I have an unused pc atx psu. I want to make a bench power supply. The usual ebay breakouts give 3v 5v and +12v and -12v. Can I use an LM317 and make a variable output up to 24v? Any advice gratefully received.

  • 2
    \$\begingroup\$ you can, but the total current would be limited by the -12 V supply current, which is very low ... you would also not be able to use the resulting 24 V output in the same circuit as the 5 V output \$\endgroup\$
    – jsotola
    Feb 28, 2021 at 21:07
  • \$\begingroup\$ If you look at the manual for your power supply, there is usually a list of the power supply's rails and if the supply is modular, often a list of which ports connect to which rail so you can balance loads better between the rails. You might be better off running an isolated 12-24V boost converter off of a decently powerful 12V rail. \$\endgroup\$
    – K H
    Feb 28, 2021 at 22:12
  • \$\begingroup\$ I'm a noob and tried to help a friend with his home-brew 3D printer. I suggested we re-purpose an ATX and while we learned a lot there was a heck of a lot of workarounds and he ended up buying a non-ATX. In my experience you'd be getting yourself a bunch of unexpected problems. If you can afford the $20 buy something designed for your project's needs. \$\endgroup\$
    – JonathanZ
    Mar 1, 2021 at 19:24

2 Answers 2


Can I use an LM317 and make a variable output up to 24v?

No. The LM317 has a voltage drop of ~2V, so the maximum you will get is ~22 V.

A better solution would be to boost the +12 V to +26 V using eg. an XL6009 based module, then regulate down to the voltage you want with the LM317.

To reduce power dissipation in the LM317 at lower output voltages you could turn the booster output voltage down, eg. to 14 V when the LM317 is regulating down to 12 V, or 12 V (minimum possible booster output) when regulating down to 10 V and lower.

  • \$\begingroup\$ I'm experimenting with TI MSP430 microcontrollers for which the power requirements are modest. But there is often a need to do something real such as automate my house climate control so higher power is required. It's interesting that you can play around with electronics at a signal level ie microcontrollers and get some useful results without really understanding the fundamental electrical properties. Sometimes you just want to experiment with a coil of wire and see what happens so this is where the higher voltage comes in. \$\endgroup\$ Mar 6, 2021 at 19:17

Maybe, depending on your needs.

The first problem is that since the -12V is very limited in current, typically few hundred milliamperes tops, it will be the limiting factor, even if there is tens of amperes of +12V available.

Second problem is that the power supply 0V is connected to mains input earth, so the outputs really are +12V and -12V, even if the total voltage is 24V. So it is not a floating 24V power supply. If use it to power up a 24V device, remember that the negative lead is still -12V, and thus cannot be connected to any other earth-referenced device. If you do, it will short the -12V output to 0V via mains earth wires.

And no, making the supply float by disconnecting mains earth or modifying the internals to have floating output can be dangerous, even if some people do it.

Third problem is that it is still an ATX supply with multiple outputs. They may not be that stable when loading only one of the positive outputs, and when the load is suddenly disconnected. The power supply output may rise high enough that it hits the overvoltage protection limit and shuts down. So since it is an ATX power supply, power must be drawn from it as per the ATX specifications mandate how it must be used, or it may not work properly.

Basically, getting a proper lab power supply is better than making a makeshift lab power supply from ATX power supply.

  • \$\begingroup\$ 3rd problem: the classic workaround was to use a 6V incandescent bulb of several W to load down the 5V rail; that tends to be most sensitive to under-loading or at least that used to be the case \$\endgroup\$
    – Chris H
    Mar 1, 2021 at 11:30
  • \$\begingroup\$ My needs are mostly experimenting with a desire to replicate some of the discoveries of pioneers in early radio. So running a current through a coil and looking for the results on my Pluto SDR is about my level. With my background in computer support I never really had a need to understand more fundamental considerations regarding power. Thanks for the advice. \$\endgroup\$ Mar 6, 2021 at 19:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.