I’m having trouble getting a Sonnet Crescendo PCI 1GHz up and running in my Power Mac 8600 (Kansas) or 9600 (Tsunami). An 800MHz version of that card in the same Macs works fine.
Symptoms:
- The Mac freezes while loading the Sonnet extension(2.3 or 3.1).
- With the Sonnet extension disabled, there is a “divide by zero” error while loading the Open Transport extension.
- With both Sonnet and Open Transport disabled, booting is normal, but of course no L2 cache, no L3 cache, and no Altivec.
Background:
This is a card that underwent a poor quality recap with MLCC many years ago, when I was first learning how to solder SMD devices. These symptoms began immediately after. I stopped using the card. Lesson learned: Do not practice on unobtainium.
Years later, I saw that R5 and C11 had each been pushed off a pad during the recap.
I recently recapped the card with aluminum polymer caps, and put R5 and C11 back where they belong. I had to put a bodge wire from R5 to a via because of a lifted pad. When that had no effect, I used a bit of electrically conductive silver bearing paint from R5 to the trace that leads to the via. The repairs made no difference.
It seems to me that the issue might be bad cache on the 1GHz card. What are your thoughts? If the cache is bad, my question then becomes: Where is the cache on the card, and can it be replaced?
Symptoms:
- The Mac freezes while loading the Sonnet extension(2.3 or 3.1).
- With the Sonnet extension disabled, there is a “divide by zero” error while loading the Open Transport extension.
- With both Sonnet and Open Transport disabled, booting is normal, but of course no L2 cache, no L3 cache, and no Altivec.
Background:
This is a card that underwent a poor quality recap with MLCC many years ago, when I was first learning how to solder SMD devices. These symptoms began immediately after. I stopped using the card. Lesson learned: Do not practice on unobtainium.
Years later, I saw that R5 and C11 had each been pushed off a pad during the recap.
I recently recapped the card with aluminum polymer caps, and put R5 and C11 back where they belong. I had to put a bodge wire from R5 to a via because of a lifted pad. When that had no effect, I used a bit of electrically conductive silver bearing paint from R5 to the trace that leads to the via. The repairs made no difference.
It seems to me that the issue might be bad cache on the 1GHz card. What are your thoughts? If the cache is bad, my question then becomes: Where is the cache on the card, and can it be replaced?
