Dead DayStar Genesis MP CPU card -- ideas?

[ThisDoesNotCompute]

I have a DayStar Genesis MP with the quad CPU card, and it doesn't want to boot. Pressing the power button gets the system to power on, but there's no chime, video, or hard drive activity. The system has a brand new PRAM battery. Swapping the CPU card for another one, like a Sonnet G3 accelerator or dual-CPU card from a PM9500, gets the system to boot normally -- so the problem is clearly with the quad CPU card.

There's no info I can find about the card itself in terms of schematics. The heatsink puts a decent amount of pressure on the PCB, enough that the PCB flexes a bit, so my initial suspicion was that perhaps there were some loose solder joints between the CPU legs and the board. I reflowed all of them, but no change. Very briefly (less than 10 seconds) running the card without the heatsink let me find with an IR thermometer that the CPU at position U2 gets much warmer than the others. But I don't know if that's indicative of a short, or if that's normal since the classic Mac OS only uses one CPU unless application software is MP-aware.

The power pigtail for the CPU card reads the expected 3.3V. Barring a failed component on the CPU card (there are no electrolytic caps on it), does anyone have any experience with these machines?

 

[Phipli]

I have a DayStar Genesis MP with the quad CPU card, and it doesn't want to boot. Pressing the power button gets the system to power on, but there's no chime, video, or hard drive activity. The system has a brand new PRAM battery. Swapping the CPU card for another one, like a Sonnet G3 accelerator or dual-CPU card from a PM9500, gets the system to boot normally -- so the problem is clearly with the quad CPU card.

There's no info I can find about the card itself in terms of schematics. The heatsink puts a decent amount of pressure on the PCB, enough that the PCB flexes a bit, so my initial suspicion was that perhaps there were some loose solder joints between the CPU legs and the board. I reflowed all of them, but no change. Very briefly (less than 10 seconds) running the card without the heatsink let me find with an IR thermometer that the CPU at position U2 gets much warmer than the others. But I don't know if that's indicative of a short, or if that's normal since the classic Mac OS only uses one CPU unless application software is MP-aware.

The power pigtail for the CPU card reads the expected 3.3V. Barring a failed component on the CPU card (there are no electrolytic caps on it), does anyone have any experience with these machines?

I wouldn't run anything like that for any amount of time without a heatsink unless I was significantly underclocking it. They put out much more heat than a G3.

Have you cleaned the connector with contact cleaner? I know it sounds trivial, but its the most common issue I've seen with that era of processor.

Another possibility is a weak PSU or dry logic board caps resulting in low quality power. That thing draws much more power than most other CPUs that go in that slot.

Quad 604e card must have a crazy power draw.

Edit - Just grabbed the datasheet - 180MHz 604e chips are rated at up to 14.2W each so thats potentially up to 57.2W for the quad card. Compared to... 11.9W for a single 500MHz G4 7410.

 

[Phipli]

Do you have a photo of the card with the heatsink removed? For my curiosity.

 

[ThisDoesNotCompute]

I don't have any photos handy, but there's a good look at the card around 3:48 in my video about the machine.

 

[herd]

If your CPU chips have legs then I'm guessing it's not the BGA version? Those thin QFP packages were pretty fragile. You could check the ceramic substrate for cracks under magnification. Here is an example:

68kmla.org/bb/index.php?threads/power-mac-7100-ppc-601-cpu-package-damaged.43507/