PDA

View Full Version : DFI NF3 LP UT hardware V-Dimm Mod (5V rail). Need someone with a MultiMeter...

celemine1Gig
07-28-2005, 03:19 PM
Hi guys,

there was a thread some time ago about the VDimm OVP on this board, here to be exact:

http://www.xtremesystems.org/forums/showthread.php?t=57825&highlight=dfi+nf3

Well, now that my exams are history :D I had the time to try some more on this board.

As I already posted in the old thread, I wanted to power the vdimm from the 5V rail.
Well, the problem is that something must have gone wrong, or I'm missing something. One of these two possibilities. ;)

As posted before, I removed the 2 mosfets directly under the dimm slots to replace them with higher rated (40A and lower RDSon in comparison to the stock ones with 20A and a higher RDSon) FETs.
That worked, but all I got after the mod was a vdimm of max. 2.08V, thus no boot up. I tried every single hardware mod to raise the vdimm (LM358), but nothing worked. It always topped out at ~2.3V (without anything installed, just the naked board, and 2.08V with RAM, CPU etc.). I think that I could've damaged an SMD part while soldering in the new mosfets (or even the mosfets) and that could be the reason for this strange behaviour.

Now I need someone with a multimeter, who could measure some things for me and tell me the values. Perhaps, I can make the 5V powered Vdimm work with your help.

Please look at the pic and do what is marked:

http://www.fh-augsburg.de/~madocer/DFI.jpg

The part where you should measure resistance is a diode if I'm not mistakin. I ask you to measure because I think I could've damaged it. I get a reading of 818 Ohm on both sides (the single contact is ground, and I get 818 Ohm from each contact on the other side to ground).

PLEASE MEASURE WITH THE BOARD NOT POWERED ON!!! NOT EVEN STANDBY POWER! No LED on the board should be lit when measuring.


In the meantime I just did the VDimm = VIO (3.3V rail) to see if the rest of the board, except for the vdimm regulation, was OK. And it works like a charm. A little test at 375MHz HTT proved that. :D

Big THX in advance!

Greets

Ingmar
a.k.a celemine1Gig
craig588
07-28-2005, 05:52 PM
The connected points are connected on my board, and my board has 818 or 816 ohms on both points as well, I don't remember which, but I don't think 2 ohms out of 800 really would matter.
persivore
07-29-2005, 12:13 AM
You need to use the diode test function on your multimeter. The resistance function of a multimeter doesn't give out enough voltage to bring a diode into conduction, so it will just appear to be an open link to a multimeter ;)
celemine1Gig
07-29-2005, 04:43 AM
OK guys, thank you very much so far. Thx craig for measuring and thx persivore for the heads up. :)

Looks like I didn't do much wrong. Seems to be a small error that makes the board behave that way. Will try to find out today. I have the bad feeling that one of the mosfets could be damaged, but I don't really know for sure. Will take care of it today as far as I can.

Perhaps I'll have further questions later. Wish me luck and thx again. :toast:
persivore
07-29-2005, 04:47 AM
What is the part number of the origional MOSFETs and the ones you replaced them with?
celemine1Gig
07-29-2005, 05:41 AM
Those were the original units (datasheet):

http://www.fh-augsburg.de/~madocer/SD(DU)9916.pdf

And I replaced them with these:

http://www.fh-augsburg.de/~madocer/irlr8503.pdf

I hope I didn't choose the absolutely wrong ones.
persivore
07-29-2005, 05:51 AM
Those were the original units (datasheet):

http://www.fh-augsburg.de/~madocer/SD(DU)9916.pdf

And I replaced them with these:

http://www.fh-augsburg.de/~madocer/irlr8503.pdf

I hope I didn't choose the absolutely wrong ones.
The only thing which I can see which might be causing a problem would be the higher gate-source (Vgs) voltage (its +/-12v on the origional 'FETs and +/-20v on the new ones). Can you check the voltage at the gate pins on the MOSFETs?
celemine1Gig
07-29-2005, 09:53 AM
OK, no wondern that it didn't work. Had a bit of solder shorting Drain and Gate on the input mosfet. D'Oh!
That's sorted out now. Now one last question: Could you please check the capacitor between the two vdimm mosfets. It's the 1000µF 6.3V cap directly under the dimm slots. Could someone please check on the backside of the board where it's soldered in. There are lines of solder (I guess they're good for keeping the signal clean, but I don't really know). What I need to know is, are these solder-lines connected to the +side of the cap only, or are some also connected to the -side (i.e. ground)???


Edit:
OK, got everything fixed that was wrong before (I hope I really got everything :D ). Wish me luck, will try it out soon. Hope that I won't ruin that good board.
celemine1Gig
07-30-2005, 09:18 AM
OK, Mod works. Vdimm is now powered from the 5V rail. Had 3.7V running a few minutes ago.
The only problem that I still see is power dissipation.

As persivore said, VGS is the problem. I'll just sink the FETs and let them be.

Another small update:

Soldered in the original Mosfet as a replacement for the new one which I thought would get too hot. To my surprise, it got exactly as hot as the new one. :( Now, it's time for some extra cooling. Will have to cut some proper heatsinks and will then test for absolute stability. Will have a guide up, perhaps until next weekend. Works fine so far, except for the increased heat that I don't really like.
celemine1Gig
08-05-2005, 06:43 PM
Update time again: :D

As I already said, the mod works great. VDimm mod finshed in the meantime,too. Now I have voltage control from 2.6 to over 4.0V with the help of some potentiometers. :)
That's simply great. Cooling mod for the mosfet's also nearly done. Will post final pics when I'm really really satisfied with the results. So far everything is looking good.
persivore
08-06-2005, 02:00 AM
Do you still get the 'cold boot' OVP problem with the Vdimm now?
celemine1Gig
08-06-2005, 04:36 AM
Had no time to fully test that out yet, but as the board previously always booted up with the default 2.5V VDimm and then did one cycle, rebooted and the bios set itself back to the previous value that had been set in BIOS, I think the VDimm mod should've fixed that, just like on the NF4 DFI.
Ahh, and no problem with high VDimm so far if you wanted to know that. Had it boot up at 3.7V and higher without a problem. So, there doesn't seem to be a real OVP on that board. More likely the PSU giving in due to the high 3.3V rail (for the poeple who had problems before).

Will try to finish the cooling mod ASAP and then I'll do some extensive testing again. If that works out fine, I'm going to write a little guide.
zaubara
08-07-2005, 07:08 AM
Can't wait to get my hands on this! :)
celemine1Gig
08-07-2005, 03:18 PM
Final tests today looked good. I'll finish the rest of the cooling mods (new nforce3 chipset heatsink and so on) tomorrow, and then I'll start writing the guide.
celemine1Gig
08-11-2005, 06:24 PM
Okidoki, as I'm running late again, another small teaser for all that are still enjoying the board, waiting for the mods:


http://www.xtremesystems.org/forums/attachment.php?attachmentid=35667&stc=1

Explanations will follow. Just for the start, the huge black wire, leading to the mosfet on the left, is what it's all about. That's where the 5V come into play for Vdimm. ;)

BTW, finished all my tests (ran Memtest, VDimm of ~3.5V for a whole day without any problems).