VP6 only posts with one processor after recapping...

[Shenes]

Hi,

I replaced all the 1000 and 1500 uF caps on my VP6 and now it won't post normally with 2 CPUs. It posts normally with 1 CPU (it doesn't matter which of the two CPUs or in which socket).

If I reset the BIOS settings it reluctantly starts with two cpu's (after a couple of resets) but as soon as I try to save the correct setting from BIOS and restarts it refuses to post.

I've had similar problems before recapping, always had to reset once after power up. But after that it ran normally (with the occasional system freeze once a week or soo). This was the problem i was trying to fix...

Update: I found out it posts "reliably" with fsb's up to 115 with 2 cpu's. (The CPU's are 1000/133mHz ones)


Any ideas of whats going on? Is it fixable?

Thanks,
Johannes

[kc8adu]

what kind of caps did you use?
did you change the ones next to the agp?
the 16v ones near the vrm's?

[Topcat]

I've had a rash if VP6's lately that ended up being unrecoverable. Many had the coil and FET's by the AGP burnt to a crisp, even after replacing them, there was still a dead short in the board which I couldn't trace out. I've 3 or 4 of these turds floating around here now. These boards are competely dead though, no POST at all.

[kc8adu]

i have 3 here like that.
the northbridge is shot in all 3.
1 is the one you sent.
i will try to get chips from cheapie junk mobo's to transplant when i have time to kill.
bga rework can be fun.

[Topcat]

I want PICTURES of that before and after!!!

I got a VP6 in yesturday, the torroid coil and FET's by the AGP beyond recognition, and along with the board being destroyed, it fried a gig of Micron ECC PC133... damn shame... Melted the DIMM's... I'll post pics soon...

[Shenes]

what kind of caps did you use?
did you change the ones next to the agp?
the 16v ones near the vrm's?

I ended up using Rubycon caps. I changed all 1000uF 16V, 1000uF 10V and 1500uF 10V. Had to tip over the two by the AGP.

regards,
johannes

[Shenes]

I've had a rash if VP6's lately that ended up being unrecoverable. Many had the coil and FET's by the AGP burnt to a crisp, even after replacing them, there was still a dead short in the board which I couldn't trace out. I've 3 or 4 of these turds floating around here now. These boards are competely dead though, no POST at all.

All components look ok visually. I redid all the soldering today just to make sure that everything was soldered in properly. How much heat can the caps take from soldering, is it easy to fry them when soldering?

regards,
joahnnes

[Shenes]

edit...

[kc8adu]

try another pair of cpu's if you have them.
sometimes bad caps damage them when the vrm's overshoot due to all the noise.

[Shenes]

try another pair of cpu's if you have them.
sometimes bad caps damage them when the vrm's overshoot due to all the noise.

I don't have any spares, I hope that's not the case. They both work just fine as single CPU's though. When running dual they run stable at half speed.

/j

[Topcat]

I had a few do exactly the same thing. I didn't test them before recapping , so I don't know if the problem was pre-existing. However, after recapping, only 1 CPU would show up.. These were a matched pair of P3 1GHz 133FSB and known good test CPU's. After scratching my head and checking the obvious things, it still did it...

The cure I found is this:

1) Remove ALL sources of power from the motherboard.
2) Remove the CMOS battery (IMPORTANT)
3) Close the clear CMOS Jumper
4) Let the board sit like that for a couple minutes to completely discharge
5) Reset CMOS jumper to normal position and reinstall the CMOS battery (actually, replacing the CMOS battery is a good idea at this time)
6) Reset all your BIOS settings and VOILA!! Problem solved!! Both CPU's showed up and running perfect!

This little procedure has worked on 4 VP6's that did this. My only explaination is that the CMOS data is getting corrupted somehow by the bad caps. The 'one CPU' problem isn't the only CMOS related glitch I've seen with this board, and completely discharging the board and the CMOS usually always fixes it.