Variable PC Switching power supply?

Is there a (simple) way to transform an AT or ATX power supply into a variable one? I mean instead of the fixed 5v and 12v, to be able to go all the way to these values.

Thanks

Hi there Pablo,

Might be a simple way if you can get a schematic.
If not, no simple way :-( except to add external control,
which may not be that hard really.

Take care,
Al

Hi Al,
I don't have schematics, just a bunch of unused PC power supplies /ubbthreads/images/graemlins/smile.gif

Adding an external control is as simple as adding a regulator (like lm317) to the output?


Pablo

It should be possible to replace the fixed sense resistor with a variable one, or maybe a voltage divider. The problem is that you need to locate the sense resistor, which is why I would want the schematic. There are probably hundreds (dozens for sure!) of different variations of these power supplies. There are generic AT or ATX power supply design modifications (mainly for ham radio) on the web that might help you find the sense resistor. A lot of applications require that the output be adjusted to 13.8 or maybe up to 15v, so locating the resistor is a common problem.

Hey PEU,
Your right on this one, PC AT/ATX power supply is cheap and commonly available. I've a 250watts AT(easier to operate) power supply with all the cables removed. At the back, I added a 240volts illuminated rocker switch which switches the power supply on and off.

First of all, the power supply is hazardous so be careful when working with it. Discharge the large capacitors, using some mains light bulbs to avoid unnecessary electric shock. Most of the capacitors in the power supply should discharge by themselves but please take precaution.

There's a lot of free space inside the power supply but make sure that you do not interfere with the existing circuit and components when adding your variable LMxxx circuits. I use two LM350 voltage regulators for dual voltage output. It has a continuous output current of 3amps. If you want to go for higher current output, use the LM338. It has 5amps continuous current and 12amps peak. My 250watt power supply can handle 8amps at 12volts so it should be fine. The side of the power supply casing is hacked with a window for a mounting a massive heat sink for the regulators /ubbthreads/images/graemlins/smile.gif. On the front of the power supply mounts the potentiometers and banana binding posts. You can even go funky with leds and a voltage meter display would be handy.

The limitation of this setup is that I can only get around 9volts max because of the voltage drop in the regulator. If I need 12volts then I have to bypass the voltage regulator circuit. RussH mention about replacing the fixed sense resistor with a variable one could be a good solution but you have to know the limitation of your power supply. Some ATX power supply requires some minimum load to work properly, and AT power supply is easier to switch on. Use plenty of fuses for your regulator input or the output, they are cheap so no excuses, better be safe than sorry. PC power supply is not designed for this job so keep that in mind.

Aside the limitations this is the ultimate and cheap way of building a variable voltage power supply with high current output.

PC power supplies have a bunch of feedback points and limit sensors. You can change the 5V rail by no more than 0.5V each way before "under voltage shutoff" and "over voltage shutoff" kicks in. There isn't all that much you can do unless you want to take care of all of them /ubbthreads/images/graemlins/frown.gif

Hi Folks,

Very interesting replies.

PEU:
Yes, i meant just that -- adding perhaps an LM317 to the
output with a pot for the lower resistor to adjust voltage
and maybe a series resistor for current limit if needed.
You'll get up to almost 10v output if you run off of
the 12 volt supply, maybe 9v with high current output.
Worth a try is running from the -5 volt supply (if it
has enough current rating) to the +12 volt supply for
a 17 volt input. Chances are however that the -5 volt
supply doesnt handle that much current.
The case might work good enough as a heat sink for the
317 too with added insulating hardware.

I wouldnt take the chance of trying to modify the PS
unless i had a schematic or felt like tracing out
the whole circuit. I've got a few of those PS's around
too, and dont feel like tracing any of the circuits out
just to make them variable, and if the circuit is too
dedicated it may be very difficult to do this anyway.

Let me know what you would like to build and i'll post
a schematic.

Take care,
Al

I already have a variable power supply, but I want a more portable not so heavy one from 0 to 12 (10 after the LM regulator) I already have all the components: at supply, regulator, female plugs, a nice case and lotsa dissipators /ubbthreads/images/graemlins/smile.gif from old pc processors and destroyed AT supplies, so when time allows I'll build the kludge and post some pics.

Pablo

Oh ok that would be nice

Take care,
Al

I've got a 'snake that swallowed a rat' supplies from a laptop I'm going to do this with. 18 VDC IIRC.

I've always wanted to try using the PC supply negative supply (either five or fifteen) to return the 317 divider to so you can dial the supply down to zero volts minimum. Anyone done it? Either a resistive divider or a 337 set to minimum might do the trick.

Doug Owen

Doug,

I've used the negative voltage rail (-5V) before with the +12V with a 317 - it seems to work, but I think typically they're limited to about 1A (0.5A in some cases)

[ QUOTE ]
evan9162 said:
Doug,

I've used the negative voltage rail (-5V) before with the +12V with a 317 - it seems to work, but I think typically they're limited to about 1A (0.5A in some cases)

[/ QUOTE ]

I've done this as well, returned the *load* to minus five (within the limits of minus five current). What I'm suggesting here is returning the load to ground but the regulator reference to a minus voltage (minus 1.25) so we can dial the output down to zero volts. Since this reference and bias current is quite low, the supply should work fine up to the full current availability of the plus 12 I would think?

Doug Owen

Ah yes.

Let me see here...

I've build the "Adjustable Current Regulator" circuit on page 10 of the National LM350 Datasheet (uses an LM350 and LM317) and used the -5V line as a reference in that case - the power supply didn't seem to have any issues doing that.

another question (I'm in the building process /ubbthreads/images/graemlins/smile.gif ) will a MOV protect the output from a short circuit? or a short will blow the lm350 and/or the pc supply?

Pablo

The LM350 should be current and thermal protected. It'll limit output current to around 3-4A, and will shutdown if it gets too hot.

Lots of PC power supplies will also shutdown if they detect a short - what I've seen is that you must power the supply down before it recovers. I'll bet that not all are built this way, and could possibly blow if they are shorted out.

I think that MOVs typically protect against overvoltage - they start to conduct after a certian voltage is reached, so you would attach them to the input of a device to protect the device from overvoltage.

ohh thanks evan, Im using an AT supply from an old siemens nixdorf PC, I see it has some kind of thermal protection in one of the dissipators. will do, will tell /ubbthreads/images/graemlins/smile.gif Thanks

Pablo

There is a diode shown in the app notes for the 317 type regulators to protect from reverse polarity (actually 2 protection diodes, one on input, one output). Otherwise the built in short or thermal protection should be enough. I slapped together a boat battery charger with a 317 regulator at a friends house using a 2 amp wall wart. It would thermal limit to a little over 1 amp after running for 20 minutes or so into a marine 12v battery. My friend used it nearly continuously for several months before the chip gave up.

One other solution for higher voltages and currents is to use 3 +5v supplies in series. This would be too bulky for some but gets you about 30 amps or so at 13.8-16v, depending on adjustment. Note that a lot of the older supplies will not come on without an amp or so load on the +5v. MPJA had some good prices on switching supplies the last time I looked.

the one I choosed had 6.3A on the 5v and 1.5A on the 12v line, its an old one...

I'll use 2 lm350 for the variable output.

An hour ago I purchased the elements I did not have at hand, so its going to be finished tomorrow or monday for sure.

Total cost of purchased parts $11, and that includes a cheap tester to measure output (like the one discused in a separate thread), the remaining was lying around in my office including a very nice cabinet...

Pablo

Hi there Pablo,

Wow, that's prety good too. Those 350's go up to 3 or
more amps i think... what are you going to be testing
if you dont mind me asking?
Normal LM317's go for about 60 cents and put out about
1.5 amps.

I've been casually working on a low dropout version of
a LM317 but it requires a separate bias supply...one that
can either be a small wall wart or a tiny switcher/boost.
Puts out 13.8 volts with a 12.0 volt input!

Take care,
Al

Nothing in particular Al, but I wanted to do this long time ago, the variable power supply that I have have 2 very heavy power transformers and is bulky to say the least, but it works great to say something positive /ubbthreads/images/graemlins/smile.gif

I had many spare and unused LM350, when I went to electronics parts street (yes there are 2 blocks full of electronics components/parts/etc stores) I asked for a LM338 to take full advantage of the 6.5A part of the supply and they asked $5 for a single unit, so I declined to use it.

I'm staying home, so its going to be build next monday at the office.

Pablo