[Xav101]

So a few months back I ended up trading some hardware for a Power Indigo2 with a bad PSU. At some point I put the PSU from my good I2 in it and ran it just enough to get to the PROM but for various reasons I shelved the machine and ran my normal I2 for a while. I recently put the PSU back in the Power I2 to try and test some things on it, and it didn't want to come up.

Symptoms:

- On pressing power button, fans began spinning. Light on HDD came on but it didn't seem to spin up. No boot chime or video output. Power button didn't respond to turn the system off.
- I reseated the R8000 card and the PSU cables. Same symptoms. At some point shortly after the psu fan stopped without me hitting the power button. Power button was unresponsive and unplugging/replugging the psu didn't help.
- Removed the PSU from the system and plugged it in, fan started spinning again (This is a non-IMPACT PSU where it will power up if power is applied but it is not connected to the system.)

Is this a PSU issue? If it's a PSU issue I can try and fix it myself or find a different one. I'm just really hoping that my R8000 card isn't dead and this is a CPU issue.

[weblacky]

Yo,
The R8000 is a rare machine so some things could be different. But you confirmed one thing, your PSU USED TO run the system. So that eliminates questions of "are you sure it's the correct PSU model...as the entire Indigo2 line has like 5 of them". So you're PSU model is okay.


If "a while" is two years or more...then yes it's likely the PSU (not 100% but that's where I'd start). Here's some things you need to know.

Fans and HDD are ran off 12V but, unlike modern PCs, most SGIs don't give two cr@ps about 12V on their CPU and mainboard. Their PSUs have HUGE 3.3v and 5v rails because SGI didn't use VRMs often on their old mainboards and ran most boards off the 3.3v and 5v circuits directly off the PSU lines!

When you "free wheel" an Indigo2 PSU (any model) it will not regulate it's voltage without any load attached. So while it's a good sign you can run it on a bench...it's meaningless to detect a hard pass/fail condition.

I suspect that under load, your PSU isn't given 3.3v and/or 5v (likely too low) and so the board doesn't have the correct power. Given that it starts at all though is a very good sign that simply recapping ALL the electrolytics on the PSU will likely result in a totally working PSU. I have seen PSUs (like Indy Nidec) start without one of their voltage lines (in my case 5v) so starting is good for checking logic/brains...but not voltage or regulation. Even after recap you may be missing a voltage rail, but normally that's just one of the smaller rails they might use a regulator for and that's not too hard to eventually figure out with testing.

I'd say you're in a good spot though (assuming the attempted boots of a failed PSU and board fiddling haven't done any harm) for a PSU recap and then to be on your way.

[Xav101]

Alright, hopefully I can try something here.

One thing though is that I know it was working in the regular I2 a few weeks ago because I was using it for some stuff. Is it maybe possible that the 3.3V or 5V rails were sufficient for an R4400 but aren't enough to properly turn on an R8000? I have no clue what the power draw difference is for a 200MHz R4400 vs the R8000.

And yeah, I'm really hoping it isn't an issue with the CPU. I've tried my best to avoid touching the cache chips around the edges because apparently they're very sensitive to static, but you do have to do a certain amount of handling the card while working on the system. Because of its sheer size, the CPU card sits upside down and it actually overhangs one of the PSU connectors, meaning you have to unscrew it and lift it out every time you plug in/unplug the PSU.

[weblacky]

Yeah, it's a mystery and you know that unless there is serial output giving more info, you're gonna have to do the spare parts dance to get more info (break the stalemate). I can say that the graphics does take a bit of power, I hate to recommend attempting to boot in a bad power situation (I would just fix the PSU, regardless of the fault as it's VERY OLD and not safe for a rare machine) but if you remove the graphics card, you may lower the system load enough for a different result. But again I'd not risk the machine in this way if it were in my collection.

The parts will only cost you like $45 shipped to rebuild the PSU (if that) and it's not worth risking a rare machine. Get a PSU up to snuff, then use it for troubleshooting.

[Xav101]

Yeah. I need to do a big cap order anyways for some random boards so I might as well throw this in there.

One thing I can't find a guide on for the life of me is the way to actually get the boards out on the non-IMPACT PSU. The IMPACT PSU, despite being terrible to work on, is fairly simple to get apart with two of the side walls being made from a single piece of sheet metal and the other two being made from a single piece as well. As far as I can tell, the non-IMPACT PSU I have is made of far fewer pieces. The left wall is the same piece of metal as the top and bottom and an overhang on the right, as well as the front with the fan grille. The only two pieces that seem to be independent are the right panel (minus the top overhang) and the backside with the PCB that interconnects the top and bottom boards.

My best guess is that once you remove the right panel, you have to slide the boards out that gap. The screws on the bottom of the PSU seem to hold in the bottom PCB and likewise the screws on the top seem to hold in the top PCB. It seems to me that you might be able to remove the back panel with a flathead, use a phillips to disconnect the PCB interconnects, and then use a flat head to disconnect the AC inlet and slide the whole bottom PCB out the side. From there you might be able to remove the back PCB and the top PCB? It doesn't seem that the front with the fan comes off - although there are two threaded holes on the top and two on the bottom around it, the fasteners seem to have their heads broken off, and on other I2 non-IMPACT PSU pictures I can scrounge up those also seem to be riveted, so probably not removable.

Any guidance on this would be appreciated since most of the guides seem to focus on the IMPACT PSUs.

[weblacky]

Here's what I have about how the IMPACT PSU case separates:

https://siliconimage.irixnet.org/index.p...y/IMG_8135
https://siliconimage.irixnet.org/index.p...y/IMG_8136
https://siliconimage.irixnet.org/index.p...y/IMG_8137