PDA

View Full Version : Troubleshooting assistance question/opinion?



dhoelzer
December 23rd, 2014, 11:31 AM
I'm looking for some troubleshooting advice before I tear something apart.

I have an SGI O2 that just walked in the door. If I plug it in, the fan spins up and power light comes on for about 1 second, then it shuts itself off. After that it is non-responsive unless I pull out the power cord and let it sit for a few minutes. After that I can get it to repeat the same performance.

My first thoughts are that there's something wrong in the switching power supply, likely capacitor related, since it powers on at all. The switching power supply is my prime suspect since I've seen them do this type of thing when there's either no load on them or there's something wrong in the feedback circuit.

Opinions, please? Normally I'd just tear it open and start replacing capacitors but this little monster has a stack of boards in the power supply.

Stone
December 23rd, 2014, 11:41 AM
It could be a short outside the power supply. You could check the 5V and 12V rails to be sure one of them isn't short.

dhoelzer
December 23rd, 2014, 12:24 PM
Same behavior with the PSU removed. I get 5 volts and 12 volts for about 1 second.

I think the feedback loop is triggering. No components look marginal, unfortunately.. :(


It could be a short outside the power supply. You could check the 5V and 12V rails to be sure one of them isn't short.

KC9UDX
December 23rd, 2014, 12:34 PM
I have no experience with SGI. But, what you're describing, I've seen some switchers do if there isn't enough load. I think it's unlikely in your case, but it might be worthwhile to try plugging in some heavier load; maybe just a hard drive or something.

Stone
December 23rd, 2014, 01:50 PM
Same behavior with the PSU removed. I get 5 volts and 12 volts for about 1 second.
No, that's not what i was getting at. :-) Check them on the motherboard. If the PSU is good and the board has a short, either 12V or 5V, you could get that same type of response. Check them (12V and 5V) with and without the PSU connected. You might also try a power-good shunt.

dhoelzer
December 23rd, 2014, 03:00 PM
Unfortunately, it's not really a feasible test with this thing. The entire thing is modular ,everything sliding into a backplane. This includes the power supply. The great thing is that there are zero wires in the case. The bad thing is that you can't get at the motherboaard (or anything else for that matter) when the stuff is slotted in.

I'm going to proceed with caps at this point.. I agree that it acts like it's not enough load, but to me that could indicate that the feedback circuit is borked. I've seen something similar with Iic computers.. I have had a few that work great until you access the internal floppy drive. When you do the current draw change drives the feedback loop wild if there are some marginal caps and the entire thing resets.


No, that's not what i was getting at. :-) Check them on the motherboard. If the PSU is good and the board has a short, either 12V or 5V, you could get that same type of response. Check them (12V and 5V) with and without the PSU connected. You might also try a power-good shunt.

dhoelzer
December 23rd, 2014, 04:10 PM
1 bad cap found, 1 marginal.

I have a 2200 uF here that reads off the charts on my cap tester. ESR was also higher than expected.

Another cap is marginal. Time to just change out the lot. Now the question is if a replacement PSU is cheaper than the caps. :)

Uniballer
December 25th, 2014, 03:52 AM
Another cap is marginal. Time to just change out the lot. Now the question is if a replacement PSU is cheaper than the caps. :)

I think you mean, the question is if a replacement power supply will also be ready to fail? Aluminum electrolytic caps sometimes dry out and fail while just sitting on the shelf.

dhoelzer
December 25th, 2014, 01:45 PM
Absolutely true but in the event that it's the high frequency transistor rather than the caps it certainly saves me a great deal of time and energy troubleshooting to have a working one to compare to. :) The schematics for these seem to be unfindable (at least for me).

Personally, I always find switching PSUs to be difficult to follow and trace. I think it's because when I started out it was using massive isolation transformers, step-downs and bridge rectifiers rather than these new-fangled switching doohickeys. :) I'm invariably more successful with them when I have a spare for comparison.


I think you mean, the question is if a replacement power supply will also be ready to fail? Aluminum electrolytic caps sometimes dry out and fail while just sitting on the shelf.

NeXT
December 25th, 2014, 03:31 PM
O2's can sometimes have a really cranky SoftPower circuit. There's a method to fix it and normally you can find it on nekochan.net but they've been offline for several weeks.

One thing to remember when working on an O2 is to NEVER REMOVE THE LOGIC BOARD UNLESS THE AC CORD IS UNPLUGGED.
Even a skilled person like me missed this important step and for the last two months I've been trying to source a replacement logic board.

Anyways, there's a single jumper on the logic board marked POWER ON. Jumper the pins and put the machine back together. When you plug it in it will turn itself on (because the jumper forces the computer to turn in whenever it has AC power). Stop it at the PROM, enter the console and type POWER OFF.
Now remove the jumper and the soft power circuit should be reset.

I'm curious about that one bad cap since the O2's are not known for bad power supplies.

dhoelzer
December 27th, 2014, 06:02 AM
I'm curious about that one bad cap since the O2's are not known for bad power supplies.

The replacement PSU came in.. Plugged it in and everything works beautifully.

I'm afraid I can't tell you the precise location of the cap since I pulled them to test them, but it was one of the 10,000 uF caps on the secondary side of the board. Of course, I'm not repairing that board at this point since the replacement PSU was only $25 and I'd be looking at more than that to replace all of the caps, but it had a substantially high resistance while one or two others were marginal.

NeXT
December 27th, 2014, 07:42 AM
That's quite bizarre. Okay I'll take a note of it anyways.

dhoelzer
December 28th, 2014, 05:43 AM
I can see why these rarely fail. SGI actually used very high quality capacitors in the PSU, especially given the age. The capacitor was C-210 on the board. C-206 was marginal, just a tad high on the ESR.

Hope this helps!

It was a 10000 uF 6.3 volt capacitor
That's quite bizarre. Okay I'll take a note of it anyways.