I'm trying to use a small ATX power supply (version 1.0 connector) as a bench power supply. I've found that it only turns on intermittently. I've tried with ~1A load on +5V, and 0.2A on +3.3V .

I accidentally shorted +5VSB to ground, and it turned on. Now I find I can reliably turn the power supply on by momentarily attaching a load to +5VSB, maybe about 0.2A.

Is it possible that this PSU needs a load on +5VSB to start up (but not to keep it running)? An "ATX Design Guide" online lists only 0.0 amps as min current on +5VSB. Is it more likely that the PSU is damaged?

  • \$\begingroup\$ I've never seen an ATX power supply with minimum current specified for the +5VSB. You could load up to 1 A or more on the +5VSB to make sure that it's working (a short is a bit severe). I'd suggest you probably have a fault in the turn on circuitry. PS_ON# should measure 5 V before you use it to enable the PS. If you pull PS_ON# low with a multimeter it should draw about 1.6 mA. \$\endgroup\$ Commented Feb 10, 2017 at 3:36
  • \$\begingroup\$ I had a resistor and LED on +5VSB and it seems to work properly. That's what initially shorted against the case ground. I don't short it any more, usually use 17 ohm resistor. The LED and resistor alone I guess are not enough of a load. I was thinking of measuring a minimum load that will have it turn on. \$\endgroup\$
    – user209759
    Commented Feb 11, 2017 at 3:57
  • \$\begingroup\$ I've measured PS_ON and it is only 0.7 V while the PSU is off. I am getting a negligible mA reading with it on, however that is with a 10A meter input (500mA doesn't appear to be working at all). \$\endgroup\$
    – user209759
    Commented Feb 11, 2017 at 4:01
  • \$\begingroup\$ ... The ATX Design Guide says PS_ON should be 0-0.8V when pulled low. It should be 5V when open circuit. When open circuit, PS_ON seems to work properly (PSU won't turn on, even if +5VSB is shorted), yet it measures like it's pulled low???? So something's wrong with PS_ON? \$\endgroup\$
    – user209759
    Commented Feb 11, 2017 at 4:09
  • 1
    \$\begingroup\$ Adding the extra load on +5VSB (to force the power on) drops it from 5.011V to 4.983V. I think +5VSB is fine (Design Guide says overcurrent protection is required on it, so I don't think I damaged it). The PS_ON componentry is the only thing so far measuring out of spec. The PSU turned on properly occasionally before, so I think the .03 V change in +5VSB is enough to overcome the fault. \$\endgroup\$
    – user209759
    Commented Feb 11, 2017 at 7:08

1 Answer 1


When you just connect your ATX PSU to mains AC, no load, the +5VSB rail should be at +5V, and the PS_ON should be HIGH, either +5V, or +3.3V.

You are saying that +5VSB is normal (+5V and holds some load), but the voltage on PS_ON (green) wire is 0.7V. If the PS_ON is LOW (0.7V as you say) from the very beginning, the PS_ON circuitry is damaged/shorted, possibly by ESD. This pin has a simple pull-up, and must be shorted to ground to turn the PSU on. It is possible that, even when damaged, the remaining PS_ON circuitry needs a power toggling to turn the rest of PSU rails on.

In modern ATX PSUs the PS_ON is managed by a separate specialized IC, and the PS_ON is a simple input, no other active components is there. If the pin is damaged, it can't be repaired, and only IC replacement can fix the problem.

  • \$\begingroup\$ Yes, it seems like the PS_ON circuitry is damaged. I don't think the PSU is very modern. It is a Samsung SFX-108 C Rev B2. I couldn't find a schematic and I would need to remove the PCB from the case to find what PS_ON is connected to. I'm curious to see if I can find damaged parts, but will probably not risk messing with it. For now it works, other than having to "jump start" it. Thanks. \$\endgroup\$
    – user209759
    Commented Feb 15, 2017 at 22:24
  • \$\begingroup\$ @user209759, the same website has older schematics as well. I did check several other schematics, the PS_ON also comes directly into some controller chip. My advice - throw this PSU into recycling, and get a new one. \$\endgroup\$ Commented Feb 15, 2017 at 23:51
  • \$\begingroup\$ Even in the modern schematic that you linked, PS_ON (labelled "ON/OFF"? in the PS_ON# circuitry in lower middle) is only connected to a resistor, as with other schematics I've glanced at. I figured I might have luck finding a damaged resistor? (or cap, such as C26 in the example) Again, I won't risk trying this without a correct schematic. I plan to keep using the power supply since its voltage is closer to spec than another older AT supply I have, and I am assuming that the damage is limited to the PS_ON circuitry. \$\endgroup\$
    – user209759
    Commented Feb 17, 2017 at 17:18
  • \$\begingroup\$ @user209759, it is very unlikely that any resistor or cap is damaged, although it happens sometimes. The problem is likely in silicon pin, inside the control IC. \$\endgroup\$ Commented Feb 17, 2017 at 17:35
  • \$\begingroup\$ You're probably right. The same chip is used for OV protection in the example, so maybe it's best to take your advice to recycle it. \$\endgroup\$
    – user209759
    Commented Feb 17, 2017 at 17:45

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.