[Trippynet]

The firm trying to repair my flaky Indigo2 PSU have just contacted me to ask the following:

We are currently looking at your power supply but are struggling to find any connection details for the unit or what we need to do to enable the unit. Usually with this type of supply 2 pins need to linked together to enable the outputs

In short, does anyone have any information on which pins need to be connected to fire the PSU up? I'd obviously quite like them to test it before they send it back to me! I doubt it'll make any difference, but it's an IMPACT PSU.

Many thanks in advance.

[Irinikus]

Is this the second time this unit has failed, as I recall you sending an IMPACT PSU in for repair in the past?

Check this thread: http://forums.irixnet.org/thread-1473.html

Luke managed to get his failed power supply to start briefly in this thread by connecting J10 pin5 and J11 pin 11.

[Trippynet]

Well, it's a longish story.

Sent it to them for repair, got it back and it still kept dying after 30 minutes or so.

Sent it back to them (1 year warranty), came back and seemed much better. But after prolonged testing, i found it keeps dying after 90 minutes or so now.

Sent it back again (keen to ensure it is fully fixed before the 1 year warranty expires). They've had it a while, but now say that they're struggling to test it. My expectation so far is that they re-capped it the first time, I know they replaced a few resistors the second time that didn't seem to give the expected ratings, but I expect up until now that they've not been able to power the thing up and give it a thorough test - they've just replaced bits that look troublesome before sending it back.

I imagine they're getting sick of me sending it back to them so actually want to power it up for a while to make sure it works, which suits me - but I wish they'd done this earlier.

I did find the bits from Luke earlier, if all else fails I can tell them that - however I thought I'd check in case anyone has a more comprehensive pin-out of the supply first.

[weblacky]

Hi Trippynet,
I'm still new at this...so don't take what I say as gospel. I'm trying to learn this stuff to perform PSU repairs myself, but I'm still a beginner.

However I recently thought about your stated symptoms and I'm wondering if it's a special kind of failure related to the components heating up. If they haven't touched any of the semiconductors, then they should. I'm currently guessing you may have what's known as a "leaky output diode". It's not a literal leak like a capacitor (something you can't see), it means it's passing voltage at a point where it shouldn't normally. The output diodes are the thing that actually makes the basic dirty DC (as we know it) at the other side of the transformer (chopper) at the beginning of the second half of the PSU (lower voltage). Each voltage line should have one or a pair of them in a single body. They allow only a range of voltage to flow, cutting off the rest of the chopped up inverted-sine wave (filtering really), which produces a gentle, narrow, wave in a tight DC range...that is later smoothed by output caps and inductors, etc.

The output diodes can just be leaky, or in very rare causes, they become leaky went heated. Most of the time the leak occurs very high up on the voltage range where they is some kind of internal breakdown of the biasing material.

I've been told by the interwebs, that the symptoms of a leaky output diode is a sudden spike in voltage, which should trigger the over-voltage protection on the PSU and do an immediate shutdown.

I took your words literally when you said 30-90 minutes, then it turns off, I assume you mean like someone pulled the plug...not like the system claims it's now shutting down :-)

I would have them replace all your output diodes, there aren't that many in the PSU, they aren't pricey...but a little tough to get to physically.

If they have replaced all your caps, and have done basic resistor checking and your system runs for some time, then I think it's a semiconductor breakdown during heating, it's rare but it does happen. And you have several of these lines, any one of the lines going high voltage will cause this shutdown.

They are asking to bench-run the PSU to be able to test for things like this (ripple, bad power, failure under load, etc)…but if we KNEW how the darn things could be started...we'd be doing this ourselves. I’d say the easiest way would be to hook the PSU up to the station and just use an oscilloscope with logging to check ripple and over current right on the chassis while it’s running.

But short answer is, if they are willing to source, insulate, and install new fast output diodes, I would do that. I have a hunch that’s the right direction to go in.

My two cents, good luck!
-Weblacky

Well, it's a longish story.

Sent it to them for repair, got it back and it still kept dying after 30 minutes or so.

Sent it back to them (1 year warranty), came back and seemed much better. But after prolonged testing, i found it keeps dying after 90 minutes or so now.

Sent it back again (keen to ensure it is fully fixed before the 1 year warranty expires). They've had it a while, but now say that they're struggling to test it. My expectation so far is that they re-capped it the first time, I know they replaced a few resistors the second time that didn't seem to give the expected ratings, but I expect up until now that they've not been able to power the thing up and give it a thorough test - they've just replaced bits that look troublesome before sending it back.

I imagine they're getting sick of me sending it back to them so actually want to power it up for a while to make sure it works, which suits me - but I wish they'd done this earlier.

I did find the bits from Luke earlier, if all else fails I can tell them that - however I thought I'd check in case anyone has a more comprehensive pin-out of the supply first.

[Trippynet]

Quick update on this. The company have had the PSU for a couple of months now after I sent it back to them again.

Finally had an update from them yesterday, they've declared the PSU to be broken beyond economical repair. Whether this is actually true, or whether they simply have run out of ideas to fix it I'm not sure. However they will be sending the PSU back to me and have offered me a full refund of the money I paid them to fix it, so it's not a complete disaster - and it was working a bit better than originally (uptime from 30 minutes to 90 minutes after they re-capped it and replaced a few other components).

So, not ideal unfortunately - but it was worth a punt! Maybe I'll see if any other firms want to have a crack at it (especially now the re-capping part is done).

[weblacky]

So I'm still fixated on your issue. Let me ask, what testing have you done? This sudden shutoff after such a long period seems weird.

You said it shuts off like 90 minutes now, right? Is that waiting on the PROM screen or idle in Irix?

Have you checked the memory? I'm seeing some people is PC-land claim they get 30 minute shut offs (even in BIOS-only) of their machines for odd RAM issues. Can you isolate a trusted minimal RAM set and see if this changes your random shutdown?

[Trippynet]

The shutting down is whilst running in IRIX, usually with apps open and whatnot. After it dies, if I power it back up, it'll run for just a few minutes before dying. I've even had it die like this whilst IRIX was still booting up.

Don't think it's memory related as I have a second PSU and with this fitted, it'll run for a good 3+ hours reliably without any issues. The problems only arise if I fit the dodgy PSU to it. Plus after the PSU repair company swapped a few resistors that weren't giving correct readings, uptime changed from about 30 minutes to nearly 90 minutes.

[Irinikus]

It actually looks like a pretty simple unit, barring the various transformers. Why don't you just replace all of the components?

[Trippynet]

It has been fully re-capped. Once I get it back I might look at another firm that could replace the other bits. Unfortunately in this case, re-capping it didn't fix the random shutdowns :(

[Irinikus]

It has been fully re-capped. Once I get it back I might look at another firm that could replace the other bits. Unfortunately in this case, re-capping it didn't fix the random shutdowns :(

You might want to replace the semiconductors too! (Especially the large ones doing the switching!)