Sciencemadness Discussion Board
Not logged in [Login ]
Go To Bottom

Printable Version  
 Pages:  1  2    4    6  ..  9
Author: Subject: PC PSU to laboratory PSU
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 25-10-2007 at 17:17


@Xenoid

The datasheet says something entirely different from that note #2 on the schematic you posted about functions for
pins 2 and 3 ....
it isn't pins 2 and 3 that are the error amp inputs but pins 1 and 2 instead ......pin 3 is for frequency compensation .

Soooo you should be able to change the biasing as I suggested at either pin 2 or pin 1 with the same effect .
As the better choice I think go with what I said first above .

BTW the internal reference is 5 v which is actually an
internal regulator providing +5V out at pin 14 , for local biasing needs ect. That is what I was guessing when
I made the first suggestion , as I didn't believe then
what the schematic said about pin 3 , and have confirmed that now .

The stated values shown on the schematic don't look
correct for the biasing at pin 1 , so I am trusting the data sheet and intuition here ...for already having caught that schematic in one blunder and expecting more :P

For example R40 should equal R41 corresponding with
R38 equaling R42 .......in order to set the operating point
properly for 5 volts . But the schematic shows R40 as 10K
instead of the 3.9K it should be :D Also the same for R39
which is wrong at 24K and should be 18.7K .......there's two good reasons I don't trust that schematic additional to the
discrepancy in note#2 about the pinout .

Wait a second ..... I see now what they are doing there , combining 12 and 5 to get a parallel sum at the 3.9K .....
a 2.5 volt derivative from the combined 12 and 5 signals .
It's okay there , they are just shortcutting having independent monitoring and regulation for the separate
12 and 5 ....must be an older design supply .

You might want to disconnect the 24K R39 and replace the
R40 10K with a 3.9K if you are going for variable voltage of the 5 volt output only .......that will give you a dedicated more accurate regulation on the varied 5v output you are using , while letting the 12 volt output go where it will .

[Edited on 25-10-2007 by Rosco Bodine]
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 25-10-2007 at 17:46


@Rosco

Yeah, I remember seeing that Note #2, I should have mentioned it!
I also don't understand those bias resistors around pin 1.
Have you managed to have a look at the pencil sketch circuit, I mentioned above, note they are adjusting the 12V supply not the 5v.
I have the power supply the circuit refers to, I'll check the circuit.

Edit: The resistor values around pin 1 are correct according to my circuit board. My supply uses a Sharp IR3M02 which is pin compatible with the TL494.

Regards, Xenoid

[Edited on 25-10-2007 by Xenoid]

[Edited on 25-10-2007 by Xenoid]
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 25-10-2007 at 18:07


I figured out what is up with pin 1 and edited my message above .

I think you need to separate your feedback for 5 and 12
and just use one or the other ...for accurate regulation
of the one you are using and varying . That could be a problem depending on how stable or wild the one freed up will be . Lightly loaded , I'm thinking the freed one
should be okay ....but can't be certain .

Where does that "power good" output go ...to an LED ?
If it loops back into some sort of power on self test , it
could be trouble and might have to go .

[Edited on 25-10-2007 by Rosco Bodine]
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 25-10-2007 at 18:28


@Rosco

Sorry to keep bugging you!

Why is it necessary to have feedback from both the 5 AND 12 volt outputs. Since they come off the same transformer. If the TL494 ramps up to correct the 5V supply, wont the 12V supply be increased accordingly and vice versa! Thats why I was worried about the 12V auxilliary supply going too low or high when the 5V is changed from say 3-7 volts.

Am I missing something here, I might be getting out of my electronics depth with SMPSs.

Edit:....."Where does that "power good" output go ...to an LED ?
If it loops back into some sort of power on self test , it
could be trouble and might have to go."
It went to the MB as far as I know, it's disconnected, the supply works fine!

Regards, Xenoid

[Edited on 25-10-2007 by Xenoid]
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 25-10-2007 at 18:40


It's just a corner cutting way of regulating two outputs at once ......not really true regulation of 5 and 12 independently , but a "presumptive" sort of limiting regulation of two outputs at once .....figuring that more or less one follows the other since they are running from the same oscillator .

Really what I get from this now is you should use the 2.2K
fixed resistor in series with a 5K pot like I said earlier ,
but use the series of those two as a substitute for R40 and just remove R39 . Set your initial series resistance
for 3.9K sum 2.2K + 1.7K "in the pot" . That will be your
nominal 5 volt out and you can adjust it up or down from there .

Allowing the 12 volt output to float , put a light load on it
to help it not "float away" :P

And you will have to disconnect the overvoltage sensing
to pin 7 on the LM339 . Just tie pin 7 to ground .

Leave the lead that went to it open and just cap it off .

I think you will be in business then .

But maybe keep a fire extinguisher handy , just in case:P

BTW here are some output ranges for different value
fixed resistors in series with the 5K pot .

Using a 2.2K .... 3.9v to 7.1v
Using a 1.8K .... 3.65v to 6.86v
Using a 1.5K .... 3.46v to 6.7v
Using a 1K ........3.14v to 6.35v
Using a 680 ......2.94v to 6.14v


[Edited on 25-10-2007 by Rosco Bodine]
View user's profile View All Posts By User
12AX7
Post Harlot
*****




Posts: 4803
Registered: 8-3-2005
Location: oscillating
Member Is Offline

Mood: informative

[*] posted on 25-10-2007 at 20:29


On that schematic, to make it adjustable;

Check the rating of C16 and C17. They must be 10V rated for the range you suggest.

Nix the +12V winding. No need to cut it all off, but do remove the filter capacitors and ZD1. Optionally, you can also remove the rectifier, which will still cause a small amount of loss, even though unloaded. (Move D11 accordingly, of course.) You still need the fan, so move it to the -12V supply. You could keep the +12V supply for this reason, but it's much more beefy than you need.

For control, nix R39 and change R40/R41 to a 10k potentiometer. Add a resistor in series with its GND leg so you can't adjust voltage to +infinity (which would be useless, as PWM would max out to no benefit -- you lose active regulation and get some unspecified output voltage). Don't muck with the 2.5V reference (R38, R42). R43 and C26 need a constant impedance to function properly.

On overvoltage, you can cut it altogether (nix R22, R19 and D14) and ground it (for example, now that the top end of R23 is loose, it could be grounded). Don't muck with the voltage divider (R32, R33), because you do want current limit.

As you adjust the input and output voltage and vary the load current, the PWM ratio changes to compensate. This circuit works by chomping a variable width "up" pulse, followed by dead time (Q1, Q2 off), followed by an identical "down" pulse and another dead time. These two pulses are flipped into forward-going and off periods by the rectifiers, and this doubled-up PWM is filtered by the choke. The percent on-time therefore corresponds to the average, which is filtered and given to you at the output. As line voltage changes, the voltage on the transformer changes, so the output would change as well. This must be compensated for with a proportional change in PWM. Resistance in the circuit (power line sag, the transistors, transformer, rectifier, choke) causes some voltage drop with increasing load current, requiring somewhat less PWM change to compensate. Since PWM is limited (up to 48% or so, depending on C27 -- see datasheet), there is only so much compensation to apply in this circuit. As such, you can only adjust the output voltage so high, given some load current and line voltage.

The practical lower limit is where PWM becomes so short that it's squegging (runt pulses, only enough to maintain voltage). This is probably at few volts. You cannot draw a constant power limit from this type of power supply: you are limited by the current capacity of the circuit. The SOA is roughly square.

Speaking of SOA, do keep in mind that this power supply was NOT designed to put all its VAs into one winding. You WILL need a beefier filter choke (MWC and L5), rectifier and power transformer secondary (ooh, adventerous this one) to pull that off.

Tim




Seven Transistor Labs LLC http://seventransistorlabs.com/
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
View user's profile Visit user's homepage View All Posts By User This user has MSN Messenger
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 25-10-2007 at 21:08


Thanks guys;

I've just had a quick go at modifying;

I removed R39 (12v feedback) and simply replaced R40 with a 10K pot, as I didn't happen to have a 5K. I disconnected the overvoltage on the LM339 and grounded pin 7.

I connected a small load (10 ohm, 20W resistor) and powered up, I was able to control the voltage from about 2.5V up to about 9.5V which is about what I expected.

The fan stops running at about 3.8V as the auxillary supply drops, and really roars away when turned up to 9.5V. I guess the 12 volt auxillary is changing from about 6V to 22V. I didn't get around to measuring it. I think it will need a seperate supply!

Regards, Xenoid
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 25-10-2007 at 21:23


The current limiter is still operational and works from sensing on the primary . You can make the current limiter adjustable by substituting a 5K pot for R33 . The supply should still deliver its rated current on the 5v output .

I disagree with cutting off the 12 volt winding and moving the fan to the -12 . I would leave that as is .

I disagree with using a 10K pot like 12AX7 is saying , going to the node pin 1 and R41 .....just pick one of the combinations of fixed value R and a 5K pot as I have charted above , substitute that series combination in the place of R40 and remove both R40 and R39 .

I do agree removing ZD1 and additionally removing R19 , R22 , R23 . They become parasites on the new configuration .

You could probably use a power resistor and a 12V 5W zener
to protect your fan or use a cheapie fixed regulator in front of it .

That's strange with the voltage going to 9.5 with the 10K pot
as it should've stopped at 8.9v . Must be some wide tolerance resistors on that board .

[Edited on 26-10-2007 by Rosco Bodine]
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 25-10-2007 at 22:28


I can't see any way around not having a separate supply for the fan.

It's no good if it stops operating when the voltage is turned down below 3.8 volts. My chlorate cells tend to run at about 3.0 - 3.5 Volts and 20 Amps. At 20A the fan will need to be going flat out.

I guess I could use a mains voltage fan, if I can find one!

By the way, this Seventeam circuit is pretty generic, and many of the early AT supplies which utilise the TL494 and the LM339 (or equivalents) have essentially the same circuit.

Regards, Xenoid
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 25-10-2007 at 23:03


If the -5 output has enough current capability
for running the fan......

Perhaps move the negative for the fan to the -5 output
and then use a power resistor and 12v 5W Zener
across the motor leads to limit it .

At 60% output , 3 volts on the usual 5 volt , you would see
-3 to +7.2 or 10.2v for the fan as minimum .....right ?

And then as you crank it up ....the power resistor would burn off anything over 12 .
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 25-10-2007 at 23:25


Yes Rosco, that sounds reasonable. I'll try it tomorrow.

Actually the fan has an added on thermistor control board which is not shown on the circuit. I'll try removing it, and check exactly what the fan requirements are.

I'll also check the current limiting idea as well.

The recycling centre has a large bin of AT type supplies, they cost about $2 and they all seem to work OK. Even if they don't, there's more than $2 worth of parts inside - incredible value!

Regards, Xenoid
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 26-10-2007 at 14:01


Not meaning to too much overstate the obvious here

* If * there is a good schematic available for any common and even higher current SMPS of the more recent ATX specification ....
a similar conversion could possibly be done there as well,
by way of this kind of unauthorized reverse engineering .

warning ! no user servicable components inside !
warranty void if seal is broken ! :P ROFLMAO :D:D:D:D

If this power supply was sentient , and it sensed your
approach with diagonal cutters , soldering iron , test meter , magnifying work glasses and a little bag of parts ..... just imagine the poor things harness grommet
would be constricted so tightly it would shear the cables
in two :D

Now just hold still Mr. PS ...this won't hurt a bit :D
Just making a little adjustment here .

Any of you fellow nerds talk to the hardware on which you are working as if it can hear you ? Hehehe , the doctor will see you now :P
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 26-10-2007 at 21:12


@ Rosco

Here's an ATX circuit:

http://www.pavouk.org/hw/en_atxps.html

I notice the "auxiliary" power is driven by a separate transistor oscillator, transformer and a 7805 "second power supply". I guess this part of the circuit could be used to drive the fan as it is independent of whatever output voltage changes one might want to make.

I have an ATX, it has 5V at 24A and 3.3V at 14A but I'm only using the 5V at the moment. To get it going all that was required was grounding the PS-ON wire, if I remember correctly. I put a small load across the output to insure it started OK when not connected to anything. It is rated at 230W but the combined output of +5V and +3.3V is only 130W. I guess by ATX, you're thinking of the more "modern" 300-400W supplies. I haven't seen any of these at the recyclers yet!

This might be a better proposition to make adjustable as I haven't drilled holes in the front panel for a switch (it has a switch on the rear) and an indicator light, so there is still room for a pot or two!

I notice the LM339 in this circuit is not used for the over voltage/current protection.

There are several web sites for modifying these supplies, they are usually modified to produce +13.5V for "ham" radio operators, like here.

http://www.webx.dk/oz2cpu/radios/psu-pc1.htm

Regards, Xenoid
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 26-10-2007 at 22:38


Actually I was thinking about those ( Broadway Com Corp) Okia branded ATX supplies that have the 50A 5v winding , and can be gotten for about twelve bucks and change shipped :D

http://www.bccpc.com/bccpc/power_all.htm

Two of the 450 series are very interesting also as two of them have an additional auxillary *high current -12v* output .

That makes a 15A @ 24v available for servo motors ,
or 15A @ 15.3v available for battery charging .....two things that any usual ATX can't do .
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 26-10-2007 at 22:49


Wow!

No, I haven't seen any at the recycling centre....:o

Regards, Xenoid
View user's profile View All Posts By User
dann2
International Hazard
*****




Posts: 1523
Registered: 31-1-2007
Member Is Offline

Mood: No Mood

[*] posted on 8-11-2007 at 17:53


Hello folks,

Would I be right in presuming that if one were to connect a variac to the input of a PC PSU that this would not work at all for varying the output?

Dann2
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 8-11-2007 at 18:05


Quote:
Originally posted by dann2
Hello folks,

Would I be right in presuming that if one were to connect a variac to the input of a PC PSU that this would not work at all for varying the output?

Dann2


Yes you would be right. Computer SMPSs use an internally generated reference voltage, and compare this to feedback from the output. They can however be fooled, that is what the previous few posts were all about. Just recently I did a simple modification to vary the +5 volt output from 4 volts to about 7.5 volts. This is a really usefull range for a perchlorate cell. I was running my perchlorate cell at about 5.5 volts. Unfortunately the fan stopped working during the night (I think) and there was a melt down of the main driver transistors and a massive short circuit. It was not the fault of the modification, however. I plan to get a couple more from the recycling centre and modify them also, the PSUs only cost a couple of dollars. The one that self destructed is not worth repairing!

Edit: Posted without actually answering the question, whoops!

Regards, Xenoid

[Edited on 8-11-2007 by Xenoid]
View user's profile View All Posts By User
Twospoons
International Hazard
*****




Posts: 1201
Registered: 26-7-2004
Location: Middle Earth
Member Is Offline

Mood: A trace of hope...

[*] posted on 8-11-2007 at 19:20


If you do have a variac, it will work wonderfully for adjusting the output of an ordinary transformer. I have an 8 amp one that I use for controlling the heater in my alcohol still.



Helicopter: "helico" -> spiral, "pter" -> with wings
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 8-11-2007 at 20:05


BTW , I found another cheap ATX supply with a 50A rated 5 v output . That earlier $12 one I posted has gotten some poor reviews . This one is higher and better rated ,
but costs $20 shipped . The specs on the dealer page are wrong , but the 50A rating is on the label and on the manufacturers website . The brand name for these
would most often be "Hercules" for those that are badged
with a brand name .

http://www.eaglebit.com/ProductDetails.asp?ProductCode=EB-46...

Also I got another mil surplus bargain on an obsolete but functional linear lab power supply 120 Amps , 0-12 volts
for under fifty dollars :D The thing weighs a hundred pounds , has a triac pre-regulation on the power transformer to minimize dissipation on the eight parallel
TO-3 NPN pass element array which is on one huge cast and machined aluminum heatsink with a five inch fan blowing across it . The machined heatsink alone looks like a several hundred dollar item . Redundant breakers and thermal protection as well as active current mode backed up by settable crowbar trips , and coarse and fine detented control pots , 3 inch meters , and neon status indicators . Did I say yippeee :P
View user's profile View All Posts By User
Xenoid
International Hazard
*****




Posts: 775
Registered: 14-6-2007
Location: Springs Junction, New Zealand
Member Is Offline

Mood: Comfortably Numb

[*] posted on 17-11-2007 at 21:26


Looks like a "slack" time for posts, so here's a little contribution.

A year or so ago, I picked up a large transformer from the recycling centre. It's a 500VA / 12 Volt / 48 Amp halogen lighting transformer. It is designed for home lighting (down lights, recessed lights etc.) and has a built in thermal cut out. The transformer was in perfect condition and housed in an unpainted open-ended mild steel box, with terminal connectors at each end. I guess it cost me about $5 - $10 as I usually never pay more than that at the recycling centres.

Well, I have finally got around to doing something with it, I'm going to use it to power a fairly large (for me) chlorate cell at 24 -36 Amps. I've spent a week (it was going to be just a day!) dressing it up. I fitted a ventilated panel, fuse and power cord to the back. A couple of chrome plated handles I had lying around went on top. A front panel was made from PVC and fitted with a switch and neon. I have used two bolts for heavy duty terminals.

Two BR354 35 Amp bridge rectifiers are mounted internally and wired in parallel. I added a pair of "home made" aluminium heatsinks (made from some scrap aluminium channel and sprayed matt black) as extra cooling. The whole unit was sprayed in black satin paint (my favourite electronics colour). The extras probably cost about $15 - $20 with most of this being the rectifiers ~$5 each, most of the stuff was lying around the workshop.

I'll be running it of my 500VA Variac for current control, when I hook it up to the cell.

Regards, Xenoid

PowerSupply.jpg - 42kB
View user's profile View All Posts By User
12AX7
Post Harlot
*****




Posts: 4803
Registered: 8-3-2005
Location: oscillating
Member Is Offline

Mood: informative

[*] posted on 18-11-2007 at 10:18


Running a couple in series I presume? 12V is good for three or four at typical voltages.

Tim




Seven Transistor Labs LLC http://seventransistorlabs.com/
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
View user's profile Visit user's homepage View All Posts By User This user has MSN Messenger
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 18-11-2007 at 11:13


That was actually my same *first* idea for an improvised power supply . And the components I chose are virtually identical , a medium sized low voltage (12v,40A) landscape
lighting transformer , also using two 35A bridge rectifiers paralleled , and a variac controlling the primary voltage for the transformer . Additionally I have a four capacitor filter bank , and I am contemplating rewiring an MOT core as an inductor to add there for LC smoothing of any ripple ...for whatever little good that may do without using absolutely huge LC values . But it would be highly power efficient ,
with the biggest loss just being the drop across the bridge rectifier . And with proper breakers / fuses for protection , there isn't anything there to break , so the thing should last just about forever , cranking out plenty of amps , but pretty much having the 1 or more pounds per amp capacity for linear supplies that seems unavoidable .
Nothing dainty or lightweight about the cores on these things .

I still have a couple of parts to scrounge before I can put mine together , but it basically follows precisely what Xenoid has there . And it can always be used as the front end for added solid state regulation . The variac can be
ganged with the control pot for the solid state regulation
so that the input is just enough to account for the drop across the solid state regulator , to minimize dissipation .
It is the same idea as using a phase control to the transformer primary for preregualtion , but of course a variac is a better and lower noise , though more expensive method of preregualtion . Variacs still rule for this sort of application , but add more weight of course to a linear .

The high current linears that are commercially manufactured are ungodly expensive , so any improvisation is sure worth consideration . I have a
50A Power Mate lab supply that is nearly thirty years old
and it was list price around $1500 *then* at the time of manufacture . And my other Hewlett Packard 120A lab supply of the same vintage , was listed price of $3800 *then* , so it's for sure these linears aren't cheaper now in todays dollars but probably three times as much now .

I've had a few used cars that cost less than these damn things :D So I too am a junkyard and mil surplus scrounger :P as it would seem is the case for Xenoid .
View user's profile View All Posts By User
dann2
International Hazard
*****




Posts: 1523
Registered: 31-1-2007
Member Is Offline

Mood: No Mood

[*] posted on 19-11-2007 at 11:48


Hello,

@ Xenoid:
You made a right dog's dinner of that transformer + rec. The whole thing looks much more exciting (and sophisticated) when spralled out on the bench in 'jackdaws nest' configuration.:D

About putting cells in series there is a diagram in the Kirk Othmer chem. encyclopedias (in reference section) showing anodes and cathodes all in the one container/cell. They are sortof stacked.
Anode then cathode then anode then cathode etc with just two (+ & -) connections going to cell.

I could post the pdf but I have just realized I am at a different pc.

Dann2
View user's profile View All Posts By User
12AX7
Post Harlot
*****




Posts: 4803
Registered: 8-3-2005
Location: oscillating
Member Is Offline

Mood: informative

[*] posted on 19-11-2007 at 14:41


Note that the anodes or cathodes need to be insulated around the edges, resulting in independent cells.

Tim




Seven Transistor Labs LLC http://seventransistorlabs.com/
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
View user's profile Visit user's homepage View All Posts By User This user has MSN Messenger
dann2
International Hazard
*****




Posts: 1523
Registered: 31-1-2007
Member Is Offline

Mood: No Mood

[*] posted on 19-11-2007 at 19:33


Hello,

I had not got my reading glasses on when looking at the cell diagram. There is a dashed line down through cell on page 112 of attached document. In effect, two (or more), independent cells as Tim said.



Dann2

Attachment: chlomend.a01.pdf (161kB)
This file has been downloaded 1046 times

View user's profile View All Posts By User
 Pages:  1  2    4    6  ..  9

  Go To Top