Just killed my rare WGS9150! Repair help?

[willmurray461]

While messing around with my Workgroup Server 9150, I managed to kill the motherboard. I was plugging in one of those PDS ribbon cables (for use with a HPV/AV video card) into a Sonnet G3 upgrade, and the connector went in at somewhat of an angle, managing to short some pins on the PDS passthrough. The machine was turned off, but I think some of those pins were still live due to the soft power-on circuitry. I heard some crackling noises and now the board doesn't chime/no video even with or without the G3 upgrade. Did I kill the CPU? Is there something else that could have died that's less annoying to replace (buffers, etc)? I've replaced 601's before, but before I go attempt another replacement, I thought I'd ask anyone here if they know some troubleshooting tips to narrow down what probably died.

 

[Byrd]

I'm guessing with a tear down, strip down to known good working basics (remove G3, cards, piles of RAM), clean it all and check it over, it'll fire up again. Add components one by one with the G3 and HPV card last.

 

[willmurray461]

I already tried that. I removed all the RAM, the G3 card, and populated the PDS slot with the terminator. I also tried swapping the cache and ROM SIMMs with no luck. Do you know if there's a way to check if a CPU is dead with a multimeter somehow?

 

[Franklinstein]

I blew out an external SCSI hard drive years ago by crookedly installing a terminator. Oops. Gotta pay attention next time, I guess.

Anyway, you'll probably want to get a bench power supply and inject voltage directly, bypassing the soft power and related control/regulators and allowing a slow, controlled voltage increase and current draw monitoring.

The 601 user manual should give a list of pins that would be useful for probing. I don't know enough about the 9150 to be able to say if it always used the 601v (the later 3.3v variant) or if the 80MHz model used a regular 601 (5v). The chip itself should be labeled though.

 

[willmurray461]

It's the 601v.

Isn't voltage injection primarily used to find shorts with a thermal camera, or is there something else that I can look for?

 

[willmurray461]

I think I may have found the problem. After inspecting the board, I noticed that the area around the main 3.3V regulator (LT1085) for the 601v looked a bit burnt. I replaced the LT1085 but it didn't make a difference.

I then decided to measure things and noticed that the input to the LT1085 was nothing. I traced it back to the PSU connector and voilà, there's a 5A fuse that was blown that's supposed to connect the LT1085 to 12V (that probably explains why the LT1085 area looked so burnt; seriously, a linear regulator dropping 12V down to 3.3V???).

Hopefully, replacing this fuse will fix the problem.

One concern I have however is the resistance between 5V and GND only being ~15 ohms. Is this normal or a sign of something else being shorted?

 

[Franklinstein]

It's the 601v.

Isn't voltage injection primarily used to find shorts with a thermal camera, or is there something else that I can look for?

It helps find shorts with a thermal camera, but it also provides safe voltages for tracing circuits and avoiding potentially troublesome devices that would normally interfere (or that may be defective) such as soft power circuits or voltage regulators.

 

[willmurray461]

It lives! After replacing the fuse, everything is back to normal. Guess I was pretty lucky with this one.

I'm still a little concerned about that linear regulator dropping 12v down to 3.3v; I'm very tempted to have it replaced with a buck converter, but for now I'll just leave it alone.

For those of you who are curious, the fuse was F5 (right next to the PSU connector). The fuse is a 125V 5A Littelfuse 459 series. They still make them, so I got a few from DigiKey.