It's time to build a dual power supply for bench top experiments. I have attempted to use an old computer power supply in the past but have found that it requires some magical input to tell the power supply to turn on. How do I figure out which wire(s) need to be switched on? Or perhaps fed with some frequency to switch it on - or whatever the case may be.

And I'd like to turn it into a dual supply providing 12V+ to ground and 12V- to ground. I know there's a 5V circuit, and maybe that can be converted into a 5V± supply as well.

It's that or scrap a transformer out of an old stereo and build a power regulator circuit for that with the ± supply.

At present I own a 13.8 VDC, 9A supply but it's single ended, not ±. Any ideas?

 

Go to Amazon and search for "atx breakout"

 

Typically to turn on a pc power supply you need to short the green wire to ground (any black wire) You'll probably find that the -12 is only good for a fraction of an amp. Here you go.. http://pinouts.ru/Power/atxpower_pinout.shtml

 

short the green wire to ground

Thanks. Gosh that was so easy. Much appreciated.

And my power outputs (if anyone wishes to know) are:
+3.3 V @ 17 A
+5 V @ 22 A
+12 V @ 18 A
-12 v @ 1 A

As long as I don't ask for more than one amp from the negative 12 volts I should be fine.

I've been trying to figure out Op Amps. Have read LOTS of material on them and most say you don't need the negative rail. However, all my experiments with just 12 volts positive have proven to be a failure. I'm SURE I'm doing something wrong, so maybe with a double rail supply I can figure out either what I'm doing wrong OR what works.

A project I've been working on - the amplification of a pre-amplified source - has been at the root of my attempts. I'm hoping to get this going soon. Maybe now I can learn a little more about Op Amps and how to make them work in my application. But my application is the subject of another thread. When my project is complete I hope to video the outcome and share it with all.

Thanks for the quick replies; I consider the matter closed. Is there any way to close my thread?

 

Go to Amazon and search for "atx breakout"

AMAZON? All I got was stuff to buy. Oh well, I got my answer. Thanks all again.

 

Thanks. Gosh that was so easy. Much appreciated.

And my power outputs (if anyone wishes to know) are:
+3.3 V @ 17 A
+5 V @ 22 A
+12 V @ 18 A
-12 v @ 1 A

As long as I don't ask for more than one amp from the negative 12 volts I should be fine.

I've been trying to figure out Op Amps. Have read LOTS of material on them and most say you don't need the negative rail. However, all my experiments with just 12 volts positive have proven to be a failure. I'm SURE I'm doing something wrong, so maybe with a double rail supply I can figure out either what I'm doing wrong OR what works.

A project I've been working on - the amplification of a pre-amplified source - has been at the root of my attempts. I'm hoping to get this going soon. Maybe now I can learn a little more about Op Amps and how to make them work in my application. But my application is the subject of another thread. When my project is complete I hope to video the outcome and share it with all.

Thanks for the quick replies; I consider the matter closed. Is there any way to close my thread?

Usually you have to draw current from the +5 rail and/or the 3.3V to get the other outputs to work. Usually, those 2 rails are sampled to control the PWM - if those rails aren't loaded; the PWM idles and the 12V rails may not even come up to spec.

There have been various projects in Elektor & EPE, some breakout types and some for more advanced constructors who might tackle rewinding the chopper transformer.

 

Usually you have to draw current from the +5 rail and/or the 3.3V to get the other outputs to work.

So you think I should load the 3.3 and 5 volt lines just to make sure the 12's are good? If so - how much should I draw? My actual reading was 11.89 volts on the 12 volt line but I think my meter is not even close to being calibrated. I once opened it up to see if it could be calibrated but the darn thing has so many trim pots I wouldn't know which to adjust. The meter measures Hertz as well as Capacitance in addition to the usual volts and ohms. It can even detect a 120 volt power line without making contact (NCVD).

 

opamp circuits from a pc supply is a bit creepy. They are noisy.

you should consider additional LC filter. Also they dont like no load. but the new ones do fine with 470R LED.
Also theres a capacitor coupling to mains, though a small one.

Threads here arent closed when the question was resolved.

 

Thanks Takao.

As for the noise factor - this isn't any kind of precision application, nor is this any kind of audio device, though its intended function is to work from a pre-amplified musical source to drive a Laser Diode focused on an offset mirror casting a circular pattern on a wall or screen. The audio frequency will pulse the laser both with amplitude and with frequency. The resulting pattern on the wall will be that of a broken up circle that pulses with the amplitude. Dependent on the frequency of the music and the RPM of the motor/mirror the dots on the wall may appear stationary or may move one direction or the other. Lower frequencies will produce dashes whereas higher frequencies will produce dots. In the course of the musical composition the dots and dashes interplay with each other to form unique but repeatable patterns on the wall. Cool to watch - even more fun when you've been smoking something. NOT THAT I ADVISE OR CONDONE SMOKING OF ANY KIND - ANY SUBSTANCE.

 

..................
I've been trying to figure out Op Amps. Have read LOTS of material on them and most say you don't need the negative rail. However, all my experiments with just 12 volts positive have proven to be a failure. I'm SURE I'm doing something wrong, so maybe with a double rail supply I can figure out either what I'm doing wrong OR what works. ................

Problems with trying to amplify AC signals with single supply op amps are often incorrect biasing of the signal.
Typically the op amp input needs to be biased at 1/2 the supply voltage using a virtual ground circuit (which can be a simple as a two resistor voltage divider) to properly amplify AC signals whose peak voltages go above and below ground, even if the op amp is specified to operate from a single supply.

 

I have seen a 10 ohm, 10 watt resistor recommended as a load on any 5 volt red wire to a ground black wire. On the last one I did, I used an automotive halogen headlight bulb. It just glows from inside as an indicator light and the power supply is happy with it.
This is a good article http://web2.murraystate.edu/andy.batts/ps/powersupply.htm

 

A 10W 4.7R resistor across the 5v rail will do as a dummy load.

 

So you think I should load the 3.3 and 5 volt lines just to make sure the 12's are good? If so - how much should I draw? My actual reading was 11.89 volts on the 12 volt line but I think my meter is not even close to being calibrated. I once opened it up to see if it could be calibrated but the darn thing has so many trim pots I wouldn't know which to adjust. The meter measures Hertz as well as Capacitance in addition to the usual volts and ohms. It can even detect a 120 volt power line without making contact (NCVD).

An ATX PSU I stripped down and traced the circuit had a resistor network sampling both the 5V & 3.3V rails into a single TL431 - you'd have to load both rails to get the correct voltage on the 12V outputs.

It was a cheap PSU, and a long time ago - later and/or better quality units may have better regulation circuits.

On one low spec unit; the 3.3V sense wire had a pin all to itself on the main connector instead of crimped in with one of the supply pins. My test load at the time only picked off one pin for each voltage and failed to link the sense wire to the 3.3V pins - the PSU went bang at switch on!

Now all my test loads are made with the power connector still soldered to a strip of PCB cut out of a scrap motherboard.

 

Which PSU is it?

 

Which PSU is it?

Which PSU is what???

 

If setting up a PC P.S. as a bench supply the common is usually connected to earth ground, which is not always desirable in a bench supply, usually a selected option.
It should be possible to separate the earth ground connection.
Max.

 

If setting up a PC P.S. as a bench supply the common is usually connected to earth ground, which is not always desirable in a bench supply, usually a selected option.
It should be possible to separate the earth ground connection.
Max.

That could be possible with a couple of the magazine projects - but those are for advanced constructors who are willing to strip the chopper transformer and re wind it.

It still isn't trivial - the re wind has to be done very professionally in the manner of double insulated equipment.

 

Which PSU is what???

Im asking the OP if he could tell us the name of the manufacturer which should make things easier because then we could find the schematics.

 

That could be possible with a couple of the magazine projects - but those are for advanced constructors who are willing to strip the chopper transformer and re wind it.

It still isn't trivial - the re wind has to be done very professionally in the manner of double insulated equipment.

??? This is the P.S. output, it is isolated up until the final output which is earth grounded?
No need to touch any of the SMPS components.
Max.

 

The PS came out of a DELL, not too old. I looked all over the labeling to see some kind of identification but didn't see any. I don't know if it's ATX - or something else. It did have 24 pins. A couple wires I didn't use were the purple (+12v) and the grey (+5v). Think I read somewhere that the grey is for sensing a voltage. But since all is working as far as the meter goes - It remains unterminated and shrink-tubed.

I've seen some documents that suggest there's a negative 5 volt line, but this one didn't have that. It also had a couple yellow wires at +12v and a bunch of white wires +12v. the 12v is rated at 18 amps but I'm not going to be using it anywhere close to that.

Looking at what others have done - someone used a 10 watt 4.7 ohm resistor as a load on the 5v line. I haven't done any loading on mine and when I turn it on it stays on. 12 volts floats around 11.9v, close enough. And keep in mind I don't have much faith in my DVM. I'm looking into getting something much better. Maybe a Fluke. Anyone know of a better meter?

I've finished modifying my PS. (-12v) (Gnd) (+12v) / (+3.3v) (Gnd) (+5v). One of the smaller gauge 5v lines I used for an LED to indicate when the unit is on. All I have left to do is stick some feet on it and I'm done. My connections are common banana plug (sockets). Since there were so many available ground leads I opted to put two ground posts. Don't know if I'll need them but they're there if ever the case comes up - like running something on 12v and on 5v, precipitating the need for a second ground.

Just for a measure of safety I put a dollop of hot melt glue on the panel where there's a slight chance (otherwise) of the wires contacting case ground.