DFI EXPERT and VDIMM ? ? ?

[Dani]

My EXPERT have some serious problems with VDIMM, check this out:

In normal room temp, around +23-25ºC, my VDIMM act like it should do, I put 3.5V in BIOS and it is after boot 3.5V

When I put some cold winter air to blow over my mob, BH-5 and TT BT, we say something around -3ºC, then I put 3.5V from BIOS, boot up and my VDIMM is 3.85V
When I put from BIOS 3.2V and reboot system, my VDIMM is around 3.55V.
Then back in to +23ºC air, everything is normal again!?
From room temp with 3.5V in windows to -3ºC, next boot I have near 3.9VDIMM measured from board

What a F*UCK is going on there

I measured those VDIMM from board.


DFI nF4 EXPERT with 7.04BTA BIOS
2*512mb BH-5
FX-57
Antec TrueControl 550W 20/4pin

[DaWaN]

coldbug

[kiwi]

Not sure about 7.04BTA bios but 12/07 had no problems with vdimm and cold air. What revision do you have?

[Dani]

Do you mean we got cold bugged AMD's, ATI's, nVidia's, memorys and now we got an cold bugged DFI EXPERT...naah, cant be

Kiwi: R.A02

[celemine1Gig]

Could be that some resistor(s) doesn't like the cold and changes its value under those circumstances. That could cause the higher Vdimm.

[IvanAndreevich]

20 pin PSU?

[Dani]

20 pin PSU?

Yes and...what?

[Lastviking]

20 pin PSU?

Antec TC 550 20 pins works fine when its are modded to 24 pins trust me , Hmm realy wierd Dani...hmm maybe my venus will do the same , i need 4.5v so hehe

[OnLine]

Could be that some resistor(s) doesn't like the cold and changes its value under those circumstances. That could cause the higher Vdimm.

Exactly, because I got the same problem with my Expert mobo.

Normal ambient temp; real Vdimm = Vdimm set in BIOS ( = perfect ).
10'C drop in ambient temp; real Vdimm = Vdimm set in BIOS +0.1V ( or so ).
20'C drop in ambient temp; real Vdimm = Vdimm set in BIOS +0.2V ( or so ).

I figured out it must be some Vdimm resistor or output inductor changing the value under cold air.

Very annoying!

[dinos22]

interesting

what does MBM5 report

[celemine1Gig]

Exactly, because I got the same problem with my Expert mobo.

Normal ambient temp; real Vdimm = Vdimm set in BIOS ( = perfect ).
10'C drop in ambient temp; real Vdimm = Vdimm set in BIOS +0.1V ( or so ).
20'C drop in ambient temp; real Vdimm = Vdimm set in BIOS +0.2V ( or so ).

I figured out it must be some Vdimm resistor or output inductor changing the value under cold air.

Very annoying!

Oh well, if you know the actual temp (i.e. the difference to normal conditions) and also know the resulting voltage offset, then it's not very serious IMHO.

[OnLine]

Oh well, if you know the actual temp (i.e. the difference to normal conditions) and also know the resulting voltage offset, then it's not very serious IMHO.

Basicly that should work, but it still makes it somewhat harder to fine tune.
One more variable to consider.

BH-5/6 mem at 3.5+V will heat up the area in which this "cold bug" appears.

Let's say You fire up 3DMark2001SE.

When benching CL/CH this area will be cold and Vdimm will be high.
At LL/LH the area has gotten heated up a bit and Vdimm will be lower.

Vdimm unstable = no good.

With everything OC'ed to the limit You don't want this temp dependent variable.

[OnLine]

interesting

what does MBM5 report

Oh, I use a DMM to measure the Vdimm and the change is real - and that's what matters.

What MBM5 or ITE SM reports isn't really important.

[Dani]

My SG and MBM5 says the same that DMM says.
From where can I measure VTT on this board?
I would be intresting to see if it is ½ of VDIMM when U put 3.5V in BIOS and it is measuredfrom board like 3.85V, is VTT 1.75V or 1.925V.

This is'nt a good thing, U can burn whole system down with simple cold air

[kcnyc]

Can it be the PSU? I recently received a PSU for a review. It came on a very cold day. I took it upstairs opened the box and took it out to jump start it. And the PSU was mad cold, like ice. Measured the voltages.......
12V rail was at 13V+!!! And so did the other voltages. They were off by at least +0.7V.

I let it sit for a few mins in my house with an ambient temp of about 68F. And after about 1 hour - voltages were spot on 12V, 3V and 5V!

I guess if it is too cold the PSU has a boost in efficiency or something, but I can be wrong. Anybody can explain this?

[Lastviking]

Dani:

http://xtremesystems.org/forums/show...09&postcount=7

[Dani]

Ok, thanks, will measure VTT next time I try with cold air

But this is very weird behaviour, how about if I test my FX57 with DI and EXPERT do like this

Hmm, maybe it's safe to stay with SLI-D

[Dani]

I gonna test today with 12/07 BIOS and some cold air.
..and my Antec PSU is not modded to 24pin, it's original 20pin with 4pin +12V

[Eldonko]

I ran my Expert @ 0C ambient with no voltage issues. BIOS E1207. That is weird, hard to say if it's the board, but my $$ is on the PS.

[OnLine]

I ran my Expert @ 0C ambient with no voltage issues. BIOS E1207. That is weird, hard to say if it's the board, but my $$ is on the PS.

You mean PSU?
We got the same one so that should rule out the PSU as the cause of this Vdimm behavior.

What revision on your DFI expert board?
Mine is the AA0.

[Eldonko]

R.A02 and yep PSU.

[OnLine]

Interesting!

What revision on your mobo Dani?

Maybe rev AA0 is flawed?!!

A $290 ( Sweden ) piece of DFI AA0 crap.
Surely that would rock.