ATX to Lab Bench Power Supply Conversion

In my sophomore year of college at the University of Minnesota, I started into my main electronics classes, and needed a good power supply for working on lab projects at home. My roommate Adam told me about somebody online who had converted an ATX computer power supply into a lab bench power supply, so I decided to do the same thing. You can also check out this link for a very similar guide by their user Abizarl. I have also documented this project on my website at http://www.mbeckler.org/powersupply/ if you are interested.

Warning! There are several large capacitors in ATX power supplies, that will store a dangerous charge for a long time. Please let your power supply discharge, completely unplugged from the wall outlet, for a few days before opening it up. You can probably be seriously hurt, so please be very careful. I am not a lawyer, but I hereby release myself from as much liability as I can, for any sort of injury you sustain, or any trouble you get into.

Step 1 Background

First, a bit of background on a typical ATX power supply:

Computer power supplies are Switch Mode Power Supplies (SMPS), which use high-frequency switching circuit elements to provide a high-quality output voltage, with good energy efficiency. One side effect of this technology is the minimum load requirement that each power supply has. In order to function properly, the power supply needs at least a very small electrical load connected to it. In other words, ATX power supplies will only work if you have something connected to it. We will be using a power resistor to provide this minimum load.

Also, modern power supplies do not simply have an OFF/ON switch, they have what is known as a "soft" power switch. This normally makes no difference to the user, as the computer behaves the same, but when you shutdown your computer, the motherboard can turn off the power supply when it has finished shutting down. This requires us to add our own power switch to the power supply chassis.

To protect our circuit from accidental (and careless!) short circuits, we will install some fuse-holders and fuses, which will disconnect the circuit supply lines if too much current flows. The size of the fuses are up to you, but a 1 amp fuse will work just fine for most circuits. You really should put fuses on all supply lines.

Update: While the diagrams show fuses on all voltage rails and no fuse on the ground line, when I actually built my power supply, I was young and foolish and only put a fuse on the ground wire. It's much safer and a better idea to put fuses on all signal lines and not the ground line. Thanks to many emails and messages on Instructables about this oversight.

Comments

[ertxz18]

Awesome tutorial, I have some questions though. First, how can I make sure my power supply is working without having to plug it back into a computer, I tried shorting the power on to ground but I know that some terminals must have a load in order to work so is there a way to get around that? Also I noticed you used one wire per post, I seen some other tutorials in which they connect ALL of the wires (of their respective voltage) to their post, is there a reason for that? Thanks!

[matthewbeckler]

You can try shorting the power-on wire to ground to see if it might turn on, I've heard that some power supplies don't need a load to keep running. If yours needs a load you could grab some LEDs and resistors, or a small DC fan, or anything like that.

I would connect as many of the wires as you can to the binding posts, but I only connected one wire per post. This determines the maximum current you can draw from the binding post (one wire can safely carry less current than a bunch of wires).

[ertxz18]

Thanks for the respond! I actually have two power supplies and I'm debating which one I should use, one is a 240 W and the other is a 145W. Any characteristics in particular I should look for when building a lab bench power supply?

[matthewbeckler]

For most people the most important specification is the current capacity at their voltage of interest. For example, if you want to charge RC car batteries, you'll be most interested in how many amps can be supplied at 12v. You might also want to test them both to see if both/either/neither require a load resistance. Good luck!

[ertxz18]

Thanks for the help! I greatly appreciate it! Keep on making awesome instructables!

[hangtime21]

(removed by author or community request)

[cubemike99]

Awesome Instructable and explaination, but why does the resistor need to be such a large wattage? Can a higher resistance/wattage be used?

[matthewbeckler]

Mostly it was what I could get from radio shack easily without having to make an order from digikey/mouser/etc. I've never heard "official" numbers about the minimum load current required to keep a power supply, but 0.5A seems to be working for most people. You could use a resistor with higher resistance to try and reduce the load current (V = I * R), and then you could use a resistor rated for less power dissipation (P = V * I) if you want to experiment. You can always use a resistor with a higher wattage, that won't hurt anything, but if you get too low then the resistor will likely heat up and might burn you (funny story: The earlier makerbot 3d printer heating elements to melt the plastic filament for extrusion were actually just power resistors that were driven far beyond their designed power rating, so they get really hot).

[salexandridis]

Hello!
I also tried to convert my 350W ATX, but there are aome problems.
First of all I'd like to mention that I didn't a 10Ω 10W resistor, so I replaced it with two 22Ω 5W resistors in parallel. Is this fine?
When I turn on the switch of the ATX, the Stand By LED lights up. But when I short the green wire to ground, the fan spins for a second (or less) and then.... nothing. The Power On LED doesn't even light up. Firstly, I thought that my PSU is "clever". I thought that the fan was heat controlled. So, in order to test that, I connected an LED to the 5V output, using also a 330Ω resistor. And guess what... The same exactly thing happened. It just blinked.
Any ideas of what's going on here?
P.S. The PSU certainly works, because I just unplugged it from a working PC!

[matthewbeckler]

Hey there, sorry to see that you're having a bit of trouble.

1. The pair of 22 ohm 5W resistors in parallel will have an equivalent resistance of 11 ohms. This will draw I = V / R = 5v / 11 ohm = 0.45 amps, making for TOTAL power dissipation of P = V * I = 5v * 0.45A = 2.2727 Watts, which is split between the two 5W resistors, so you should be just fine.

2. That's a good idea about the PSU being too clever, and I've heard from other people that that happens. It sounds like the PSU thinks it has no load (and shuts off), but if you have the 0.45A load on the 5v line that should work ok I would think.

Perhaps you can try moving that load to 12V or 3.3V, or adding extra loads to those lines?

Good luck, let us know if you solve it!

[salexandridis]

Thank you! I'll try!

[dcoates1]

Thank you! Out of all the instructables I found, yours was the most helpful with your wire diagram. Thank you! You can check out an image of my painted and modded power supply here if you're interested.. -> http://mixologicaldisaster.wordpress.com/2012/04/09/atx-benchtop-power-supply-v2/ :D I tried posting the image below..hope it works..

[matthewbeckler]

Wow, that looks really well done and nicely polished! I like the excess of power connectors in the front of the frame (power pole or something like that?) and the speaker-style wire terminals are a really good idea. Nice work, thanks for sharing your photos and experience!

[mlinhart]

Hi,
I have an AT power supply, it was creating 11.43v from the yellow 12v wire. I wired a 12v globe to one of the black wires and one of the red (5v) wires. I put the multimeter on the 12v and ground wire while the globe was on. it read -12.21 I was wondering why it said negative? and also my lipo charger can handle 12v-17v, would it be ok if I put another globe on the 5v rail? or is one enough already?
cheers

[Elokuu]

Thanks for the great tutorial... I just "converted" an old PSU, i found in the trash, to an nice power supply for an upcoming project... It works fine, except that i didn't had a 10ohm/10W resistor. So i used two 22ohm/5W resistors in parallel. They get pretty hot, hope that's normal!? I also had to put these resistors between GND and 3,3V (Orange) to keep the PSU running (as you explaind)... However, when I measure the voltage on my +12V cable (yellow), i get 13,14V. Is this normal because there is no load on it or is something with my PSU wrong? Thanks again

[matthewbeckler]

Hey there, glad you liked the tutorial! So with two 22ohm resistors in parallel, the equivalent resistance is 22/2 = 11 ohms. This means that there is I = V / R = 5v / 11ohms = 0.45 amps of current evenly divided across the two 22ohm resistors. The power dissipated in each of these resistors would be P = V * I = 5v * 0.45A = 2.27 Watts of power (heat). It seems like that should be reasonable for 5W resistors but maybe you can put them by the PSU's fan so they get cooled a bit?

I've heard that on newer PSUs you also need load resistors on 3v3 so that matches with what you've said.

Yeah, with your 12v line being 13.14 volts that makes sense based on what I've heard and what I know about how these switching supplies actually operate. If you add a small load it should drop down to 12 volts. You can use a computer fan as a good 12v load if you don't have anything else. Hope that helps, good luck!

[Elokuu]

I'll try to put it close to the fan :)... Thanks

[tinkerist]

i'm very sad. i spent the whole day working on this project, and was rather proud (as it's the prettiest electronics project i've put together, to date). but i don't have power to anything except the LED on my rocker switch (green wire) which i used in place of the toggle on yours. the pcb in the atx power supply did have a "3.3vS" wire, which i took to mean that it would need a 3.3v load as well as a 5v, so i added that to the circuit, but it's otherwise identical. any ideas?

[matthewbeckler]

Hey there, sorry to hear you're having trouble with your power supply! You're right, that "3.3vS" wire is probably the 3.3 volt sense wire, so it probably needs to be connected to the 3.3 volt wire. What sort of load did you put on the 5v and 3.3v lines?

Just to confirm, you are still connecting the green wire to ground to turn it on, right? Even with the LED in your rocker switch, the green wire needs to be connected to ground to signal the PSU to turn on.

[tinkerist]

it's the same 10 ohm resister as is on the 5vS line and then to ground. does it need to go to the 3.3v+ instead? should i put a lower resistance sandbar on it? and the ground wire is on the switch.

[matthewbeckler]

Hmm. I'm certainly not an expert, but here's what I would try:

Connect the 5vS wire to one of the 5v wires, and then connect one of the 10 ohm 10 watt resistors between 5v and ground.

Connect the 3.3vS wire to one of the 3.3v wires, and then connect another 10 ohm 10 watt resistors between 3.3v and ground.

Use your rocker switch to either leave the green wire disconnected, or connect it to ground. If your switch has an LED you can power it from the 5vSB (5v standby, which is always on) wire if you want it to be lit up whenenever it's plugged in, or power the rocker switch LED from the regular 5v line if you only want it on when the switch is set to "on".

Does that make sense? Good luck, hope you get things working!

[tinkerist]

ok, so here's what i've got. i've got 5v from the 5vSB and PS ON (purple and green perspectively). no combination of grounding or loading these makes the grey wire (PG) go live. i've found that by grounding both of them i get fractional voltages (less than .02v) from some of the supply wires and the grey PG wire. i'm at a loss. i tried banging my head against it, that didn't work. any ideas?
you're working on a PhD in electrical engineering. if your not an expert, who is? :)

[matthewbeckler]

Ok, thanks for the update. You should never ground the 5vSB line, that is just an indicator line that the PSU is connected to AC power. You can put a resistor+led on that line and it should light up when it's plugged in. You don't need to do anything with this wire, however, it's totally optional.

You probably need to have a 10 ohm 10W power "sandbar" resistor between the regular red wire 5v line and ground. You might also need to do the same on the 12v wire or the 3.3v wire. Separate resistors for all lines you want to add a constant load. Different PSU's have different loading needs, you'll have to experiment a little bit. I've even had some people report not needing any load at all!

Some power supplies have "sense" wires which need to be connected to their corresponding voltage wires, but these are uncommon. I think you said you have one, so connect the 3.3v sense wire to the 3.3v supply line.

If the power supply doesn't come on (meaning you can measure proper voltages at the 5v and 12v lines, as well as the PG wire goes high (5v I think), then your power supply might be broken. Good luck!

(also, there are lots and lots of different areas in EE, and my area of digital electronics and IC chip test are pretty far from analog circuits and power electronics :- )

[tinkerist]

thank you! i'll try all that and get back to you. i had the 5v (red) and the 3.3vS running through 10Ω sandbars. I also tried running 3.3v and the 3.3vS wire together and then through the sandbar. i'm not sure i understand how the 3.3vS wire would get a signal from a 3.3v wire that has no voltage. but i understand practically nothing about electronics, this is part of my learning process. the atx did power the computer the last time i used it, but that was like 3 years ago. thanks again for the advice, i'll let you know how it turns out.

[aclark17]

I didnt use the -5v SB (Stand By) line at all, its a stand by signal wire, and when you put load on it (I think... I had this issue the first time I converted a psu but it might not be universally true) it puts the psu into stand by mode, when you put your pc into stand by mode, which generally turns things off but keeps minimal power for resuming pc use immediately. Thats why youre getting slight voltages, its in stand by.

[seraine]

For some reason, my standy led takes about a minute to turn off when it loses power because it takes a while for the voltage to drain off. Is anyone else experiencing this, and is there a simple solution?

[matthewbeckler]

I see the same exact behavior when I unplug my PSU. I don't know if there's a way to change this behavior, but I also can't think of a reason why it would be a problem.

[seraine]

What exactly does -12 volts mean? Can I connect the -12 volt lines normally (as in, positive goes to -12 volt and negative goes to ground)? Also, there's a couple extra small pink and brown wires that were connected to 3.3 and 5 volt rails.
I've looked around, and I think I could just connect them the wires they were connected to.

[matthewbeckler]

I'm not exactly sure what you are asking about the -12 volt line. Whereas the +12 wire provides a voltage 12 volts above the ground wire, the -12 wire provides a voltage 12 volts below the ground wire. On most PSUs I've seen the -12 wire can't provide very much current at all, so it's pretty much worthless except maybe as a voltage reference?

The extra pink and brown wires are probably "sense" wires that are used to read-in the outputted voltage to make sure each wire is reasonably close to whatever voltage it's supposed to be. I would connect them to the 3.3v and 5v rails again as they were.

[seraine]

Also, I found a small 12 volt, 5 volt and ground wire going to the fan. I'm assuming I just have to include the rest of the wires in order to power the fan, correct?

[matthewbeckler]

Both 12v and 5v wires going to the fan? That seems a little odd. Maybe one of the wires (the 5v wire maybe) is actually a "fan speed sensor" wire? You could try measuring the voltage when it's running to see what it's doing. Good luck!

[milkywayman]

Hi all, just wanted to give you my 2 cents :)

I used a 350W ATX power supply to power a CD car radio and an automated aerial over 12V and here's the results:
- the 3.3V line had to be loaded with almost 2 Ohms of resistance!
The reason is this: the 12V are supplied from the same transformer as the 3.3V and are not regulated as opposed to the 3.3V line. So when I cranked up the volume the PS would shut down since the load on the 12V caused the 3.3V to dip. But since the 3.3V line had no current going out the PS decided to shut down. So in order to get any good power out of the 12V line the 3.3V line should be loaded with at least 1 Amp.
Now I can draw almost 3 Amps on the 12V line without a problem.

PS: your mileage might wary with the load resistor :)

[matthewbeckler]

Another good data point, thank you very much for sharing your experiences with us, I'm sure it will help someone who's seeing a similar problem.!

[pcpowerstereosystem]

and this is a high end stereo system, my power supplys are 30a at 12v and 36a at 5, it will be drawing around 35-40amps, im just powering a car audio amp with these, headunit is on a different smaller one

[pcpowerstereosystem]

i have two 12v 650w power supplys and im looking to power a stereo system, is there anyway i can run them both at like 15v?i have thought about using a red for postive white -5v to the red wire of second, then ground from the sencond, would this give me 15v?

[bbeckwith]

Awesome instructable BTW. So i have built one of these with success about 3 days ago. tested the PSU before "moding" it and it powered on by jumping the "PS_ON" to ground. opened it up, performed the mod exactly as yours minus fuses, and a couple of minor wire color differences. Unit works like a champ. i noticed that this unit has an internal "sandbar" type resistor built into it. it is currently being used to power a fan on my transmission cooler im using to cool my mineral oil submersed PC.
So my brother-in-law is an R/C plane hobbiest, saw my little project last night, and really wanted one of these to charge his batteries with. so with another PSU i had laying around, i first tested it by jumping "PS_ON" and ground, and it worked like a champ.
after all internal mods were done, i went to test the PSU before giving it to my brother-in-law and nothing......my +5vSB LED comes on but thats it. i did notice that this PSU had a +3.3vS line so i connected it to the +3.3v wires in the PSU, no worky worky. it puzzles me that this PSU would turn on before taking the PSU "apart" but not now. im going to take the switch out of the loop tonight and solder the "PS_ON" to ground, but other than that any ideas? maybe i have to connect the 3.3vS wire to the +5.5v line?
just a little background on my self, 9.5 year aircraft avionics tech with basic Electronic Principals knowledge. I can use an "O" scope and DMM like nobody business, and know OHM's Law very well. i appologize for not doing anymore troubleshooting before posting this, but im at work and would like as many ideas as possible to go home with to troubleshoot. thank you everyone in advance!

[bbeckwith]

added a pic for everyone of my set-up. the PSU conversion powers a 12v Transmission cooler fan from the 5v rail. it is supper loud if i have it connected to the 12v. the cooler cools the mineral oil for the PC :)

[matthewbeckler]

Hey there, thanks for the comment! Sorry for the delay in getting back to you.

Yeah, that's a strange case for your second PSU, that it would originally work just with shorting PS_ON to ground, but then when you did the mods that same thing doesn't work. Odd.

Perhaps you created a short by accident when you did the modifications? Did you cut down the extra wires near the PCB? Could a stray strand of wire shorted against the case or another wire? I'd look for that.

I've heard that some newer power supplies might need loads on the 3.3v line so that might be something else to try?

PS. You did plenty of troubleshooting, definitely more troubleshooting than most people do before posting. Good luck, let us know how you get it working!

[bbeckwith]

come to find out it was a bad switch from Radioshack. the first thing i did when i got home was short the PS_ON to Gnd and like a champ the PSU turned on. I did however find that even though the PSU doesn't require a load on the 5v rail, the power that comes off the 12v rail is "cleaner" (less ripple) when a load is applied. so i did in fact add a 10ohm 10w resistor on the 5v rail. thanks for the reply and Happy New Year!


P.S. do you have any experience with server PSU's? im about to buy one i found online for $15 that has a 32amp rating on the 12v rail. for R/C hobbyists that charge batteries with these its perfect. wish me luck :)

[matthewbeckler]

Groovy. I've heard similar things to your "load causes less output ripple" in the past, good to add another data point to that concept.

I haven't played with server PSUs before, but I'd hope they operate in a similar fashion, but might have a different pin out, you know? Probably would still have a pin for "ground to turn it on" but it might be in a different place in the main connector. Let us know what you find out!

I'm also building a reprap prusa, which needs a lot of current at 12v for heating coils and the like, so I picked up this 12v 30A dedicated PSU from Amazon for about $40 including shipping. If you can get your server PSU working that'd definitely be a better deal. Good luck! http://www.amazon.com/Universal-Regulated-Switching-Computer-Project/dp/B004CIW8LA/ref=sr_1_1?ie=UTF8&qid=1325513725&sr=8-1