[synthetix]

I just tried to start up my R10K Indigo2 after a few months (it was working fine previously), but it won't start now. When pressing the power button, the CPU fan twitches, then nothing. I tried unplugging the power, waiting, then trying again, but same deal. Fans twitch for a fraction of a second. That's all.
I took the power supply apart and aside from some slight corrosion on I believe the pins of a couple LM358s, I didn't see any obvious problems. No obvious leaking or bulging caps. I cleaned up the corrosion and tried it again, but no dice. Just the fan twitch-n-stop.
The power supply is the 060-8002-01 model. My machine is the Solid Impact R10K 195MHz.
Is it possible the problem is elsewhere? Again, everything was working fine a couple months ago. It was just sitting in my office in the corner since then.

[Raion]

It's a bad PSU for sure. They have power good signals, if that doesn't come online then the PSU shuts off to protect components. So yeah, caps, diodes, mosfets, could be anything.

[synthetix]

It's a bad PSU for sure. They have power good signals, if that doesn't come online then the PSU shuts off to protect components. So yeah, caps, diodes, mosfets, could be anything.

Whew, that's good to hear. I'm happy it doesn't sound like a motherboard problem!
I ordered some new caps from Digi-Key and will try replacing them before giving up and replacing it. I found a link on I think Ian Mapleson's site where it shows which caps to replace on the high voltage half of the PSU. Since the caps are only a few dollars, I'll try this first.

[Raion]

get a desoldering electric pump. I'm getting a good one to finish my O2 project.

[synthetix]

get a desoldering electric pump. I'm getting a good one to finish my O2 project.

Good call. This problem was actually the final straw in holding off on buying a vacuum desoldering gun, which I just did. I managed to survive without one for years, but every single time I'd repair some random device around the house I wished I had one. Ahh, those years of frustration will soon be history...

[sunray]

get a desoldering electric pump. I'm getting a good one to finish my O2 project.

Good call. This problem was actually the final straw in holding off on buying a vacuum desoldering gun, which I just did. I managed to survive without one for years, but every single time I'd repair some random device around the house I wished I had one. Ahh, those years of frustration will soon be history...

There are many tools and tools are fn to get but for this kind of work nothing beats this simple classic tool

[synthetix]

I started removing caps from the power supply, and noticed some trouble on the low voltage board. There is a group of six 6800uF caps on that board, all of which had green corrosion around the leads and a couple with traces of liquid under them. One was so bad (C451) that it ate away some of the mask on the PCB, exposing copper (see pic). Fortunately, I don't think this will cause a problem as it's very minimal.
I cleaned up the mess, and will order replacements.
Also, I didn't see any obvious trouble on the high voltage board. No corrosion or bulging caps. I did order new caps for this board, so I'll replace them anyway even if they're not causing the trouble.
More to come!

[synthetix]

I've got an update on the power supply repair. Unfortunately, replacing the caps didn't do anything. The machine still tries to start, then stops after the fans twitch for a second. At this point, I need to delve deeper and see about testing/replacing other things. If anybody has any intel on where to start looking for other failed components, let me know.

In the mean time, I started a list of all the caps on the low voltage side of the power supply. This isn't a comprehensive list, as there may be some caps not listed here. Here it is in case someone needs it:

SGI Indigo2 IMPACT Power supply
Zytec 060-8002-001, rev A

Low Voltage Board:

C353: 6800 µF 6.3v (ø16mm)
C354: 3300 µF 16v (ø16mm)
C451: 6800 µF 6.3v (ø16mm)
C464: 6800 µF 6.3v (ø16mm)
C342: 6800 µF 6.3v (ø16mm)
C452: 6800 µF 6.3v (ø16mm)
C344: 6800 µF 6.3v (ø16mm)
C454: 6800 µF 6.3v (ø16mm)
C379: 2200 µF 16v (ø12mm)
C385: 3300 µF 10v (ø12mm)
C370: 470 µF 35v (ø10mm)
C400: 470 µF 35v (ø10mm)
C363: 47 µF 35v (ø6.5mm)
C364: 47 µF 35v (ø6.5mm)
C349: 47 µF 25v (ø5mm)

[Raion]

Trace the power good circuit back. I bet there's a problem somewhere along that circuit.

[weblacky]

I haven’t gotten to these PSU models yet. However, most of the time, the twitching or attempt to start indicates a problem on the secondary (low voltage) side of the PSU and/or somewhere along the feedback circuit. So likely the high voltage side isn’t the main issue (doesn’t mean it’s without any problem, just ones that prevent initial startup).

While I know this PSU is a pain due to its design. If it were on my bench I’d do two things immediately to check.

Test the feedback parts (assuming it has an optocoupler or a mini transformer) for functionality and that it’s still connected (through conductivity/ohm tests) to the power management IC. And secondly, I’d do a resistance test (need a high quality ohm meter for inductance behavior) and test each dc voltage rail in the PSU connector (when PSU OFF) to test for bad secondary diodes. A single bad secondary diode can cause this problem in a power supply. Take your red and black probes from you Ohm meter and test by getting a reading in ohms for a voltage rail (again from pinout, PSU OFF AND UNPLUGGED). First put red in V rail and black on PSU connector ground. Record the reading then reverse the leads and do it again. You should get a second (different) reading that’s often higher.

The reason for this is you’ve bypassed the diode by reversing your probes and are now measuring the winding of the secondary transformer coil. Which should give you a little higher reading. If you get roughly the same reading in both directions, then you know you have a shorted diode on that rail and can go investigate (assuming it’s it a rail that uses a transformer, big wattage rails should)

Those two things are easy to check without removing much and are both power-off testing so safe for you and your meter.

Also I’m unsure about Indigo2 PSU, but I have discovered the at Indy PSU using IC voltage regulators to create the smaller capacity voltages (like ones that can do barely any current, -12v I think is one). So it’s possible that not all voltages are created equally!

Another test you could consider (but has risk) is leaving a meter on min/max voltage and doing a startup test in each voltage rail while the meter is connected and running and try to “catch” a voltage reading during that brief twitch. This might tell you which rail to focus on (assuming it’s not a logic issue).

Good luck.