4
\$\begingroup\$

I have an old switching power supply for HP 712/60 workstation (+5V, +3V3, +12v). It was working about ten years ago when i've last tried to boot the system, but now one of the output voltages is at half of the nominal value (12V output is at 5.6V). I have replaced all of the electrolytic capacitors in the SMPS, and resoldered all points on low voltage section - still 12V output is at 5.6V. What to look for next?
General question - SMPS has output voltage lower than nominal - what troubleshooting strategy should i take to find the culprit?

\$\endgroup\$
9
  • 3
    \$\begingroup\$ Does the 5.6V output voltage stay (roughly) constant when you put a load on the rail? If yes, it's probably the feedback circuit. If not, it might be the secondary rectifier diodes. \$\endgroup\$ Nov 1, 2017 at 18:59
  • 1
    \$\begingroup\$ Put a load on the 5V rail yet? 10% This is the primary regulator. The others are just tracking. \$\endgroup\$ Nov 1, 2017 at 19:07
  • 1
    \$\begingroup\$ the measurements are: 5.8V on 12V rail with no load. 5.4V on 12V rail with mainboard connected. 5.25V on the 12V rail with mainboard and HDD connected (sorry for the original descripton of 5.6V on 12V rail - it does not actually have this value at any time). Other voltages are: 5V rail is 5.85V with mainboard connected (seems a bit too high, but...) and 3V3 seems fairly stable. Mainboard is actually working ok (enters PROM, and waits for the boot source. Disk is not spinning up - probably cause there is no 12V), so 5V and 3V3 are ok. \$\endgroup\$
    – woytekm
    Nov 1, 2017 at 19:11
  • 1
    \$\begingroup\$ @Fizz the 12V not regulated, it just follow the 5V because of the transformer ratio. \$\endgroup\$
    – Jasen
    Nov 3, 2017 at 5:32
  • 2
    \$\begingroup\$ 5.85V seems high. Should be closer to 5.1V \$\endgroup\$ Nov 3, 2017 at 7:01

4 Answers 4

1
\$\begingroup\$

One of the most important troubleshooting procedures I do, is to measure the resistance between output rails to the ground. If the resistance is too low, then one of semiconductor parts on the secondary side; e.g. diodes or MOSFETs might be broken. Otherwise, which I'm not so practical with is the capacitors also might be faulty.

Good luck,

\$\endgroup\$
0
\$\begingroup\$

Chapter 7 of this document MAY have useful information:

https://parisc.wiki.kernel.org/images-parisc/a/a8/712_technical_manual.pdf

I can't read German well enough to tell if this document contains useful information:

https://computermuseum.wordpress.com/2006/05/09/model_712_60/

This website MIGHT offer some replacement parts:

https://www.partsit.com/product/hp-712-60

This website offers some manuals:

https://www.manualslib.com/products/Hp-Model-712-60-Workstation-3127355.html

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Thanks for the links, i went through all of them, and one useful thing is the PSU description in the technical manual: "It's a 100Khz single FET forward convertor with +5V and +12V as a the two main outputs. Both the main outputs are sensed and fed back to provide an average regulation for the two outputs." \$\endgroup\$
    – woytekm
    Nov 3, 2017 at 7:38
0
\$\begingroup\$

Apart from the obvoius things like checking electrolytics and for dry solder joints and for broken windings in the transformer, you should also check your feedback circuit. If something is wrong with the resistor voltage divider or optocoupler in the feedback circuit, this could have the effect of a lower output voltage than expected.

\$\endgroup\$
0
\$\begingroup\$

A simple answer is: measure as much as you can :-)

If you have a clamp amp-meter and the PSU output wires are exposed enough, check the currents out the 12V and 5V rail. What about power consumption at the primary side?

Is it possible to disconnect the PSU from the motherboard? If yes, can you start the PSU disconnected from the mobo? If yes, can you still see the disbalance in voltages, 5V vs. 12V?

There's a single feedback loop, sensing the two output rails via a simple resistive divider (hence the "average" regulation). I don't have a schematic of your particular PSU, but take a look at R25/R26 in the schematic here, near the label saying "FEEDBACK". I would expect the feedback to pass through an opto-coupler, maybe preceded by a TL431 reference (also working as a comparator) but in general outlines it's a fairly basic PC PSU arrangement. The disbalance that you observe likely means that the +12V rail is heavily loaded and the +5V rail is not loaded at all. Look for a short to ground in the 12V section. Hopefully it's not in a secondary winding in the main transformer. Or it could indeed be a rectifier diode, conducting both ways. Some do this only under RF load (not if you check the diode alone with a tester). If there was indeed a short, I'd expect something to smell and burn. If there's no "Ampere odeur", that's weird...

BTW, if your 12V rail is actually lower than your 5V rail, it might be a clue that each rail has a dedicated secondary winding. I.e. the windings are not connected in series. Hmm... have you tried measuring voltage between the 5V and 12V outputs? Could there be a short between those two rails? After rectifiers, or before rectification? What if the secondary windings are in series, and the section "between 5V and 12V" is shorted? And the difference after rectification could merely reflect a difference in load (in the diodes' forward voltages).

Are you saying that you've replaced all the elyts? With what family / model?

Do you have an oscilloscope? That's a key tool when debugging SMPS...

Unfortunately the resistance in secondary windings is so small that checking it with an ohm-meter yields no useful data :-(

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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