Computer PSU not reaching 12v when powering on

Discussion in 'General Electronics Chat' started by RogueRose, May 15, 2015.

  1. RogueRose

    Thread Starter Member

    Oct 10, 2014
    189
    4
    I have a standard PC PSU that I pulled to use as a bench supply unit. It seemed to work fine for a while and then I think something shorted while using it so I set it aside to look at it later. What happened was the resistive load wire broke so it wouldn't power on. Fixed this.

    I pulled the unit today and it seemed to be working and I pulled the board from the metal case to put it on my project board and now it isn't working. I turn it on and I get about 6-8v on the 12v rail and it drops about .75v every second or so until it hits .49v and sits there. Power off the unit and after about 20-30 seconds I hear a little click and the voltage starts to drop to zero. I don't get any voltage on the 5 or 3.3v rails. I think this PSU has some jumpers that run the power from the 12v, through a resistor, into the 5, through the jumper, into another resistor, into the 3.3. IDK if that is standard on other supplies but I was able to trace that on this unit.

    Can anyone give me any insight on what might be the issue or where to start looking here? This is a nice supply that has adjustable voltage so I'd like to save it.
     
  2. BR-549

    Well-Known Member

    Sep 22, 2013
    2,004
    394
    How did you fix the resistive load wire and how long did you use the supply after you fixed it?
     
  3. Hypatia's Protege

    Distinguished Member

    Mar 1, 2015
    2,803
    1,237
    For what it's worth, IIRC ATX (and their ilk) derive regulation for all 'channels' from the +5v rail...

    Best regards
    HP
     
  4. studiot

    AAC Fanatic!

    Nov 9, 2007
    5,005
    513
    First step is to ensure that the +5 volt independent supply is working. This is pin 9 the purple wire in a modern ATX based supply. It is called the 5V sb (standby) pin.

    This is not the normal 5v rail, but is always on (if the mains is on), regardless of the state of the rest of the PSU.

    The system works like this

    The 5V sb wire powers some sensing circuitry on the motherboard, and some in the psu itself.

    When you press the power on switch you momentarily short the green wire on pin 16 to ground (black wires).
    This provides an active low signal to start the power supply running.
    The power supply monitors its own rails.
    When these are stable it puts a signal on the grey wire pin8 called PS_OK which tells the sensing circuitry on the motherboard to start the motherboard.
     
  5. RogueRose

    Thread Starter Member

    Oct 10, 2014
    189
    4

    I am getting voltage (5.10v) to the purple wire (I cut the 24 pin connector off btw). I never had the grey wire hooked up to make this work before.

    When I fixed the resistive load, the original axial lead had broken off so I got a new resistor and soldered it in place. Once it started working again I tested it about 3-4 times powering it on and off, unplugging and such, and making sure that it working.
     
  6. dl324

    Distinguished Member

    Mar 30, 2015
    3,250
    626
    Have you tried grounding the "power on" wire as suggested? That signal is held high internally and needs to be low for the supply to turn on.
    A don't care for your application.
    I've used ATX power supplies to power bare motherboards and I've never had to mock up a power switch to get them to turn on.
     
  7. ScottWang

    Moderator

    Aug 23, 2012
    4,855
    767
    If you trying to fix the psu then you have to attach some clear photos.

    When I see a psu have some problems, what I thought was that probably some capacitors was bulge or the capacitor liquid was dry out.

    Check the surface of each capacitors, if you could then separate from pcb, and measure their capacitance, sometimes you have to using the ESR meter to test the resistance of capacitor.
     
  8. ian field

    Distinguished Member

    Oct 27, 2012
    4,415
    784
    AFAIK: the 3.3 rail is the most critical, the ones I've traced had a resistor summing network to sample both the +5 & +3.3 for the PWM controller.

    There's usually a quad comparator watching the most important rails for out of spec voltage, an error here will shut the PSU down, I would imagine the more expensive units probably have a latching shut down that you have to unplug it to reset. Most just shut down and then the fault condition disappears so it starts up again (pulsing).

    Most types rely on tight coupling in the chopper transformer to regulate one or two rails and have the rest come in "close enough" - but this relies on all the outputs having an "expected" appropriate load.

    In the simplest case, only the +5 & +3.3 rails are sensed for regulation, the PWM responds by increasing the mark space ratio to supply any current draw - unless that PWM response is made to happen, you won't get much out of the +12 rail.
     
    Hypatia's Protege likes this.
  9. Hypatia's Protege

    Distinguished Member

    Mar 1, 2015
    2,803
    1,237
    FWIW: I can attest that such is indeed the case for over-current (triggered via low EMF detection?) on (at least) the 5v rail with every PC supply I've thus encountered -- all the way back to the AT:)

    Best regards
    HP
     
  10. ian field

    Distinguished Member

    Oct 27, 2012
    4,415
    784
    I've encountered a few that sat there "ticking", every time the sensed over voltage deacayed it tried to start again.

    One of the most frequent faults was the start up resistor going high or open. The top reservoir electrolytic has a bleed resistor across it, the bleed resistor on the bottom capacitor is usually returned the the base (or gate) circuit of the bottom transistor in the half bridge - that makes sure one transistor conducts at switch on, that kick puts enough energy through the transformer to start up the control chip on the secondary side.

    For a while, I had a run of blown up PSUs come in that had a 150V MOV across each of the reservoir electrolytics - rectified and smoothed 230VRMS comes to somewhere in the vicinity of 320VDC, so the MOVs were a loud bang waiting to happen.

    One came in with incinerated insulation laquer on the secondary side toroidal choke. A sheet aluminium heatsink bolted to the steel casing didn't have matching ventilation slots to line up with the ones in the case. A scrap unit from the "graveyard" donated a slightly bigger choke, it took a while to work out how many turns to remove to correct the +3.3 rail though.
     
Loading...