Power Mac G4 MDD CPU card problems

[kilo]

Hi,
A couple years ago, I bought a dual 1.25GHz MDD. It has always had a very weird problem, the CPU speed would sometimes show as 917 MHz and it would crash every so often. I found out that heating up the CPU board had an effect on this but up until recently, I never really took the time to look into the problem any more than that.
Yesterday I decided to try to overclock it to 1.5 or 1.42GHz. I removed the necessary resistors and reflowed the rest thinking a cracked solder joint caused the issues it previously had. It booted up and showed the CPU Speed was 1.5GHz. I ran geekbench to see what the performance was and... it crashed again.
Since that crash I've tried reflowing more resistors and capacitors around the jumper area but it will now only boot if I point a hair dryer at it for a while. Then it usually just kernel panics on the Apple logo. After changing the speed to 1.42GHz and heating it up more I've gotten it to get to the login screen but it just froze there (probably cooled down too much by then).
Besides the CPU speed jumpers I know nothing about how these cards work so I have no idea what could be wrong. I'm pretty sure it's cracked solder joints but I have no clue which ones could be at fault.
I have attached pictures of the front and back of the card but I can send more detailed ones if anybody wants them.

I'd be interested to see if anyone here has an idea what could be the problem.

 

[mikes-macs]

I think that the issue is related to contact problems with the processor card and the logic board. I had this issues with mine. After cleaning the slot, the card and some finesse, I was able to correct the issue.
Try to solve the issue before trying to over-clock the processor is my advice. My dual 1.42 GHz G4 runs very well still in its stock form.

 

[kilo]

I think that the issue is related to contact problems with the processor card and the logic board. I had this issues with mine. After cleaning the slot, the card and some finesse, I was able to correct the issue.
Try to solve the issue before trying to over-clock the processor is my advice. My dual 1.42 GHz G4 runs very well still in its stock form.

I forgot to mention, but I tried that not long after I got it. It didn't work then but I'll try cleaning it again.

 

[kilo]

I forgot to mention, but I tried that not long after I got it. It didn't work then but I'll try cleaning it again.

Cleaned it again, no luck. A hair dryer still gets it going though. I had it running stably for over an hour. Unfortunately it won't post after cooling down again.

 

[MBongo]

You might check for warping or bowing of the CPU’s PCB (daughtercard) itself, especially along the long edges… with a straightedge of some sort. The use of the hair dryer to pre-heat the board suggests that it might be bowed. And once heated, that “flex” in the board might be diminished enough to allow it to then make sufficient contact and boot?

I have seen this before in various Quicksilvers. In one case, applying downward pressure (by hand) on the heatsink, would then allow the machine to boot. Remove that pressure and the machine then shut down.

 

[kilo]

You might check for warping or bowing of the CPU’s PCB (daughtercard) itself, especially along the long edges… with a straightedge of some sort. The use of the hair dryer to pre-heat the board suggests that it might be bowed. And once heated, that “flex” in the board might be diminished enough to allow it to then make sufficient contact and boot?

I have seen this before in various Quicksilvers. In one case, applying downward pressure (by hand) on the heatsink, would then allow the machine to boot. Remove that pressure and the machine then shut down.

Interesting. The card did seem to be bent slightly, likely from the pressure of the heat sink. Pushing on the heatsink gets it to boot. If I let go, it freezes. Though I still think this is caused by a cracked solder joint.
Interestingly, when I booted into OS 9, I got a cache error during RAM check.