help rescue a lightly fried dfi ut nf4 ultra d

foxt

Member
Mar 18, 2006
59
0
0
I posted this over at diy-street as well, but no one has responded so I figured I would try here too.

I am helping someone with a lightly fried nf4 ultra d, and was wondering what you guys think the chances are of rescuing it. Board is 2 years into ownership, so RMA is not an option.

The long story is here, complete with a picture of a blown voltage regulator (I think).

The short story is that the fella connected up a 4 pin berg floppy connector from his psu to a 3 pin fan header connector (the one in the corner, near the base of the edge-side ram slots). Pop!

Remarkably, the board booted. Even more remarkably, it runs prime stable at stock settings (16 hours of the torture test, 24 hours of memtest in the setup in my sig).

Aside from the torture test, a quick check of just a few features reveals that both NICs work, all 4 ram slots work, sata works, audio works, firewire works. Diagnostic leds all behave normally. 2 amber leds are ok. Haven't tried the raid, etc, figured that was all secondary (read on).

The one fan header where the psu was connected is dead, but the other 4 work. Most bios options can be exercised. Now the bad news:

1) vdimm is no longer adjustable (stuck at 2.6). All other voltages can be changed, and I can even *select* a different voltage for vdimm, but whatever I choose, when I hit enter it goes back to 2.6, sorta like vcore behaves on my bent-pin mobile pentium
2) cpu fan speed is fixed at around 1200 rpm as reported in the bios. core temp hit a high of about 46c with that winnie and a stock amd cooler during torture test, sounded like the fan never moved off of about 1200 rpm. I couldnt tell for sure, since speedfan isnt reporting any fan speeds, although they are reported in the bios.
3) speedfan reports -12v at -9.6, and -5v at about -0.5. Something is either borked with the speedfan linkage (maybe, since it wont report fan speeds), or voltage control took a hit during the fireworks. I need to put a voltmeter to work, if I decide to check it any further.

I stopped checking stuff at this point. Just for grins, I did try to bump the cpu freq a bit, but system wouldnt post anywhere past about 210, and I figured it was because I couldn't bump vdimm. I could go back and drop the dividers to rule out the ram and make sure that the board can still run stable on increased freq, but I figured, so what, it's still got a stuck vdimm.

Question is: what to do next? The board does seem to be fine for basic s939 use (if you count 8 hours of torture test as "fine", I didn't really go past that) as long as you don't want to flex its legendary oc muscle. I thought about flashing the bios or some other tweaks, but I'm a little leery of taking a chance on something and actually leaving the board worse off than I found it.

So, what's your advice?
__________________
DFI LanpartyUT nF4 Ultra-D
AMD Athlon 64 3200+ winnie
2x 512 kingston hyperx 2-2-2-5
x300 pci-e
WD caviar 40g
FSP 450w
 

sieistganzfett

Senior member
Mar 2, 2005
588
0
0
:Q leave the board as-is and say use at "your own risk" if its prime stable now, thats a very good sign, but i don't think this board will last another full year of use. it may fail at any given time, just leave everything at stock since its working.

edit - in addition to at "your own risk" when this thing goes, it could potentially take other parts with it.