41 got very hot at the high setting, may need to increase the 100 ohm resistor. I will mount the two transistors on seperate heatsinks when I get around to building it.

Aqua_Fortis_100% - 19-1-2008 at 08:43

@Dann2 ,

The notes on the bottom should be next to this:
"PS: the two trasistors should be fixed on individual metallic heatsink and properly insulated"

Yes is exactly the same that you and Xenoid stated..

Note, As I've said early this circuit WASNT made by me.. Was a reply from a friend which know much more than I about e-stuff, and this was when I asked at a local eletronics forum..

I tried and worked well to regulate the current...

But as Xenoid said the TIP41C gets too hot within short space of time, so both transistors need of a heatsinker...Specially the 2N3773 or similair 'brothers' which need of a large heatsink.. Dont let the two heatsinker contact each other and/or touch the PC PSU..

Among the 4 transistors said there, I've used 2N3773 , but you can use others as 2N3055 , MJ802 , MJE13007 ..

The 2N3055 isn't realiable because that is very weak comparing with others and gives less current range.. The best said by the author is the MJ802 which is the strongest..

Yes , Xenoid is again correct.. Mica insulators are very good to do the job, authough I cann't buy it at the momment..

@Xenoid , about 5K instead of 4,7K pot , what max. current you get with your circuit? Seems very promising :D:D:D

Although I cann't got very high currents , since my PSU is quite old and poor (250W) and I dont want to "abuse" it..

The only problem I get when doing tests with my early chlorate cell was that the current wasn't that great.. I wished at least 7-8A , but the max. I got was 4,6-4,7A(because I will use that in a 1000-1500mL cell) .. I believe that is because of resistence created by poor connections and cell desing.. I will try that after finishing the treatment with my new anodes and soldering all connections and diminishing the anode-cathode spacing to increase current..If gets too much current then I should only manually control the current on trim-pot and its done , probably without mess..


Ah, I've also put a PC cooler in front of the heatsinkers to help it to avoid overheating..


[Edited on 19-1-2008 by Aqua_Fortis_100%]

[Edited on 19-1-2008 by Aqua_Fortis_100%]

Xenoid - 19-1-2008 at 10:53

@ Aqua_Fortis_100%

I have only wired up this circuit in an experimental fashion at the moment, just to "try it out". To increase the current, I used two 2N3055s wired in parallel (I just happened to have two mounted on a heatsink), and was able to control 0 to 8 amps into a "dummy load" of 0.4 ohms. To wire 2N3055s in parallel I believe you need to put a very low value resistor (say, .05 ohms) in each emitter circuit so they turn on evenly. The TIP41 got hot only at the very top of the setting (so hot, I burned my finger, it wasn't on a heatsink). I haven't got any further than this, at the moment. because I've been so busy with processing various solutions, and "tending" my cells. I will certainly go ahead with this, I think it only requires a little circuit "tweaking". It is a very useful, simple and cheap addition to a computer PSU.

Perhaps someone "more skilled in the art" could comment on the circuit (12AX7, Rosco, ....)

12AX7 - 19-1-2008 at 11:54

Exactly what circuit are we talking about?

Xenoid - 19-1-2008 at 12:12

@ 12AX7

The simple circuit for current control of computer PSU posted by Aqua... bottom page 8 of the "KClO3 by ...." thread.

Just below where you told me to "point my nose down away from the clouds"!

http://www.4shared.com/file/32991572/7dcdf563/Regulador_0_a_...

It needs a bit of a tweak !

Actually, I guess a MOSFET would be better in this application, and control it with a simple voltage divider network.

[Edited on 19-1-2008 by Xenoid]

[Edited on 19-1-2008 by Xenoid]

12AX7 - 19-1-2008 at 14:07

Oh, that's on my page 14 (most recent). Just what posts/page settings does everyone think they use here?...

A .RAR wouldn't have been necessary if the originator had any minutae of sense with regards to photography. What a horribly sized photo, and a line drawing saved as JPG? Sheesh! I was able to salvage them:

http://webpages.charter.net/dawill/Images/Adjustable_Regulat...

Edit: big image made link.

Drawn with my circuit symbols,



Picture:



That power supply is operating, not only open frame without a fan, but has no RFI filtering as well?! :o Yeesh...

Anyway, a regulator it is NOT. It appears to be a nearly darlington emitter follower, which since the TIP41 is supplied from extra voltage, explains why it gets so hot. (A 2N3055 might need around 0.3A base current for an output around 4V, causing the TIP41 to dissipate >2.1W, definetly needing a heat sink.) The rheostatic control method sucks, especially if the pot goes open circuit (as pots are prone to -- ever had a stereo with a scratchy volume knob?), putting full voltage on the output. The 1k resistor makes for a whacky control curve -- why couldn't the designer (so to speak) settle for a single linear potentiometer?

Far too much current can be supplied from the 100 ohm resistor -- if the TIP41 base is at 5V and the pot is fully open, there's about 80mA base current, and respectively, about 2A collector current flowing, for a dissipation of 14W!

The circuit obviously doesn't regulate, it only adjusts. The output impedance is roughly Rin / hFE, which is about Rin/400 after both transistors. For a Thevenin resistance of 90 to 600 ohms (highest in the middle of travel), the output resistance ranges from about 0.23 to 1.5 ohms, which is on par with typical load resistance: hardly a regulator!

To use a MOSFET in the same circuit, you can certainly control a large power MOSFET from a potentiometer, using the extra voltage supply to counter the gate offset voltage. But remember that power dissipation is still a problem, parasitic oscillation becomes a concern (at the very least, put a gate resistor of say, 100 ohms, as close to the gate terminal as possible), and output resistance is still nonzero, in this case being roughly 1/Gm, which might be 1/10 ohm or less -- better, but still not good enough to call regulated. At least you can saturate it effectively (7V should be enough for most MOSFETs to saturate to near rated Rds(on) at the currents we're talking about).

Where paralleling is concerned, BJTs are better, given an emitter resistor per (0.1 ohm, 10W is typical for something like 2N3055). MOSFETs have much more dramatic differences in Vgs(threshold), as discussed earlier, the best solution being an op-amp per.

In regards to electrochemical control in general...I use a big honking resistor. I have these things. And they work. Quite well.

Tim

[Edited on 1-19-2008 by 12AX7]

bio2 - 19-1-2008 at 14:25

.......The 2N3055 isn't reliable because that is very weak comparing with others and gives less current range.. ...........

This is just the circuit I've been looking for but in the 20 amp range @ 12 volt.

IIRC the 2N3055 in the TO3 package is good for absolute maximum 15 amps or so.

I have a few of these laying around and it seems that on a decent heat sink with a fan that maybe 10A continuous, should be attainable, given a reasonable voltage drop.

Aqua Fortis....do you have a schematic of this circuit as well as the point to point
diagram which you so kindly provided? Not that it's any trouble to derive myself, just wondering because I noticed your 4shared folder has multitudes of files and my Spanish ain't any good, lol, more like non-existent exept for the similar words with the O's on the end, lol.

Xenoid - 19-1-2008 at 15:12

@ bio2

Tim has reproduced the circuit, above, and a link to the point to point.

Tim, I realise it is not a constant current regulator, I pointed this out in another thread. But a simple circuit like this is ideal for use with a bunch of small chlorate cells where you want to manually adjust the current, from day to day, especially for testing purposes. It SAVES fiddling around with high wattage resistors to get the right current/current density for experiments!

BTW it is not Aqua's circuit, he only provided the link, don't blame him ... :)

Post/page settings? I haven't changed mine, so I guess they are default!

Rosco Bodine - 19-1-2008 at 16:29

It would probably work out better to use an LM334 three terminal constant current source with maybe a 5K pot across
it for adjustment of the current , which is used as the base current for the first transistor in a NPN Darlington pair . The LM334 would supply base current to a NPN Darlington pair , the LM334 and NPN Darlington pair being supplied from the +12V rail , and the emitter of the Darlington pair then driving the base of the NPN Output transistor whose emitter is connected to the load and whose collector is connected to any + supply rail . The load would be between the emitter
of the output NPN and ground .

The load current would simply be product of the hfe values , the transistor Betas of the three (or more) transistors in the stack multiplied by the set current for the LM334 .

Now that's current amplification :D via ye old stacked emitter follower .


[Edited on 19-1-2008 by Rosco Bodine]

Attachment: LM334 constant current source.pdf (759kB)
This file has been downloaded 862 times

Aqua_Fortis_100% - 19-1-2008 at 16:46

Quote:
Originally posted by bio2:
... your 4shared folder ...


Please , note , this is the thing I'm trying to say all times : This ISN'T my circuit , ISN'T my photo and ISN'T my 4shared folder.. I just don't have one.
Xenoid understood this and also has already replied your comment.

And yes, Tim, thanks by the comments.. Criticism is what is needed for that circuit and all other things, since almost nothing is improved without it .

Quote:
Originally posted by Tim :
A .RAR wouldn't have been necessary if the originator had any minutae of sense with regards to photography. What a horribly sized photo, and a line drawing saved as JPG? Sheesh! I was able to salvage them:

http://webpages.charter.net/dawill/Images/Adjustable_Regulat...


Tim , please, note , that image/circuit the original author just made quick & dirty , just to reply my doubt at local "orkut" eletronics community forum.. Not to post on this international forum :D Because of this it wasnt made in any fashion , is just simple. So anyway I just posted that link to see if is anything wrong with that circuit and ways to improve/change it.. And now as you and others give the viewpoints , I will be more carefull of what I'm doing with my cell suply/circuit , as seems that was quite controvertial..

Quote:

That power supply is operating, not only open frame without a fan, but has no RFI filtering as well?!


IIRC , as I said in "KClO3.." thread, this was a QUICK & dirty test , where the author just used things laying around him , not to made anything special like chlorates cell , but just to test that circuit , so you can note that the TIP wasnt on any heatsink , the 2N3773 was on a small and not proper heatsink , the small lamp that he used to test and also the poor open frame PC PSU , all that for a quick test..
Quote:
Originally posted by Xenoid :
BTW it is not Aqua's circuit, he only provided the link, don't blame him ... :)


Oh god.. Maybe was better if I just dont have posted that link.. :( I dont like bothering anyone in any discussion, even more that serious like this..

I just talked here because Dann2 asked what was written.

Again, please sorry guys.



[Edited on 19-1-2008 by Aqua_Fortis_100%]

Xenoid - 19-1-2008 at 17:02

I've just had another look at this circuit, not sure what I was doing wrong last time!

I used 2 2N3055s in parallel and was able to supply 0 - 20 amps into a .15 ohm gazillion watt carbon gouging rod resistor (dummy load). At the 20 amp stage the TIP41 was pulling about 2 amps, needless to say I didn't leave it at this setting for more than a few seconds. I left out the .1 ohm equalising resistors in the emitter legs of the 2N3055s and they still seemed to be getting equally hot! You'll need a big heatsink for each 2N3055 or fan cooling. I changed the 100 ohm resistor to a 150 ohm.

Works well, just what I wanted!

Edit: Hmmm.. strange things happen at high currents .. :o

With the 100 ohm resistor replaced with a 680 ohm, and using a 5K pot. and a reasonable quality analogue 20 amp meter. I got about 0 - 20 amps full adjustment of the pot. with the gouging rod load. At 10 amps the two 2N3055s on the same heatsink were quite warm and the TIP41 was pulling 200 mA, at 15 amps the TIP41 was pulling 510 mA and at 20 amps the TIP41 was pulling 970 mA. The TIP41 was on a small heatsink and didn't seem to get overly warm!

[Edited on 19-1-2008 by Xenoid]

bio2 - 19-1-2008 at 17:27

Thanks AX7 for the schematic. I was also wondering how this circuit could function
as a CC regulator.

You said.

.......in regards to electrochemical control in general...I use a big honking resistor.........

I've been using a piece of NiCr wire with adjustment by a sliding alligator clip.
Crude yet effective high power pot. At 15-20 amps it glows pretty bright, melting
wire nuts and welding the copper to the spring, lol, in my cobbled together test rig.

Is there a simple (low parts count) way to drive some high current transistors
from a LM317 wired as a constant current regulator to achieve a CC output?

Maybe some type of SCR circuit would be preferable however as I want to scale this up to 40-60 amps eventually.

The circuits I have seen for doing this (National Semi AN) are a little complex.

They use paralleled PNP bipolars driven by an LM317 but the current is only about
5A in the one I can't seem to find right now on National Semiconductors site.

You probably know the circuit I mean it's been in their AN forever but seems they have revised the LM317 Application Notes layout.

The problem I have been having is that it takes hours for the cell to fully heat up with
a consequent rise in current. Then depending on the ambient temperature it
may trip out on a hot day as I am trying to max out the PS output.

So the goal is to limit the current to the maximum the PS can deliver continuously
so the cell current is independent of temperature fluctuations.

12AX7 - 19-1-2008 at 18:40

Why obsess over the LM317? Not an appropriate choice here.

My cell rises to a temperature and current suitable for the ballast resistor (which is around 0.03 ohm) and supply voltage, which is just fine with me. You might preheat the solution, you might watch it more closely in the first couple of hours, you could even put a cooling bath around it, etc.

A purely electrical solution would certainly turn up a more or less constant-current supply. In that thread, I would go with a N channel MOSFET or PNP pass transistor with a series current sense resistor. An op-amp controls current, and the voltage drop can be quite small (especially with MOSFETs).

SCRs have no place here. You could build a phase controlled power supply (I've seen them before: quite beefy with the power transformer and filter choke!), but that's obviously no use after a low voltage DC supply. The only use an SCR finds in a DC supply is as a protection crowbar, to precisely blow a fuse, thereby protecting the sensitive, say, logic circuit which follows.

Edit: concerning my criticism, it was directed at whoever created said media. I don't particularly care for badly adjusted photos, and obviously, I'm not afraid to say so. If it was not yours (Fortis), you probably agree with my criticism; if it was yours, then I trust you got the message and will be more careful in the future. In either case, I thank you for accepting it. Criticism is one of those things that's ever so useful, but hard to apply, and even harder to accept!

Tim

[Edited on 1-19-2008 by 12AX7]

chromium - 20-1-2008 at 00:32

I wonder why people do not want to experiment with simple (or simplified) switchmode circuits. Everything that goes switchmode ends always so damn complex that no one ever dares to repeat original design.

About constant current supplyes: LM317 is ideal to 1 amp. I have not tryed to expand it with transistor so i can not comment how good this is. I suppose it could also be paralleeled - not that mad idea considering its cost.

I have noticed that simple constant current supplyes with FET and resistor or with trans and zener do not hold current that constant. It seems to depend a lot from junction temperatures (or whatever) and if one goes to 5 amp or more changes with time are more than just noticeable.

Opamap and transistor based constant current supply is simple to build but these tend to oscillate at some currents and load impedances - especially if darlington is used.

Btw there are DC to DC converter circuits that use thyristor based oscillators but for some reason they are not widely used (or known).


[Edited on 20-1-2008 by chromium]

12AX7 - 20-1-2008 at 06:03

Switchmode circuits can be very simple: all the complexity gets tied up in a chip. Take the UC3842 series for instance: current mode flyback converter. Just add MOSFET and a few resistors and capacitors and you have everything you need on the primary side. TL494s are very common in computer PSUs, which are only a step up in complexity -- though they do use a few more resistors in the process! Mostly it's that people are intimidated by the total complexity, not realizing the value of viewing it from a level of abstraction, a viewpoint which is valuable in reading any schematic.

The LM317 drops at least 1.25V, and more like 2.5V for any real application. That leaves only 2.5V from the 5V rail! Of course, the 5V rail needn't be used here, the 12V rail could be used at this current.

Constant voltage regulators do not parallel. Use a power transistor to bootstrap the extra current flow, as detailed in the data sheet. Constant current sources can be paralleled, though I don't know why you'd want to when the same premise can be applied.

Op-amp circuits only oscillate when the user doesn't realize the value of compensation -- something fundamental to any feedback circuit, which the user ought to already know!

Tim

bio2 - 20-1-2008 at 08:44

........Mostly it's that people are intimidated by the total complexity, not realizing the value of viewing it from a level of abstraction, a viewpoint which is valuable in reading any schematic..............

So true! I have often been originally intimidated by certain circuits I've built but when viewed as you say and taken piece by piece when assembling, they don't seem that complex after all.

Thank you 12AX7 for your input which I am considering for different applications and yes admittedly, I do tend to fixate on the LM317 due to it's extreme simplicity in CC control. They did used to make 3A and 5A units in the TO3 package in the past.

That being said, I recalled last night while considering how to control a 1200 watt cell I am starting to build ,what may well be the panacea for high output or otherwise
Constant Current control for electrolytic cells. The only simpler way would be the
series resistor but the method presented below is voltage independent, powers from the line and gives true CC control.

The old tried and true method of placing series capacitors in the AC feed to a bridge rectifier connected to the 120V or 240V house power is very simple and also bullet proof.

I use this method of CC control to charge batteries of any voltage at 6A from the line. It is reasonably accurate and is short circuit proof as well, with the voltage virtually collapsing under a shorted condition.

To prevent the voltage from going to peak without a load
another capacitor can be placed in parallel with the AC line downstream from the seriesed caps. Otherwise it's easy to get a very nasty zap if the load is lost
and accidental contact is made.

This method is elegant in its simplicity and the capacitors hardly get warm regardless of the voltage drop. The output is about 1A per 16 uF and I use some large motor run capacitors, the old oil filled type, that I have scavenged over the years. Old air conditioner carcasses are a good source for these with even a 1HP unit having 45uF caps or bigger. These old oil caps are very large compared to the more modern foil/polyester type and are virtually indestructible.

I have and excellent article from EDN that has all the calcs to analyze this type of set up and various improvements that can be made by simple modifications to the DC side of the bridge although it works just fine in it's simplest form using 1 cap (or bank) feeding a bridge. Unfortunately my disc drive is on the blink so I can't upload the article but if I am able to find it again on the net then I will provide the link.

I encourage every one to try this method. It is suitable from milliamps to many, many
amps. This method is often used to supply small consumer devices obviating the need
for a transformer. I suppose if one could get their hands on those large pole mounted power factor correction capacitors the sky is the limit!

12AX7 - 20-1-2008 at 08:53

Don't use capacitors -- use inductors. Much more economical for large supplies, and capacitors make big current spikes when connected suddenly.

You can use Ls and Cs in a 60Hz resonant impedance match, with whatever characteristic you prefer (for instance, the output matching network on my induction heater goes from about 60VAC, 10A from the inverter to over 100V, 160A RMS resonant energy in the tank circuit), but without a transformer on the line, you will never have isolation. Line isolation is one of the primary reasons to use a transformer.

Incidentially, somewhere around here I have notes on putting "soft start" on a light bulb. You might figure a 1:1 match, since the 60W bulb wants 120V, and the thing operates from a 120V supply. Fair enough. It has to be resonant at 60Hz. It has to hold a lot of VARs, since the bulb takes about 60J per second, so you need to store about that much to get a time constant of around a half second. The Q needs to be around 120, so the tank's reactance needs to be around 2 ohms at 60Hz. We're talking big reactive components here! It's no suprise such a thing has never been built -- but it would be neat to see a true passive soft-start light, wouldn't it?

Tim

bio2 - 20-1-2008 at 10:52

Yes, good concept. Thinking that a microwave ovens transformers primary winding
would make a decent output inductor. I will have to play around with this, as there are several old micros laying around in my junk pile.

Am I the only one that can't bear to throw away old electrical junk, lol? Years ago my garbage man neighbor would bring me old thrown out appliances. One day he comes over and says, "Hey man look what I found today, it looks almost new" It was a big all SS Kenmore 1500W microwave with hardly a blemish. Turns out that a blown fuse was the only thing wrong because a new one got it functioning again. I was tempted
to return it to the owner but my friend said that "That grouchy old bitch is so rich
she probably already has a new one".

The lack of isolation doesn't concern me as no one else will be operating the equipment and a series resistor does a half decent job of suppressing start up spikes which don't seem to adversely affect batteries or electro cells anyway. I have noticed that a noisy 120Hz wave from the rectifier may have even beneficial effects on an
electrolytic cell using SS electrodes.

Would you happen to know what the approximate inductance is for a 1KW or 1.5KW
micro transformer? These are somewhat deficient in iron mass compared to the typical
transformer of equal KVA.

I also have a few decent sized battery charger transformers laying
around which would also suffice, maybe in combination with caps for this application.

12AX7 - 20-1-2008 at 11:26

An MOT is good for up to, say, 10A primary current (with fan), which is 1200VA, but not 1200VAR. To get that, you need to gap the core. As-is, the core probably saturates right around 120V (at the peaks), at a reactive current of only a few amps. You need to increase saturation current and decrease inductance to get a useful power transfer to whatever's after it. Cutting off the "I" section and putting it on an adjustable spacing to the "E" section will work. The least inductance you'll get (which may not be small enough for your load, but that's to figure out) is with the I completely off. You should use something stocky (like heavy solid cardboard -- not corrugated!) to space the pieces, and hold it with a clamp.

I will never condone offline circuitry. Surely it has occurred to you that you're connecting a *bath* of liquid to 120V? Doesn't that seem the least bit stupid? Not to mention, by using a full wave rectifier, you *guarantee* that bath has full AC on it!

Tim

bio2 - 20-1-2008 at 13:56

OK, I will try the gapping at different spacings to see the effect.

BTW, I just tried it with a small MOT (maybe 800W) with the secondary winding removed and produced 2.6A. The cell voltage was 5.6V (4 series water electrolyser) I only left it run a few minutes so the ultimate
internal heating effect was not determined. It was barely warm after 5 minutes or so.

........I will never condone offline circuitry. Surely it has occurred to you that you're connecting a *bath* of liquid to 120V? Doesn't that seem the least bit stupid? Not to mention, by using a full wave rectifier, you *guarantee* that bath has full AC on it!.......

I hear you and I wouldn't expect an "electronics" guy to approve anyway.
However, there are various applications that run even thousands of AC volts into a liquid bath.

There would only be stupidity in the operator causing an accident
if adequate safety precautions are not observed. I have seen some pretty sharp
"electronics" guys do some pretty stupid things around line voltages as I am sure
you have as well. Ever close a 2KA 12KV breaker into a fault? That's very exciting
yet done all the time.

With a sealed tank, bubbler buffer, dual flame arrestors and good fusing my experimental rig is safe enough given constant observation by me.

Needless to say commercial units have multiple layers of safety devices for
the obvious reasons. Many of these units run line voltage into 60 or 120 cell
banks and industrially thousands of DC amps are run into electrolysers full of
very conductive electrolytes as the norm, I once worked on an 8KA unit and
that is a small one. So no I'm not worried.

I do appreciate your concern and to those youngsters out there or those without
relevant experience; please do not try this at home as it may kill you.

How's that?


[Edited on by bio2]

[Edited on by bio2]

12AX7 - 20-1-2008 at 15:32

Quote:
Originally posted by bio2
How's that?


Still no excuse! ;)

I haven't been around 2kA breakers, but I've been around 480V, 1.6kA breakers. Big, warm and humming. Lots of power flowing there. Good thing they're safely behind panels!

Rosco Bodine - 20-1-2008 at 15:32

Quote:
Originally posted by 12AX7
Op-amp circuits only oscillate when the user doesn't realize the value of compensation -- something fundamental to any feedback circuit, which the user ought to already know!

Tim


A little knowledge is a dangerous thing .
Try harder not to be such a scary guy :P;)

Op amp circuits will oscillate when the frequency response of
their feedback loop doesn't rolloff to 0 dB before the phase shift in the feedback path is sufficient to become positive and regenerative as opposed to attenuating . You won't get a stable servo lock in a control loop unless the Bode plot of the frequency response for that loop shows that stable response envelope is present for the whole loop , and each element in the loop contributes it's own phase , not just the op amp or the way you have it compensated .

Controlling a hugely capacitive Mosfet with an op amp is
about like making a ten foot flyrod into a bow with a heavy
rubber band from tip to butt , and having a bamboo hoop
on the tip end of that lightweight bow , loosely equatoring
a bowling ball which you wish to push and pull across the
floor in a precise motion :D It presents similar difficulties
as would be seen with the grappling of the Hubble space telescope by the shuttles spindly robotic manipulator arm ,
where the mechanical forces involved with positioning become a complex mathematical calculation that is much more involved than may be apparent . I mean it's just
simply reaching out and grabbing something right :D

Controlling that big mosfet with an op amp is slippery business , like trying to push a catfish around by his whisker tips .

bio2 - 21-1-2008 at 12:55

An update on incorporating 12 AX7's suggested improvements to an inductor
seriesed with a bridge rectifier for Constant Current electro cell control.

I took an about 1.2KVA MOT and cut a 1mm gap in the I of the core and
about 4mm of the laminations below the cut were removed on both sides (these fell out anyway). Connected to the 120V line this arrangement gave 7.2V
6.8A on the unfiltered DC output to the cell. Without modification this MOT only produced 1.8A at about 5V. The same electrolyte concentration and temperature
was present.

The winding quickly heated but no temp rise measurements have yet been taken although it seems that a cooling fan will probably be needed. The current and voltage are very steady and according to 12AX7 this is a better method than using capacitors alone so it should give reasonably accurate CC over a wide output voltage range as do caps.

The energy density (space required per amp) is much better than caps so this
method should serve well for the 15A @ 80-100V cell I am building.

dann2 - 21-1-2008 at 13:59

Hello,

15A @ 80 to 100v?

Are you putting a number (say approx. 12 X 6v cells) in series or just one cell at 6 volts, the rest being dropped accross the inductor?
The biggest disadvantage/danger with this system is that if one of your connections gets dogy the voltage (and power dissapated) will start to increase accross the dodgy connection and it will go red.

A similar thing happens if you are using a welder (+ rectifier) to drive a cell.
You set the current with the welder controll (say 30 amps) and the voltage across the cell will be approx. 6 volts. The welder inductor drops the rest, approx. 74 volts. The open circuit voltage of the welder (a common mannual metal arc type welder) will be approx 80v ac.
If a connection gets dodgy, the voltage will start to rise and rise accross that connection untill its gets red hot and there will be melting/ignition.
If you had a variac in front of the welder you could bring the open ciruit voltage down to some sensible level, like 10 volts so that you will not get red hot/ignition problem.

This happened to me BTW, I used to run cells using a 120 amp welder. Make sure connection etc are top notch.

Hope I am not misreading what you are doing.

Dann2

bio2 - 21-1-2008 at 15:40

Yes, you are misunderstanding, please reread my previous
post on using capacitors as constant current control.

I have used this method for years and it works very well. The voltage will automatically change with the current remaining constant over a very wide range. Inductive Reactance is what determines the current.

I have charged batteries at constant current from 6V to 48V at 3.7A using 90uF capacitance connected as previously described.

12AX7 said inductors work even better in this application and I have no reason to disbelieve him but am now doing testing.

My water electrolysis unit is run at 2V per cell so 40-50
are needed for 80-100V derived from the 120V line thru
seriesed caps or inductors. As I already posted my test rig
is 4 cells in series.

I would run about a 60 cell but am short of stainless steel for the electrodes. I will expand it later when I find some more suitable SS scrap. Right now I am cutting up a large SS muffler and tailpipe scavenged from a boatyard.

Your bad connection warning is well advised and for this reason they are being welded to the electrodes. Good fusing
is also used to preclude unexpected failures. The connections
are also made at the bottom of the electrodes to prevent any possibility of a spark as they are submerged in the electrolyte.

Other safety devices needed are not being described at this time. I get real tired of repeating myself because the posts
haven't been read using any attention.

I suggest you try the capacitor method. About 1A is produced for each 24uF capacitance. My previous statement of 16uF
was in error. Just gang line rated caps with adequate current
capability to acheive the desired current.

I still haven't looked for the EDN article describing in detail the theory behind this method. It was entitled XXXX Rectifier
not much help, sorry. Although I have never seen one the article says commercial battery chargers have long used this method for current controlled charging and in my experience it works very well.

I have a 1320AH battery bank and use this to charge CC at 6,12,18 or 24V depending on how I have them connected. So you see the current staying the same means it is advantageous to charge at 24V due to the higher available wattage. Then I reconnect to 6 or 12V and use the same unit as a trickle charger.

dann2 - 21-1-2008 at 16:30

Hello,

Apologies, I though you were going to run a (Per)Chlorate cell.

If anyone else is thinking about running a (Per)Chlorate cell using this method the problem I outlined will apply. The bad connection can (will?) come at the top of the anode and or cathode if salts are getting to the connection.

Also if/when the coating falls of a Ti substrate anode a large voltage will appear accross the cell and eat away the Ti.
In my neck of the woods it's 24OV mains so that a bit too big a voltage for using this method for just about doing anything with.

I am not saying that the method will not work, just these's problems may and can occur with (Per)Chlorate cells.

Dann2

[Edited on 22-1-2008 by dann2]

bio2 - 21-1-2008 at 17:53

The max voltage produced can be limited by paralleling
another capacitor of appropriate size (much smaller) across the mains downstream of the seriesed caps. This is normally done to protect against peak voltage with loss of load and the voltage will collapse in a short circuit so it is self limiting.

Put the caps in the hot line preferably although it will work
in the neutral being AC of course.

So 240V should not be a problem. For example if the desired current produces say 12V it can easily be limited to say 18V or any other value.

I have done this and it works very well. To get a feel for it
just get a handful of small across the line rated caps and
a small bridge and experiment with maybe 100mA or so
into some little resistors..

dann2 - 22-1-2008 at 08:06

Hello,

Thanks for that Bio2.
It took a bit of hammering home.

Dann2

No User Serviceable Components Inside

Xenoid - 1-2-2008 at 17:33

Here is the ATX PSU I have modified with current control, using the simple circuit from the site originally linked to by Aqua_Fortis_100% and subsequently reproduced by 12AX7 a few posts back. This simple little modification is great if you are working with electrochemical cells or doing electroplating. It allows one to dispense with "dropping resistors" to get the appropriate current. Note, this is not a constant current circuit, and current will vary somewhat as cell parameters change, it is however easily readjusted using the potentiometer control.

I used an ATX purchased from a recycling centre for $3, the only additional expenses were for the two heatsinks for the dual 2N3055s, these cost me $7 each. All the other components I had on hand in my electronic parts collection.

With dual (parallel) 2N3055s I can control up to 20 amps into a suitable load. I specifically built this with (per)chlorate cells in mind and I found that replacing the 100 ohm resistor with a 680 ohm gave me a good control range (the 1K resistor was dispensed with, see note below). Image 1. shows the modified ATX delivering 15 amps (20 amp meter on top of the case) into two of my carbon gouging rod resistors wired in series, as an approximate simulation of a chlorate cell. The multimeter is indicating a voltage drop of 3.66 volts across these resistors, indicating a total resistance of .24 ohms. In doing so the 2N3055s are dissipating 20.1 watts ((5 - 3.66) x 15) and are slightly warm. With the heatsinks shown, about 40 watts is the maximum I would want to dissipate on a continuous basis.
The yellow terminal is connected directly to the 3.3 volt output of the ATX, it was easy to do, but I will probably never use it!

The two 2N3055s on their respective heatsinks are mounted on the sides of the ATX cover, and the base, collector and emitter connections are joined in a terminal block (Image 2). This allows the lid to be completely removed by unscrewing the connecting wires. I did not use .05 or .1 ohm resistors in the emitter legs of the 2N3055s because with the two sets of transistors I have tried in this circuit they were not needed. The current seems to be split quite evenly between each transistor.

The internal arrangement is shown in Image 3. The red line points to the rear of the 5K potentiometer. The yellow line points to the TIP41 mounted on the black heatsink, which is screwed on to the front of the case. In this position it catches good air flow through the front vents. The 680 ohm resistor is mounted on a small piece of strip-board screwed to the bottom of the heatsink. Just to be safe, a piece of plastic (blue in image) is inserted to separate the high voltage components.

Note: In my earlier ramblings on this subject, I mentioned that the circuit worked well. 12AX7 suggested that the 1K resistor was not required and would effect the control circuit (whacky control curve). This only became apparant to me when I built the final version for mounting in the ATX. My earlier test circuit was built as a "rats nest" on the bench and I grabbed the first 1K resistor I found, when I found my final version was not behaving properly, I had another look at the circuit, did a few calculations and then realised I had used a new 1K resistor. The first 1K resistor had in fact been a 10K and had an orange multiplier colour band that looked red. So the circuit worked fine with the 10K resistor, it works better if you leave it out all together!

ATXmods.jpg - 81kB

dann2 - 2-2-2008 at 16:28

Hello Xenoid,

Wondering what is the max. voltage out.
Will it be high enough to drive current into a Perchlorate cell?

Cheers,
Dann2

Xenoid - 2-2-2008 at 20:35

@ Dann2

Well there are perchlorate cells and perchlorate cells, mine have operated in the 4 - 4.5 volt range at currents up to 5 or 6 amps. I have never actually run a high current perchlorate cell.

Unfortunately you will take a big performance hit having those old 2N3055s, even when fully turned on they treble the internal supply resistance from about .02 - .03 ohms to .06 - .09 ohms according to my measurements. This limits the current available. Really, when I built this I was more interested in controlling in the 0 - 5 amp range for testing anodes. I'm sure the circuit can be fine tuned a little more, the 680 ohm resistor could be lowered and alternative transistors used, its just that I had those on hand.

I have some comparative figures for the modified variable supply compared to a "plain" 5 volt supply, using different load resistances. The voltage is measured across the load resistance.

.27 ohms
Variable (max. setting) (3.70 volts - 14.1 amps)
Plain (4.44 volts - 17.1 amps)

.4 ohms
Variable (max. setting) (4.16 volts - 10.6 amps)
Plain (4.63 volts - 12.0 amps)

.5 ohms
Variable (max. setting) (4.35 volts - 8.6 amps)
Plain (4.71 volts - 9.3 amps)

.8 ohms
Variable (max. setting) (4.67 volts - 5.6 amps)
Plain (4.87 volts - 6.0 amps)

So depending on the type of perchlorate cell you have it is unlikely you would be able to control more than about 0 - 10 amps into the cell. If you want more than this use an unmodified supply with a very low resistance in series.

Actually, I am thinking I might replace the 3.3 volt output terminal with a 24 volt / 20 amp switch and so be able to switch out the transistors for when I don't need adjustable current.

DerAlte - 2-2-2008 at 22:35

I don’t usually disagree with Xenoid’s posts, but to call an emitter follower circuit a “current control” violently offends my engineering sense. The cell controls the current! It is at the mercy of a low impedance supply generously provided by dissipating volts in a transistor instead of a resistor. There is no current control at all – you are at the mercy of cell temperature, electrode level, electrolytic composition, and the phase of the moon.

If you want current control, use a circuit like those proposed by 12AX7 and Rosco Bodine. I’d rather use a junction transistor (with collector output) for this but I am convinced MOSFETs can be made to work, with some care. (I suspect Rosco might agree after we chatted about these matters on U2U).
******
I wasn’t going to post this, which I wrote some time ago, but it is partially germane to the topic and may help a tyro or two:

Re: PSU’s in general, electrolysis and Lamps as current limiters.

I hope this might be of some interest to the non-expert unversed in electronics. It started out as a quick note about using lamps as ballast and somehow grew… Rather than start yet another thread in a crowded field, I decided to tack it on to this one.

Some of you may have been intimidated by the electronic content of this thread and suppose that the whole business requires complex circuits that you don’t understand, and that the crude old method of using your battery charger as a source of DC is completely passé. Don’t be. You can use one to do many electrolyses but they it not work quite as nicely as the stabilized supplies promoted by Rosco or 12AX7.

Computer PSU’s give you fixed controlled voltages: 12v, 5v, and 3.3v normally- unless modified as suggested in the thread. You can control current drawn by electrode spacing (wasting the rest of the power in ohmic dissipation in the cell and heating it), with a series resistor, or varying the immersion depth of an electrode in the electrolyte. But really what you want is a prescribed current density most of the time. This determines the cell voltage*.

If all you have is an old auto battery charger the available current is usually limited to about 5-6 amps or so and it probably has 6/12v capability. (The charger usually has a current limiter as protection. You disable this at your own peril. You may blow the diode rectifiers or overheat the transformer). The voltage waveform is that of a full wave rectified sine wave, not ideal for electrolysis. The current density varies wildly over a cycle. Never slap a charger across a cell and hope! The current must be limited somehow.

First, you need to smooth it by using a shunt capacitor/series inductor/shunt capacitor filter using at least 1000 uF capacitors (rating at least 15v; eg Radioshack) and at least 20-30 millihenries inductance. Getting (buying) the inductor will not be easy. It should not have more than 1 ohm series resistance. Wind one on an fat iron core using adequately thick wire, leaving an air gap of 10-20 mils in the magnetic path. This will reduce ripple to adequate levels. (I’m not going to give any formulas but this core must not saturate under maximum DC load current – technically the air gap increases the reluctance of the path and ensures the flux does not saturate the core. The real design is complex and only able to be understood by an EE or a practical physicist).

A good alternative is to use a 12v lead acid battery across the charger to smooth it. This acts like a giant capacitor, and also allows currents vastly in excess of the charger’s 5 amps to be used by a cell (while the battery remains charged). The capacitor causes the peak current from the rectifier to be high, but the same is true of the battery, and chargers take this into account.

To get a given current density you must know the immersed area of the cathode or anode, at whichever the desired product is produced. The other electrode is often the same or larger in area. Only a cylindrical set up allows this to be estimated accurately. Parallel plates do conduct on both sides even if the other electrode is facing only one. The density will vary some over a flat plate but an fair average can be easily estimated.

The internal resistance of a cell is difficult to compute. The expected voltage drop (from SEP or experimental evidence) needs to be estimated. The rest of the voltage from the PSU must be dissipated in a series resistor R, computable by Ohms law. Without current control, the current drawn is then the I=V/R, where V = (nearly) the average PSU voltage (under load) minus the cell voltage, and I the desired current. If the cell voltage varies, the current varies*.

As an example, suppose the charger or PSU produces a nominal 12 V. The drop in the cell is estimated to be, say, 3V and the desired current 3 amps. The total cell ‘resistance’ is then 1 ohm. We need a total of 4 ohms to produce 3A. So the series resistor should be an extra 3 ohms, but we have to include our smoothing inductor and source resistance, so we’ll need a bit less than that. You need an amp meter to adjust it.

R ought, then, to be a resistor which can be varied from say, a maximum of 5 ohms down. If you have a rheostat of this size and rating, fine; otherwise you could use a length of suitable nichrome or other resistance wire and a slider, as I do. Or, for higher currents, a stout carbon rod with a slider works nicely.

For the same example, what happens if the cell voltage increases to 4v? The current in the 3 ohm resistor becomes (12-4)/3 =2.67 amps – not a huge change. If cell volts drop to 2v, we get 3.33 amps. The series resistor limits the current change considerably compared with what you would get by slapping, say, a 3.3v PSU source straight across the cell..

So where does the light bulb come in? It can act as a current variable resistor. As the filament heats up, resistance increases, very approximately as the absolute temp in deg K. (R(T) ~ = R(300)*T/300).When taking no current at room temp it has minimum resistance; at full rated voltage, resistance is maximum (it’ll melt if you go much higher!). This change is about 9 times from zero to rated volts for a tungsten filament.

Making the (gross!) assumption that the filament obeys Stephan’s law of radiation and its resistance follows temp. as shown, and that all the heat goes into the filament and is radiated, I derive that the I/V law is about I=V^0.6 or V=I^1.67 (a dedicated experimental scientist could find out what it actually is for a given lamp. A higher exponent can be expected for the actual I V equation. Heat conduction and convection in the gas inside the bulb mean that my figure is only a gross approximation).

Let’s see what we could do with a light bulb as the series resistor R. The idea was suggested by, I believe, Twospoons in this or some thread recently. It is a method I have often used in the past in spite of being an electronics engineer by trade. I never felt the need to build an exotic stable current controlled source for this application.

Suppose we took an auto lamp rated at 12v (for a reasonable life on our 12v charger). Suppose we could find one with a 9v drop at 3 amps (unlikely!) – it would then fit our requirement. At a full 12v it would consume I = 3*(12/9)^0.6 = 3.57 amps (43 watts – a 12v, 40w lamp is close).

At a cell voltage of 3v, then, I=3 amp. At a cell voltage of 4v, I = 3*((12-4)/9)^0.6 = 2.8 amp, and at 2v the figure is about 3.2 amps. Not a bad control. Much better than just a series R alone, but not as good as a settable current source..

Of course you can use any voltage stabilized PSU the same way. For massive cells taking 20-40 amps like 12AX7’s chlorate cell you’ll probably need to build your own crude transformer/rectifier set-up – or find one as suggested in the thread above. I doubt you’ll find a suitable lamp for that.

For voltage setting, if you happen to have a variac (expensive) you can vary the input to your crude PSU to vary the output (not a computer type or a stabilized lab supply, though). I’ve never tried it, but a suitably rated light dimmer might work, (SCR’s generally don’t like inductive loads). Or you can use a transistor circuit like Aqua_fortis showed above to adjust the output voltage (but not stabilize it to a fixed level). A similar circuit with collector output can be used for current setting.

*{Note: The electrolytic cell voltage depends upon many factors, current density, temperature etc, anode and cathode material and preparation, separation of anode and cathode, whether or not a diaphragm is used and so on. To understand what determines the relationship between voltage and current density and the relationship to SEP you need to research. Look under ‘Tafel’s Law’ or ‘Tafel equation’. A good starting resource is

http://electrochem.cwru.edu/ed/encycl/art-t01-tafel.htm}

Der Alte

Xenoid - 2-2-2008 at 23:38

@ Der Alte

Clearly you are not reading my posts correctly .... :mad:

NOWHERE have I claimed this is a CONSTANT current control! In fact I have gone out of my way to stress that this circuit is little more than a variable resistance on several occasions, I won't quote them all, but in my earlier post, note the fourth sentence.
Quote:

Note, this is not a constant current circuit, and current will vary somewhat as cell parameters change, it is however easily readjusted using the potentiometer control.


I am fully aware of its limitations, but it is a useful circuit for varying the current through a cell. Constant current is not even NEEDED in a (per)chlorate cell. I have just finished a run with my 10 litre cell which used just a variac, halogen lighting transformer and a couple of bridge rectifiers. Unfiltered DC and I barely had to adjust the variac over the entire 21 days!

I am not at all sure why the term "current control” violently offends your engineering sense, it doesn't offend mine. I set the control to minimum, attach a chlorate cell, turn the control to the desired current, "voila" "current control. What you are referring to, is constant current or current limiting. I have never, ever, claimed that of this circuit!

In fact it works so well I'm going to build another, I may experiment with different components however.

LADIES PLEASE *

dann2 - 3-2-2008 at 20:47

Hello,

Actually I don't defend Xenoid too often!! but he is operating on a circuit (as he has said) that has been described as being really like a handy variable resistor.



Cheers for input.

Dann2


* coppycatting, but just could NOT turn down that opportunity..............:D

DerAlte - 4-2-2008 at 18:58

@Xenoid,

Sorry a few ill chosen words caused offence! I have every respect for your experimental technique and achievements so well displayed in these pages.

And "violent"offence was putting it a bit strongly. I merely meant to indicate that this circuit was not a stabilized current source but actually a low imedance voltage source. You did not explicitly claim this I agree, but I guess I got the impression you had. I am certain you are very congnisant of its limitations, and they aren't that severe.

Also since current and voltage are roughly represented by the equation V(cell) = A + B*logI, (Tafel's Law), actually the voltage across the cell is not a particularly sensitive function of the current density- unless passivation or electode erosion takes place. Hence even a low impedance voltage source is often quite satisfactory in a well behaved cell.

The current density (current) once set by a variable voltage is likely to remain reasonably constant at least while reactants are not fully consumed. And the transistor voltage dropper is much neater than a fat old resistor. I am not denigrating it at all. As Dann2 says, it a very convient way of dropping volts.

A fixed current supply can be made almost as simply, with junction transistors, should one desire.

The rambling post above was actually intended to give a boost for the light bulb as a half-arsed current stabilizer cum voltage dropper, superior to a fixed resistor, and to point out to the unitiated that the humble battery charger or simple rectified supply were still good for elecrolysis. It sort of grew under its own momentum.

Regards,

Der Alte

12AX7 - 5-2-2008 at 10:44

I already gave an extensive analysis of this circuit; further analysis and arguing is wasting extra posts (and you know how *I* hate wasting posts).

Tim

DJF90 - 2-5-2008 at 08:30

I'm kinda stuck. After removing the PSU from my old pc I have sorted out the wires. Green needs to be soldered to a black, and a 10ohm 10W resistor needs to be soldered between a red (+5V) and a black. All well and good. However the problem arises in finding "terminals" that can handle the current. The best I've found so far are 6A rated 4mm test sockets and matching 4mm banana plugs. But is this sufficient enough? I cant find anything more appropriate :( The PSU is rated 235W, and has a maximum current output of 22A (at +5V, 14A at +3.3V, and 8A at +12V). There is also no 3.3V sense wire (although one of the orange (3.3v) wires is thinner than the others and is connected to a different part of the circuit... guessing thats it?). A wire that I have no idea what to do with is a white on, stated on the label as being P.G. Signal. I figured it was nothing important as it has not been mentioned in any of the methods I've found online so far.

Any help would be much appreciated :P thanks in advance

12AX7 - 2-5-2008 at 09:17

1. Then don't use them for more than 6A each.
2. Ignore the warnings and ease off on load if they start getting hot.
3. Use twisted connections. My lazy, informal power supply is wired this way.
4. Get better terminals. 1/4" quick-connect lugs are good for 30A or so. Binding posts and screw terminals come in various sizes. Bolted lugs are available well into the kiloampere range for power distribution; check with your local electrical supplier.

Tim

DJF90 - 2-5-2008 at 09:26

I intent to purchase the required components from maplin (www.maplin.co.uk). If you could select the appropriate parts then that would be very helpful, and much appreciated :P
Quote:

Then don't use them for more than 6A each.


How is this possible when the whole idea of using a PSU is for the high current properties (i.e. electrolysis)? Please elaborate.

I intend to make my PSU much like the one on Woelens webpage, so maybe he could provide the specs for the components he used?

[Edited on 2-5-2008 by DJF90]

Xenoid - 2-5-2008 at 13:51

6 Amps seems very conservative for a binding post, I have used these at 15 Amps with no problems and I am sure they could handle 20 Amps. As a rough guide, a copper conductor of 1 mm^2 cross sectional area can carry a nominal current of 10 amps. And about 2.3 mm^2 is required for 20 Amps.
A 4mm diameter binding post has a cross sectional area of say about 7 mm^2 (allowing for the depth of the thread grooves). If you are still worried, make sure the internal connections are soldered on, and dispense with the banana plugs and wrap your wires around the post or put through the hole if it has one.

As an alternative you can use bolts, nuts and washers from a hardware store. This is what I used on my 48 amp halogen lighting transformer (see earlier in this thread). Brass would be best, but zinc plated steel is OK, make sure they are mounted on a sturdy piece of insulating plastic of course.

The PG wire can be disconnected. The 10 ohm / 10 watt resistor is to provide an initial load to enable the supply to start up. It seems a little excessive, try higher value resistors (and consequently lower wattage). Some of the "newer" supplies need very little load to start, my last supply did not require any additional load resistance at all.

12AX7 - 2-5-2008 at 13:58

Quote:
Originally posted by DJF90
Quote:

Then don't use them for more than 6A each.

How is this possible when the whole idea of using a PSU is for the high current properties (i.e. electrolysis)? Please elaborate.


Each, as in, use several in parallel.

Tim

DJF90 - 2-5-2008 at 14:32

Thanks Xenoid. I think I'm gonna go ahead and order the plugs/sockets, heatshrink and a resistor. What sort of current should I expect on the +5V line with a 10ohm resistor in place? I'm unsure about how much load the PSU will actually need to power up, its from an old pentium II pc.

@12AX7; Using two(or more) binding posts per voltage (in parallel) is not a bad idea but that would mean doubling or possibly tripling the cost of the components (as i would need more sockets and plugs). I think I will just see how things go and if overheating occurs then maybe your suggestion will come into play.

Xenoid - 2-5-2008 at 15:14

Quote:
Originally posted by DJF90
What sort of current should I expect on the +5V line with a 10ohm resistor in place? I'm unsure about how much load the PSU will actually need to power up, its from an old pentium II pc.


Ohm's Law: E=IR or I=E/R therefore I (current) = 5 volts / 10 ohms = 0.5 amps
Wattage = EI = 5 x .5 = 2.5 watts, a 5 watt resistor is probably OK.

There is no problem using that resistor, it's just that high wattage resistors are fairly expensive, you may get away with using a higher resistance 1 watt resistor or even no resistor at all. With 20+ amps available from the 5 volt line, loosing .5 amps to the load resistor is no great loss!

If the supply is starting OK with the 10 ohm resistor, try replacing it with say a ~50 ohm or higher. A 50 ohm resistor will only draw 100 mA and only .5 watts will be dissipated.

DJF90 - 2-5-2008 at 15:37

Thanks alot. I have an A2 physics exam soon and I know all this, for some reason I thought that the I=0.5 amps was what would be output current. Guess I didnt learn it well enough :(

Xenoid - 2-5-2008 at 16:07

Yeah! The resistor is a "dummy load" to draw a little current and insure the supply starts up. It means you can switch your supply on without having "your" load connected. If you connect "your" load (eg chlorate cell) first, the supply will start up without the resistor anyway! It's just a convenience feature really!

Don't go connecting your cell or whatever in SERIES with THIS resistor.

However you may need a "dropping resistor" of some sort in series with your cell to limit current (see discussions earlier in this thread).

Good luck - :D

DJF90 - 2-5-2008 at 16:16

Thanks :) I read about the dropping resistors on woelens site, very helpful resource that... So the resistor just needs to be in the PSU connected between one black wire and one red wire correct (black is gnd, red is +5V). And then the other black wires are connected to the gnd terminal, and the other colours are also connected and wired to their own terminals right? the reisistor itself is only 40p. Thats about 70c IIRC so its not all that expensive. I'm just hoping that the plugs and terminals hold up, thats the only place where I can see this project falling on its face :(

Xenoid - 2-5-2008 at 16:38

Here's how I did it on one of my old supplies!

Disconnect all the existing wires, use a thick red wire for the +5V and a similar wire (blue) for the -ve (earth). Connect the startup resistor across the output terminals (in this case I used a 33 ohm 5 watt resistor - overkill). Those binding posts will easily handle 20 amps.

Note: I'm only using the +5 volts, not bothering with the +3.3, +12 or -12 volts etc.

[Edited on 2-5-2008 by Xenoid]

PSU.jpg - 44kB

saki2fifty - 23-10-2008 at 18:16

I noticed that the last message on this particular thread was over a year ago, so I may be replying to something that live in a void somewhere.

I have been searching the web to locate a lab power supply, stumbled across the million conversions involving the ATX Power Supply, but not really what I wanted... I was looking for something a bit more functional, more control, and by chance, I came across this site. I am directly replying to the schematic that Rosco posted on the first page (a goldmine I say). Rosco posted 3 revisions of this current control / limiter, voltage control to turn your typcial ATX power supply into a fully regulated supply, which is exactly what I've been searching for.

So... my question is, and after reading the entire thread many times over, not fully understanding all that has been fought over, has it been decided that this last .pdf posted on the first page of this thread, that it should be functional with no issues? I am wanting to build this. I am eager to put an order in for the parts and get the project going...

Questions:

-Has anyone built it?

-Is there a PCB layout available? (planning on using perf board, just curious if there was a layout available)

-If I wanted to regulate from a different power source other than an ATX power supply, what is the max current (guessing by the title 50a) and voltage?

-I am also wanting to build another circuit with only the current control / limiter only with only (1) 12v power source... i guess it would be too much to ask for an altered schematic of just that?

-Last, the components listed, are there other parts equivalent to whats listed that would be more readily available? I've found the parts with not much problem online, but locally, the parts stocked seem to be limited... mostly NTE stuff.

Thanks,
Michael

12AX7 - 23-10-2008 at 19:28

Quote:
Originally posted by saki2fifty
So... my question is, and after reading the entire thread many times over, not fully understanding all that has been fought over, has it been decided that this last .pdf posted on the first page of this thread, that it should be functional with no issues?


Uhhh?

The entire thesis of my point is that it will not work. This will;



You will probably need a small capacitor from each op-amp's output to its -in to compensate the response, otherwise it'll just oscillate, and that ain't no good. Easiest to analyze with an oscilloscope, but simply measuring AC voltage on the output (after checking that your meter doesn't misread DC as AC, because the cheap ones will!) will tell you if it's bonking out.

Alternatives include hacking the PSU's voltage regulation, which works to some extent. There should be a trimmer potentiometer on the board, probably gobbed up with glue, that sets the voltage. Un-gob it and you can adjust it over a small range. You can follow nearby tracks to see if there are resistors limiting its range (there should be); altering these allows much more adjustability. DO NOT set any voltage greater than the rating of the capacitors on that rail (I believe they are usually 6.3 or 10V for the 5V rail, and 16 or 25V for the 12V rail).

Tim

Rosco Bodine - 24-10-2008 at 00:28

At the point where I left off working on this there were still
concerns about the stability of the output stage. The module
which seemed most likely to work was based on using the LTC1152 as the mosfet driver. This would have been used in the place of the output stage shown in the complete schematic, but I haven't had time to go back and edit the ECAD schematic to reflect that change.

It would be interesting for someone having the time to prototype and evaluate the stability of the LTC1152 driven Mosfet and its damped feedback loop which I modeled,
and compare it to Tim's solution, see what is stable.

I don't have time to work on this. And since I already have
three huge working power supplies , a 50A, a 120A, and a 150A , there are other things I don't have that I will be building when I do have time.

I would say that the "engineering" on either of these is
unusual to say the least :D and there are uncertainties
which will probably lead to debugging / revising the actual
designs . No guarantees on this stuff from me , and if
you get them from Tim ......he's lying :P

As for substitution of equivalents, there will be little joy there.
The specs are pretty sensitive on everything I surveyed as
candidate parts with few substitutions even possible.

[Edited on 24-10-2008 by Rosco Bodine]

saki2fifty - 24-10-2008 at 06:43

Well, i'd be more than interested in putting in the time and resources in building both circuits, testing, logging my work and construction, etc. It would be great however if I could get a complete and modified schematic and possibly the cad file itself?

If on Tim's solution there were exact part numbers of the IC's of choice, as well as all parts (I trust your judgment over mine), much like what Roscoe has, I can start on ordering the parts, building and testing.

I'm not wanting to hack the power supply but rather build what you 2 guys have argued over :)... im excited.

Thanks,
Michael

Rosco Bodine - 24-10-2008 at 09:11

Quote:
Originally posted by saki2fifty
Well, i'd be more than interested in putting in the time and resources in building both circuits, testing, logging my work and construction, etc.

Great. I was hoping that maybe these postulated circuits would intrigue interest for someone having the time to actually lay 'em out on a board and see what they will do.
I know the DC logic is valid on the layout I proposed, but
what curves will be thrown by noise and AC components
which could adversely affect stability is the unknown for
either my design or Tims. I know I could "make it work"
purely by slowing down the response enough through various revision schemes, including using vactrol signal
buffering in the feedback loop . And mosfets are not essential for the output elements either, big NPN's could be used with different drive schemes. The DC logic for the
control loop being valid, goes a long way towards the solution....giving the front end controls, and then you
have a starting point for control of whatever ends up being the output stages that work.
Quote:

It would be great however if I could get a complete and modified schematic and possibly the cad file itself?

The cct file I have is for an old drawing program called DesignWorks which is a freebie. I haven't transferred the entire model in total to a SPICE simulation for the whole shebang, but just run sims on "blocks" , mainly the output module. That LTC1152 would substitute for the mosfet driver U5 and U6 in the corresponding parallel elements shown in that last pdf on page 1. The dual LT1218's are an active filter which provides the needed damping in the local feedback loop from the voltage across the current sensing shunt. That dual op-amp arrangement replaces U7 or U8 as shown on the last pdf page 1. I haven't redrawn the schematic to reflect the changes because I have to rescale the entire drawing to fit in the changes .....I ran out of room for making the changes without rescaling the whole damn drawing. It will take hours I don't have to redo the whole thing to fit the edits.
Quote:

If on Tim's solution there were exact part numbers of the IC's of choice, as well as all parts (I trust your judgment over mine), much like what Roscoe has, I can start on ordering the parts, building and testing.

You might be able to get free samples for parts from the
manufacturers if you are building a novel experimental circuit.
There was possibly a better choice of output Mosfet available from STM and they are the *only* company I found which has really looked at linear region operation of Mosfets.
Quote:

I'm not wanting to hack the power supply but rather build what you 2 guys have argued over :)... im excited.

Thanks,
Michael


It is actually a whole lot easier to hack an existing supply,
than to try to devise a more universal kind of solution .....
it's like the difference between minutes of hacking what's there, and months of abstract analysis and theory of what could be put there to do something analogous as an add-on.

The general concept of an adapter for an ATX might be implemented in a different way more easily in a conventional positive regulator configuration, and just go on and have
a local discrete 16 volt or so power supply for separately
powering the control elements, eliminating any noise
interaction from trying to power the control elements from
the same ATX supply as is being controlled. The scheme
of either Tim's design or mine is defintely not illustrative
of the "finest principles of quality engineering" but is more
a kind of what is the minimal thing which might work ......
rather than a more elaborate and conventional configuration
which could be better guaranteed to work....but at greater expense and number of parts. So you would correctly
view these configurations as some minimalistic afro-engineering of a "self powered adapter kind of circuit"
rather than either Tim's or mine "best work" which would involve a more sophisticated scheme. The parts expense
adds up fast as the quality of the thing is increased.
So if economy was no object, there are *definitely* better ways to do this.

12AX7 - 24-10-2008 at 11:03

Quote:
Originally posted by Rosco Bodine
The parts expense
adds up fast as the quality of the thing is increased.


Only if you are HP. ;)

We ain't HP, and we don't need their kind of specs, so we can make do quite well here.

For IC1, try LM837 or MC33079. I don't think an LM324 is really suitable here, but if you're feeling really cheap, you can. You'll need more compensation somewhere, which means the whole thing runs slower, which means more ripple and slower response time. Probably not an issue, but in principle you might as well build it better right?

And for IC2, a TL071 or LF411 is fine. If you do chose LM324 for IC1, you might as well use a (cough) LM741 here.

Tim

[Edited on 10-24-2008 by 12AX7]

saki2fifty - 27-10-2008 at 09:05



On lunch break at work...

Just laying out the parts on a perf board app in my spare time... but question regarding your "module", which Mosfet did you have in mind as well as alternatives to use? Also, you mentioned to place a cap on each op-amps output to -in... just want to make sure you were saying across the out / in of the actual IC and not the "module". What size cap?

Roscoe, can we possibly get the cad file uploaded so as not to redraw the shematic. No biggie, shouldn't be too hard. I did locate the sofware you were talking about...

Thx,
Michael

Rosco Bodine - 27-10-2008 at 10:41

Yeah , here is the cct file . I'll try to attach it here and if that doesn't work I'll put it on the forum FTP . I think I found the
cct file where I started to rescale it , but didn't do anything further on it. I can also upload the DesignWorks program
to the FTP if it is needed . It is really an easy to use entry level drawing program which is why I used it...but unfortunately it doesn't seem to export the drawing to more capable SPICE higher end programs like MultiSim ect.

I still don't fully trust the sims on this thing anyway, even on the output module alone.

Attachment: Converter for 550W ATX Power Supply to 50A Laboratory Power Supply.cct (46kB)
This file has been downloaded 1062 times

Rosco Bodine - 27-10-2008 at 10:44

And here is the one where I was rescaling the drawing to
provide room to fit in the larger LTC1152 output modules.

Also I ran across an ATX PS that seems to have some "heft"
to it and decent reviews , at a decent price . It might be good for this sort of project or just a good backup PS for
the usual labeled use.

http://www.newegg.com/Product/Product.aspx?Item=N82E16817101...
I chuckled when reading the reviews mentioning having to cut a hole in the top of the case for the computer to match the fan on the PS :P ....like Duh it would help if the genius
installing the PS would not install it upside down :D .....
You would think that the fourth screw hole not matching up
would be a clue there that the damn thing was flipped :D
Can just hear the guys old lady at night ...
saying "wrong hole" hehehe :o
Press any key .....but I looked everywhere and my keyboard doesn't have an "any" key ;)

[Edited on 27-10-2008 by Rosco Bodine]

Attachment: #2 Copy of Converter for 550W ATX Power Supply to 50A Laboratory Power Supply.cct (46kB)
This file has been downloaded 1137 times

Sedit - 22-7-2009 at 18:27

I was woundering if anyone here could help me out with a little something.

I just took the power supply out of my computer... Then put it back in so I could type this message:D j/k

Anyhoo I have not read this whole threed as of yet because it appeared to be turning into a willy waving contest but reading the page that Org/ posted gave me a rough Idea of what I need to do yet I am still having issues. My main question would be, is it normal that the power supply kicks off after a few seconds? I don't think its a bad supply as the computer stayed running right before I ripped it out but I feel that the load across one of the red and black is not enough because for now until I find/buy a suitable resistor I have a couple of small fans and a large coil of wire place across it.

Doing this, where as it appears to helped some, still does not stop it from kicking off after 20 seconds or less of run time.

The main thing that is throwing me off is after the power is completely cut the PSU fan comes back on in pulses so it appears a capacitor is resonating with something but Im unsure how the safety mechanics of the supply work in preventing operation without a large enough load.

Any suggestions?
And is this normal?

12AX7 - 22-7-2009 at 20:15

Sounds like overload to me.

Sedit - 22-7-2009 at 20:56

Thats kind of what I was thinking at first but I appeared to have magicly fixed my problem by placing the fan I was using as a small load across the blue and ground and feeding the cell with the red and ground. Since im not a big fan of magic in my chemistry im going to run more test on the various output wire tommorow with a multimeter and try to get a better idea of whats going on and if there is a spike or something when the other arrangment kicks out.

All and all though even though im learning it as I go it looks as though its going to shape up into a nice little setup considering I would like to be able to use the terminals im using for the load to power an overhead stirrer for the cell.




[edit]
Overload it is I believe. There must be some safety protection inside of the unit that forces shutdown in the case of an overload. I have to have a load of some sorts placed across either of the two main wires beit red/ground or blue/ground to prevent this. The Electrochemical cell is acting as one load while the cooling fan from another powersupply is acting as a second load across the other terminals.

I found this which may be of some interest to folks here and I apologise if its been posted already.

http://www.wikihow.com/Convert-a-Computer-ATX-Power-Supply-t...

[Edited on 23-7-2009 by Sedit]

Sedit - 24-7-2009 at 07:12

Well I once again am having problems with the PSU because even with the fan across it kicks off after about 10-15 minutes of run time.

Is this normal when one has an improper load or would I be best to try to use a different powersupply from another computer?

Im having alot of problems diagnosing the cause of the shutdown because my multimeters have on the best one a 15 amp fuse and Im not prepared to destroy my multimeter in order to make this thing.

All I have noticed is that when it gets to about 3v after a second or so is when it kicks off but this tells me next to nothing about whats really going on.

12AX7 - 24-7-2009 at 22:21

When I tried using an ATX supply on my cell, it appeared to work for a little while, but failed while I was away, probably as the cell heated up and current consumption rose from 20A towards 50A.

Shutdown is probably current controlled. These supplies usually have a current transformer monitoring what the transistors are switching. Too much for too long and it kicks off to protect itself. Much more and the transistors themselves cook, or some other part, like the filter coils or diodes (especially if just one output is heavily loaded).

Tim

Sedit - 25-7-2009 at 07:57

Thats what I was afraid to here because it seems as though im having the same issue. Its very frustrating because I read all these writeups and stuff telling how to do it, follow it to a Tee, and have it not work time and time again. Not only that but I think I killed my best supply. I have been tinkering with this for two or three days now with little to no improvement on its performance.

The wires I have are such,

Blue: +12v 8A
Red: + 5v 23A
Orange: - 5v .5A
Yellow: +12v .5A
Black: Grn
White: ??POK??

Thats what the white is labled and I don't know what this means so im doing nothing with it at the moment.

I have a 10ohm resistor placed across Red/Grn and Blue/Grn placed across the cell. This has so far provided the longest on times I have been able to get. Placing the Resistor across Blue/grn and using Red/Grn for the cell appears to work better but the resistor gets very hot very fast doing it this way.

Im under the same impression as you are that after a while to many amps start flowing and kicking on a safe gaurd kill switch. How would I go about preventing this? Perhaps a dimmer switch across the AC in to slow the power in the first place?


Its just frustating because I have heard of people using these for cells before yet as it stands now it does not seem possible.

The_Davster - 25-7-2009 at 08:42

This link http://web2.murraystate.edu/andy.batts/ps/powersupply.htm from the original post tells exactly waht needs to be hooked up to make a viable PSU. It has worked for me without fail on numerous computer power supplies.

12AX7 - 25-7-2009 at 09:37

Quote: Originally posted by Sedit  

The wires I have are such,

Blue: +12v 8A
Red: + 5v 23A
Orange: - 5v .5A
Yellow: +12v .5A
Black: Grn
White: ??POK??

Thats what the white is labled and I don't know what this means so im doing nothing with it at the moment.


Surely you have that backwards? Yellow is pretty standard as +12V, moderate amps. And surely you meant -12V 0.5A?
Quote:
I have a 10ohm resistor placed across Red/Grn and Blue/Grn placed across the cell. This has so far provided the longest on times I have been able to get. Placing the Resistor across Blue/grn and using Red/Grn for the cell appears to work better but the resistor gets very hot very fast doing it this way.


If blue is indeed delivering 12V (of whichever polarity), you would need a 14.4W resistor for that, so yes, it should get hot.
Quote:

Im under the same impression as you are that after a while to many amps start flowing and kicking on a safe gaurd kill switch. How would I go about preventing this? Perhaps a dimmer switch across the AC in to slow the power in the first place?


EWWW! Dimmer switch + electronics = let out all the magic smoke!

And besides, reducing the input *increases* the current drawn, since it's a regulating supply after all. But that's still beside the point.

Anyway, just use less electrode, or more distance between them. You might preheat the cell so it doesn't come up to operating temperature when you're away.

Tim

Sedit - 25-7-2009 at 09:52

Thats what im following along with some other writeups that state the same thing yet I am not having success at all. Im afraid its not that simple as there must be some other variable that that writeup is not accounting for such as the current that the cell itself is drawing.

POK(white) is the only wire I don't know what to do with. The on switch in mine is also different then stated in the link provided as its nothing more then a heavy duty white/black wires placed across the AC plug that must be jumpered in order to achieve startup.

As it stands now following everything to that writeup I can only achieve 10 seconds or less of power before it goes into shutdown.

[edit]

Yes I ment -12v sorry about that, No Im positive that the yellow is the -12.
Quote:
EWWW! Dimmer switch + electronics = let out all the magic smoke!

And besides, reducing the input *increases* the current drawn, since it's a regulating supply after all. But that's still beside the point.

Anyway, just use less electrode, or more distance between them. You might preheat the cell so it doesn't come up to operating temperature when you're away.

Tim


:D Yes I noticed it shuts down quicker when placed across the AC input.



[edit#2]

Ahhhh yess indeed Tim. After increasing the distance of the plates it appears to be working fine so cross my fingers and pray that it stays that way and life will be good:cool:

[Edited on 25-7-2009 by Sedit]

watson.fawkes - 25-7-2009 at 12:07

Not that this is immediately relevant, but computer PSU's have a rather restricted operating envelope because the loads they serve are rather well-behaved, as loads go. While the power draw from a computer isn't quite constant, it's still in a small window, almost purely resistive (power factor very close to 1), doesn't suddenly change by an order of magnitude, etc. There's always a minimum load (the CPU board), the maximum load is constrained by the number of devices in the box (barring failures), and what motors are present have minuscule startup torque. This is not at all the kind of load profile that a lab PSU sees: no load, shorted load, sudden changes, and generally all sorts of crap. So while you can press a computer PSU into service as a lab PSU, you've got to be careful with it.

I've half-thought about making a little adapter box that would add some of the basic protections needed to perform the computer -> lab conversion. It would have an ATX connector as an input, and provide minimum current draw and overvoltage protection in either crowbar (react by turning off) or current limiting (react by dropping voltage) mode. Crowbar is really easy to implement on an ATX supply because of the /PS_ON pin that allows software shutdown of the power. @12AX7: If you're considering alternate PSU projects, think about something like this rather than a full-blown supply.

Indeed, there may be some trickery available to coax more than 5.0 V out of the +5 V (nominal) supply by putting a programmable voltage reference across the sense return, but that's not broken out on the ATX connector and may not even by implemented in some PSU's. That, however, is not a product opportunity but a hacking project.

The_Davster - 25-7-2009 at 21:21

Sedit: brand of power supply you are playing with? some have odd color coding. Best is to get a motherboard ATX connector pinout, and connect what is necessary. ie red/black via 10W 10ohm, and appropriate wire to ground via a switch for power on.
Exactly what do you have hooked up to what currently?

I tend to think something is wrong with yours(or simpler and too much current is being pulled), I have abused at least 2 ATX power supplies, pulling at least 50% more amps than rated for days. The only failure is when much too much current is pulled and it turns off, easily fixed by a flicking off and on again. Well I also have had a failure when I pulled too many amps and burned all the insulation off the 5V red lines...:P

could this be relavent to your supply? from your link above
"# If you DO have a sense wire for the 3.3v. , connecting the the 3.3 v. part of the supply, using the 3.3v. voltage as a buck voltage against, say the 12v. to get 8.7v. will not work. You will see 8.7 v. with a volt meter but when you load that 8.7v. circuit the power supply may go into protective mode and shut the whole supply down.
# You can add a 3.3 volt output (such as to power 3V battery-powered devices) to the supply by hooking the orange wires to a post (making sure the brown wire remains connected to an orange wire) but beware that they share the same power output as the 5 volt, and thus you must not exceed the total power output of these two outputs."

POWERSUPPLY2.JPG - 48kB

[Edited on 26-7-09 by The_Davster]

Sedit - 25-7-2009 at 21:52

Very nice little setup you have there Davster I have high hopes of getting mine to look that nice soon enough.

I will get all the information tommorow on the brand of powersupply but after doing what Tim suggested of moving the electrodes in the cell itself further apart and playing around with how much of the electrodes are in the solution I managed to get it working perfectly(or so it appears for now:P). I would have more then likely been able to come to this conclusion myself if I was able to use my multimeter here but the 10amp fuse was highly limiting in what I could diagnose.

I have some pictures I will post tommorow as well so you can see for your self before I start locking everything in its place and drilling out for the grommets. Basicly I just have a 10ohm resister across the red and ground strapped to the fan pretty much how yours is setup there to keep it cool and am also currently running the cell off of red and ground as well. All other wires except the on switch are just loose at the moment and not doing anything one way or another.

I admit im not the greatest with electrical theory and stuff but I can still manage to make quite a few things using it without electrocuting myself...badly anyway(except that one time im lucky to have lived thru):D

dann2 - 27-7-2009 at 06:09

Hello,

Just another two electrons worth on the sudden shutting down of the supply at startup.
In order to get the supply to operate you must connect the green wire to the black (ground the green wire). If you were to do this with a switch, instead if permanently connecting green to black this may allow for a more gently startup/ramp-up of the supply????? (guess)
Instead of having the supply 'on' all the time and connecting the load to the output whereby you may be getting surges etc which may be causing shutdown problems.
Perhaps having the supply connected to the load (with green and black permanently connected together) and starting the thing up by pluging in the supply may be causing problems.
I have a supply there that I use to supply a small load (12V 3 amps very intermittent). I decided to disconnect the fan so that it would run more quiet. The supply will not start up with the fan disconnected. If I connect the fan, start up the supply and then disconnect the fan it runs OK.
These supplies seem to have their own little eccentricitys (and indeed their own little electricities too!).
Dann2

Sedit - 12-8-2009 at 07:48

I must make mention of something I haven't seen much talk about. You would think its common sence but apparently its not:D.

I got my power supply running nicely for me and it has so far served me well in running my magnetic stirrer when it needed a bit more muscle and running my cell with little to no problem.

The one issue I did have was I made the mistake of placing one of my power supplys to close to the cell and the sulfuric acid mist from the anode shorted one of the capacitors inside.

So even though you would think it would be common sence Im always here for ya to screw up what should be a simple process. Either way thought I would make mention so no one repeats my stupidity.

dann2 - 4-3-2010 at 13:08

Hello,
Since supplies are being discussed in another thread I thought I would bump this one as it may be useful.
A good add-on for a computer power supply is shown below. It gives constant current out which is usually what you want for electroplating/electrolysis etc. I have built it and it works well.
If you want more current that your (free) computer power supplies can give on their own you can use the cell to parallel them up as per the second diagram. It should work OK but I have not tried it myself.

Can speedometer and rev counters from modern cars (no mechanical cables coming to dash) be used as current or Voltage meters? They would make a great meter as they have 320 degreese of their faces in use as indicator space unlike cheap meters that have only (say) 80 degrees.

Dann2

ccs.jpg - 19kB paralleling.GIF - 7kB

[Edited on 4-3-2010 by dann2]

quicksilver - 4-3-2010 at 15:30

Dann2:


Have you tried this with a AT supply or just the ATX?

Are we going to use this one (thread) for the power supply stuff?

dann2 - 5-3-2010 at 03:26

Hello,
I have used the constant current (CC) add-on with an ATX supply. It will work with any type of Voltage supply. The size of the MOSFET and how hot it gets (which depends on the Voltage you are asking it to drop and the current you are asking it to pass) is the main limiting factor. Use a heatsink or/and a bigger device. [or/and use a few of the add-on's in parallel (if you cannot get bigger MOSFET's) as per the diagram in the post in this thread on 27-10-2008]
The resistor in the 12V line coming into the CC module is simply to drop some Votage so that the MOSFET does not have to do all the Voltage dropping and therefor heat up too much. The '5-Volts-out' option is a bit too low for Perchlorate cells and you cannot get a constant current into them as the max Voltage is too low. You need to switch over to the 12-Volt-max-output-Voltage option.
I have used a CPU heatsink with the small fan still attached to the heatsink to keep the MOSFET cool and it works well. The fan is powered from the 12 Volts from the computer PSU.

I have not tried parallelling computer suplies by using a number of independent Cathodes in the same cell but it should work OK, IMO (with or without the CC add-on). You could put a diode in series with each supply if you think that the supplies will interfer with each other but a diode will drop 0.7 Volts which is sometimes needed for running the cell etc.

Always use a switch between the Green and Black wires for switching the supply on and off. I find the supply 'behaves' much better when you do this, perhaps it's just me though.

Some info. here:
http://www.oxidizing.110mb.com/chlorate/cpsu.html


Dann2

[Edited on 5-3-2010 by dann2]

quicksilver - 5-3-2010 at 06:57

Gotcha! Makes sense.
I was noticing that (the resistor "helping out the load" from the MOSFET) and one of the reasons I asked is (like everything else) computer PSU got more cheaply made with time. I'm playing with an old one that has some of the better quality heat sinks I've seen in these and it's made in the USA. I have no "issue" with China but let's face it, they just don't turn out the top grade stuff that Japan or the EU/US does.
I've used the 5v on one that was just for fun as it was way too small -and I think the huge heat sink saved it.

quicksilver - 7-3-2010 at 11:41

Some one once asked why AC could not be used with a cell but thinking about that question made me think of an idea and that is reversing polarity in a pulse. For those who use carbon or graphite that MAY put a slow-down on the amount (level) of graphite crud in the chlorate material.

I actually don't think it's really worth the time IF someone has access to better quality anodes, etc. However it may (again, I'm just guessing) create a faster result. I don't know this is feasible to do given demands of current above 3-5a but whether it's worth the effort is another question.

I have used the graphite gouging rods after treating them w/ 2-5% HNO3 and linseed oil; that seems to make a difference in THEIR breakdown. I made a larger cell with 5 gallon bucket and the ones I'm using fit snugly in a 1/2" PVC coupling(s) epoxied to the lid. However I came across a picture of a home made PSu wherein the maker used solid state relays and a re-wound MOT & that made me think of pulse current feed on a variety of levels especially reversing polarity on two simple bars of graphite (or whatever; MMO...). The idea being not to put too much draw on one. The LD "plated" anode MAY have not broken down if it "shared" it's load with a twin.

dann2 - 7-3-2010 at 14:58

Hello,

I think the best waveform for these cells is just plain old DC.
If you reverse the current on a 50/50 basis then both Anodes will last twice as long but two of them are now wore out (6 of one, half a dozen of the the other).
The only place I know where current reversal may be useful is making Chlorate and Perchlorates where hydroxide is inclined to get formed on the Cathode and stop current from flowing (like, say, Magnesium Chlorate or Perchlorate). A spinning Cathode has been used to stop this problem occuring. Handier to reverse current every few minutes and that will do the same thing.............I think.

You cannot use Lead Dioxide as a Cathode as it will dissolve quickly. You would have to use Platinum for making Perchlorate with the current reversal scheme.

Dann2

quicksilver - 8-3-2010 at 08:00

I understand what you're saying (it IS a 6 of 1 - 1/2 dozen of the other situation). When I have seen the formation of salts on the cathode it's been when Ive used a shape that would hold them (screen-type). The less expensive route of using gouging rods (given their cylinder shape appears to minimize this (however more time needs to devoted to that simple observation to determine if it really has validity.

I recognize that the continued complexity of the cell may have minimal returns with more expensive materials (MMO, etc) Originally I thought that salts building up were a problem of the current dropping from computer PSU sensing differing conductivity as time went on.
 Pages:  1  2