ABit KG7 Memtest Bugs Fixed by Recap

[bgavin]

I just finished recapping a KG7 with the usual bulging/leaking Nichicon 3300uF in the VRM. I replaced all 1500 and 3300uF caps with Rubycon MCZ. As an experiment, all TEAPO 1000 and smaller were not replaced.

The interesting part was this board no longer fails Memtest86 with the usual Athlon test failures. It runs all the way through without error. I run the Driving slew rate=7, as I do with all KG7 boards. This is the only deviation from stock clocking.

It will be interesting to see if recapping has also affected the need for the fast slew rate.

[bgavin]

Re: ABit KG7 Memtest Bugs Fixed by Recap

I went back and tried it with non-ECC memory, and it fails the usual Athlon tests.



I guess the only memory that Memtest86 likes in this board is ECC.

Sorry for the false hope...

[WNG]

Re: ABit KG7 Memtest Bugs Fixed by Recap

Not claiming that the 1000uF caps are at fault, but I've seen these bulge and fail.
I would include 1000uF caps as a value to replace.
I believe a good number of these are used around memory slots and AGP/PCI slots.

[willawake]

Re: ABit KG7 Memtest Bugs Fixed by Recap

agree

also which memtest are you using, memtest86 or memtest86+

[bgavin]

Re: ABit KG7 Memtest Bugs Fixed by Recap

Normally I replace everything from 470uF and up. Got lazy on this board to see what happens. I have several KG7 that require recapping, all have the same 3300uF VRM cap failures.

All the memory caps are 1500uF, and were replaced. As noted above, the 1000uF are AGP/PCI located.

I use Memtest86 v3.3 on bootable CD. I have not tried the "plus" version.

[bgavin]

Re: ABit KG7 Memtest Bugs Fixed by Recap

Adding to this thread, I have a different KG7 board under test as of this writing. The board fails on Test #6 of both Memtest86 and Memtest86+.

I don't remember test 6 being a problem. The docs for the old Memtest86 note tests #5, 8 being problems.

[gdement]

Re: ABit KG7 Memtest Bugs Fixed by Recap

How many memory modules are you using? If I'm not mistaken, the AMD761 doesn't officially support more than 4 banks of unbuffered memory, yet the ABit board can hold 8 of them (4 dimms).

[bgavin]

Re: ABit KG7 Memtest Bugs Fixed by Recap

I must be dyslexic... the failing test is #5. I recapped another KG7 today, and it fails the same as the other two, test #5.

Memory is Micron 256 DDR as one stick, DIMM #4 closest to IDE connectors. Book says install #4 first, 3 next, etc.

I tried to download the source for Memtest86+, but the .TZ file is corrupt. Request sent to the board owner..