Convert an ATX Power Supply Into a Regular DC Power Supply!

A DC power supply can be hard to find and expensive. With features that are more or less hit or miss for what you need.

In this Instructable, I will show you how to convert a computer power supply into a regular DC power supply with 12, 5 and 3.3 volt outputs. For about $10!

Why use a computer (ATX) power supply? Well, they're available everywhere, and they can output tremendous amounts of power in a small form factor. They have overload protection built right in, and even a 500W model can be reasonably priced with high efficiency. The voltage rails are incredibly stable. Giving nice, clean DC current even at high loads.

Plus, it's likely that many of you simply have an extra one lying around doing nothing. Might as well get the most value for your investment.

Step 1 Getting Started

The first order of business is that of safety. While I'm reasonably sure that there isn't enough residual energy to stop your heart, those capacitors can still bite, and that can cause significant pain and maybe even burns. So be paranoid when getting close to the internal circuitry. It would probably be a good idea to put on some insulating gloves. Also (obviously) make sure the thing is unplugged. You are responsible for your own safety!

Here are the tools/parts needed:

Drill
Needle-nose pliers
Soldering iron
3 x "Banana Jack" Insulated Binding Post sets
1 x bag of "#6" Ring Tongue Terminals (16-14 gauge)
Rubber feet
Small bit of heat shrink.
Screwdriver
Wire strippers

Ok, let's get to voiding some warranties!

Comments

[junbata]

Good day !
Can Anybody Please
help me to Idfentified the pin out of my ATX COMPAQ
model PS 5201 4T2.
I like to make power supply also. from that Model.
The position of the green or Power ON is not in Position 14.
Please help me to solve this.
and do I need sense wire also. to be able to run it.
I have 20 pins out.

GOD BLESS.

Thanks

[altricious]

Depending on your PS, there may be "Power Sensing" wires that also need to be connected. You can tell because there will be two wires to a single terminal on the 2x10 or 2x12 plug. Mine had a brown wire that needed to be tied to the orange otherwise it would not run.

[Surg]

Hi guys.

Thanks for this great info. I just converted my first ATX 600w PSU model ULT-VX600 yesterday. It is working good thus far, and is so much quieter then the marine battery charger that I was previously running. I use it to power a Hyperion LiPo battery charger for my RC hobbie. My charger is capable of accepting 12-28 volts on the DC input side. I did some test charging pulling 25 amps from the PSU and am noticing the output voltage of the PSU drop to close to 11 volts and as a result my charger stops charging one leg. I have 10g wire soldered to the power bus on the inside to provide power to my charger. It shouldn't be a voltage drop issue.

Is there a way to tweak it to make it put out over 12 volts?

I was also considering wiring two ATX's in series to get 24 volts, but the problem would be that they share case ground with - DC. If I put the + from one the ATX to the - of the other it would short out since they are using a common AC source for power. The -12v line is only rated at less then one amp, so that is worthless to me.

Any suggestions?

[techxpert]

lesson for everyone don't put your hand on the big yellow transformer while pushing the power cord into the socket or big jolt tingling up your arm before jumping back just in time to save your life and live to type this up on another computer!

[Blacklight [HUN]]

Nice job.
My problem is that I can't run a zvs driver (pulls around 5A at 24V), a plasma speaker either.
It just simply turns off.
What could be the problem? :S

[louige99]

well you have 24V at 5A which is 120W, i assume your PSU is rated above that but that doesnt really mean much at all. most PSU's only allow certain max power for a combination of rails. For instance, a PSU with a 12V, 5V, and 3.3V rail might be rated at 250W, but the combined wattage of the 5V and 3.3V rails might only be 110W.

In short, check what specific ratings are on the stickers on the PSU and search the internet for the data sheet if you can find it. If you cant, i would say thats your problem and you can really only fix it by getting a new PSU that has a 25V 5A rating on it

[soshimo]

Do you have any problems with arcing on the low voltage terminals? Most ATX PSU's put out over 30A on the +5v line. That could cause arcing or smoke your circuit. Have you had much trouble with this? I'm wondering if it's not better to just pony up another $100 for a lab PSU. The potential loss of some high priced IC's I have (and some of which are not available anymore) is not worth saving a few bucks IMO. Good tut anyway. One last point, those caps inside the PSU can hold a SIGNIFICANT charge. They will do more than bite, they could kill you. I've blasted molten pieces off a screwdriver when trying to discharge them. FYI.

[shussain13]

Hi I want to use an Esc to control a motor. Instead of providing power to Esc through Lipos, can I use regulated Dc power supply?

Actually I want to make a test bench and I want to do series of experiments relating to my project, for which I cannot wait to charge the Lipo again and again

Any guidance in this will be much appreciated

Regards

[diy_bloke]

great stuff. I used something similar to feed a bunch of flashes during a photoshoot

[nsharma11]

can i take 24volt output from a normal atx computer smps...can any body help me for this

[bwdane]

Thank you for the instructable. I used this as a basis for a slightly different project. I am using a 500W power supply to power a radio in my garage consisting of old car audio bits. It is able to power a 500W amplifier that is pushing two 4.5" component speakers and a 12" subwoofer as well as an Alpine head unit. Now I just have to figure out how to add a small battery to hold the radio's settings.

[trosa3]

Great instructable! I am a R.C. enthusiast and I have had many problems finding a reliable 12v dc supply for my battery chargers. This really worked out great for me, It only costs a few bucks and like you said almost everyone has a pc power supply laying around.

[AntonMadness]

I would like to share a rule of thumb we use at my company concerning maximum current vs. wire thickness:
1mm^2 10A
1.5mm^2 16A
2,5mm^2 25A
4mm^2 32A
6mm^2 40A
My PSU is rated:
-12V 0,8A
-5V 0,5A
+3.3V 26A
+5V 35A
+12V 15A

I'm not thinking of using the negative voltages. The wires are AWG16 = 1.310mm^2. Because I don't think I will go to the max, 2 wires will give me enough room to pull 25A per channel(NOT simultaneously!).

So using two wires will give you 200% current without melting your wires and it will save you some room and time grouping ALL the wires.

[JTreehorn]

Thought I would post these pics of my finished product. Give your supply a name, take some measurements, and with a little photoshop you can give your power supply a professional look. (Sorry about the poor quality images. Cell phones were not made for photography)

[Sitnalta]

That is awesome! How did you make the labels?

[ninjatails]

That looks fantastic!

[emitsorrels]

that looks delicious

[JTreehorn]

P.S. Almost forgot to say thanks for the instructable. So... thank you very much. If it wasn't for people like you who post these things, I wouldn't know the first place to begin. Keep 'em coming.

[rcarter5]

hi, this is grate, i fly rc helicopters that run on 12s lipo batries, (44.4v) my battery charger needs a 400 w power suply
http://www.hobbyking.com/hobbyking/store/__17753__Turnigy_Mega_200Wx2_Battery_charger_discharger_400w_.html
i will be charging at 10 amps x 2

i have a 600w psu, will it handle wot i need?

thanx, wuld apriciate it if some one culd email me direct, i dont get mutch time to get on line at the mo
[e-mail]

[LasVegas]

If you start that supply without a minimum load of a few amps on the +5 volt line, your going to simply burn out the supply. Also, why didn't you make the -12v line available? While limited, it's still very useful.

[Gelfling6]

most of the time, these pans usually include a 10-33 Ohm wire-wound 10W or 5W resistor across the +5V supply wires, to maintain an ample load, with minimal load to detract from the +5V voltage. (Though, a 33-Ohm looks more like a 5-Watt film resistor, in my 400-Watt supply, looks pretty small, but appears to provide this load.) I've noticed some of these supplies have the load resistor already soldered right to the board, across both the +5 & +3.3 voltages. I agree on the -12V line.. there are still plenty of projects that requires even the old EIA RS-232 voltages (+12V & -12V)

When I look for supplies to convert, I'll actually scrap supplies that don't have the -5V live, for parts.. those with everything, get converted.

[sschoemann]

no the supply will not burn out if started without a load. it just wont start up worst case scenario....

[Sitnalta]

I'm not so sure about the "minimum load." I've been running a ton of light stuff (100mA range) on it and so far there hasn't been any issue. Even if it's ultimate lifespan is diminished, better this than a $120 bench PSU.

[Sitnalta]

Is it? For what?

[Gelfling6]

I stand corrected on the part#.. LM317 regulator. and, Ironic, the instructions are right here, on Instructables: http://www.instructables.com/id/Convert-A-Computer-Power-supply-to-a-Bench-Top-Lab/

[Gelfling6]

I can't remember where, there was instructions out on the net, same basic conversion, but as sschoemann says, you can draw tops +24V by using the -12V as GND, and +12V as +V., and it can be varied using a variable regulator. (LM320K, I believe.) You can also get the same by linking through the various voltages.. 8.3 using the +3.3 and -5, 10V using the +5 to -5, 18V (for those lovely home heating system projects) going from either +12 to -5 or +5 to -12.

Right now, Just picked-up a pair of discarded 235W 'Power Man' supplies, , just begging to be converted Provided they start-up. :-)

[sschoemann]

-12 plus the +12 equals a pos. 24 volt supply if you are using it to power mobile military or aircraft electonics... simply use the -12 for the ground and the +12 as 24v positive.

[albylovesscience]

if there still is some energy how do i discharge it to make it safe to matinenceit and also how potent? enough to kill,knock out, or just a nasty zap

[Gelfling6]

Simple enough, place a 10W wire-wound resistor across each voltage to GND, and wait a minute.. (literally 60 seconds, on each voltage, to GND)).. most of the time, these supplies discharge within that 1-minute anyways. But, if You're totally paranoid (well, honestly, who isn't?) leave the device disconnected for the whole day before opening it. I've yet to get zapped by any ATX supply, that I've opened, even within 10 minutes of running it. (right out of the machine, immediately dismantled and scavenged for parts.) Even the old switching supplies from XT & AT machines (Yipes! Did I just show my age?) were pretty discharged once disconnected from the AC cord, after 5-10 minutes.

[junits15]

usually just a nasty zap, nothing too dangerous, but you really should discharge them beforhand

[zack247]

you could plug it into your computer again, and turn on the computer a couple times, then it will make it safer to open the power supply. then, connect the green and ground wires, and plug in a cdrom drive and turn on the psu. that should suck all the power out of it.

[Robot Lover]

You should have made a -12 volt binding post for op amps and the likes. ;)

[Spuzzum]

Thanks man.. going to wire a high powered led panel with this. :D

By the way.. as for amperage... it's a heck of a lot higher than just 3,3Amps.

Look at the specs for a generic 350w psu:

Rosewill RV350 350W ATX 1.3 Power Supply

+3.3V@28A
+5V@35A
+12V@15A
-5V@0.5A
-12V@0.8A
+5VSB@2.5A


There's quite a lot of juice available :D

By the way... the +3.3v and +5v share the same amperage source, while the +12v is separate. So, if using +3.3v, and +5v.. be aware of how much is actually needed.. might be better to just draw from the +5v, and add resistors or regulators for the +3.3v feed. :)


Cheers.........

[sschoemann]

if the 3.3 and the 5 volt lines share the same source in this supplies example, they must be using a voltage divider to come up with the 3.3 volt supply since it's seven amps below the level of the 5 volt source, so using resistors to come up with the 3.3 volt supply would just be repeating whats already been done. As for doing it with regulators, to come up with it with solid state regulators would be adding way to many components to be practical so again, using what is already available is the only realistic way to go. BUT unless you are worling with a discrete CMOS based circuit, you really wouldnt have much need gor the 3 volt supply to begin with, UNLESS you were using it in parallel with another source to increase that one's current level.

[Spuzzum]

But if you only need 3v, then a resistor to cut .3v is better than cutting 2v from 5v. And if using for led's as I am, I need a 3v source for my timer. I believe some arduino circuits require both 5v and 3v.. but I haven't completely looked at that design.

[sschoemann]

Remember. LEDs dont worry about voltage, they are current sensitve devices. So unless you are using an LED Driver, that controls the current, you are more worried about limiting the current flow through your LED than the voltage across it. That can be found by simple application of Ohms Law. (voltage divided by current equals the required resistance) Since the newer High Output LEDs have different current limits/requirements, you will have to plug in the apropriate numbers to come up whth the correct current limiting resistor value for each LED. ALSO remember that you can tailor brightness by putting a pot in series with the current limiting resistor to turn down the brightness of the led...

[Spuzzum]

Yep.. already using a 25 turn 10K trimpot.. have capability of less than 100mA to more than 1500mA.. and this driver handles 1 led just as fine as 6 led's.. very adaptable.

I needed to tie loose leg of pot to output though.. otherwise it would slowly rize on it's own after 500mA.

I found 5ohm and 10ohm trimpots as well at digikey. $0.50 each. Both 25 turn units.

[sschoemann]

the multi turn trimmers are typically wire wound, which could be why you are noticing a rise in brightness... wirewound pots and resistors are also inductors. Inductors act as lump resistances to DC. So until it reaches magnetic saturation it will act as a resistor. when it hits full saturation, it will pass the majority of DC applied to it. This is why they work so well in the flyback circuits in highvoltage supplies. try subbing them out with standard single turn carbon type pots once and see if it still rises. If it does, then it is interacting with the caps in the powersupply and you may need to actually add a lump inductance to counter the effect. It's a matter of the old ELI the ICE man saying, (ELI= voltage leading current in an inductive circuit, and ICE = current leading voltage in a capacitive circiuit... In short the resistance is causing a charge/discarge rate that is being displayed by the LED's brightness due to the capacitive nature of the circuit. Remember, LEDs work off of current flow, and dont care about how high the voltage is... You MAY also be able to isolate the powersupply and the LED circuit by inserting a Silicon, rectifier diode in between them. But in many cases this wont work but is worth a try, since 1n914 and similar are very cheap.

[Spuzzum]

I've already got it going with a tester design, using a modified version of "this" design:
http://www.instructables.com/id/Circuits-for-using-High-Power-LED-s/

You'll see in my comments, I originally used a 2-pos dip to select between 2 different feed resistors.. 100K was for input feeds of 10v and above, and 22K was for input feeds of 10v and below.. as per the guy's instructions. Only needed the 100K. He also used a zener.. also not needed.


The board in this picture still has the zener, the 22K resistor, and the 2-pos dip.. they've since been removed.

I'm quite aware that led's are current driven.. that's why I built a "constant current driver". I just wanted the actual current level to be "adjustable".. I hate making things for a "single" purpose. The driver's current is normally set with a "fixed" resistor value.. and it's around the 1ohm and below range. So if you want to change from 350mA to say 700 or 1000mA at a later date.. you'd need to make a whole new circuit. Or replace the resistor. That's why a 25-turn 5ohm or 10ohm trimpot is "perfect". The 100K works fine.. but 5 or 10ohms, with 25 turns to adjust.. would be a lot more precise. Or at least a lot easier to "be" precise.

I'm making a 127 diode fixture, of 8 different wavelengths, between 4 different current requirements. So I'm making 16 different driver boards, and dividing the runs between them.. setting each current accordingly.


Cheers................

[myronj]

Does anyone know what Amps this unit would put out and if there is a way to make that amperage variable?