DFI NF4 the UTT/VX/BH5 KILLER READ IF YOU HAVE DFI

Quote:

---

Originally Posted by wtz54321

Although most of the RAM module maker gurantee their memory to be rated at 3.5V , I don't think it is correct for user to think that it's a safe and problem free behavior to work 7/24 under that voltage ... Most of the DDR SDRAM only gurantee 2.8V maximum absolutely rating working voltage ... Anything beyond that is an unknown status ... :fact:

---

Nor does any cpu manufacturer endorse extra 100mhz.... :stick:
This typical sympton of UTT, BH-5 dying in relatively short time with *low* voltage (by some standard :P) is something that hasn't happened as frequently before. It's gotta be something with that 5v VRM design DFI used in conjunction with rest of component... I can't say what it is specifically but after repeatedly running memory well around 3.5v on VX, BH-5, quite a few UTT for months at a time, and having first time of this issue on DFI NF-4 ultra makes me be wary of using it's 5v dip. Right now I am sticking to what Big Toe suggest for needed voltage. Let's see how well these 4 sticks last over months now...

Another possible reason is the heat under high voltage which will cause/accelerate the electron migration behavior in semiconductor and finally killed the dram , we all know that using 5V Jumper in DFI NF-4 will generate extremely high temperature compare to using directly 3.3V DRAM mod or by using OCZ DDR Booster ... And the hot spot area is very close to the DIMM1/DIMM2 ... If the DRAM is killed by improper voltage , K8 DRAM controller should be more sensitive to be damaged first ... Also why there is only one slot killed the DRAM ? :confused: :confused:

Applying high gate voltage will cause the semiconductor to be dead instantly by " Punch Through " effect , this is a immediately situation once the voltage is too high for the gate oxiode even it's in a very very short time period ... However the high voltage will not cause the electromigration behavior , only the heat followed by the high working voltage will cause this kind of semiconductor behavior which will last for a certain period until the semiconductor finally shotred caused by electromigration ... :fact:

Quote:

---

Originally Posted by wtz54321

Applying high gate voltage will cause the semiconductor to be dead instantly by " Punch Through " effect , this is a immediately situation once the voltage is too high for the gate oxiode even it's in a very very short time period ... However the high voltage will not cause the electromigration behavior , only the heat followed by the high working voltage will cause this kind of semiconductor behavior which will last for a certain period until the semiconductor finally shotred caused by electromigration ... :fact:

---

Not even in consideration when memory is running cooler than "stock" setting at supposed stock voltage with no cooling due to good 2x60mm fan righ above it. I've had way too many memories that didn't have issues such as this at HIGHER voltage to think it is only issue with memory.
One thing that will kill memory other than extreme voltage or extreme heat is extreme fluctuation (rapid) of voltage supplied. (2-3 months of 24x7 isn't that long) To think all 4 sticks went out at same time is fault of memory alone is.... :stick:
Something happened that shouldn't have happened. Also, this issue happens when cold booting as I had them running well until I shut down and upon start... no boot...

:confused: Extreme fluctuation (rapid) of voltage supplied is the samething as extreme voltage when we are talking about the punch through effect of transistor . Even in a every short time period , once the voltage exceed the limit of that the gate oxide can handle , punch through will happened immediately .... If the memory is killed by the voltage not the heat after a certain time period , the only possible reason is that after 7/24 usage lasting for a long time , the VRM of the board start to become unstable as the beginning ~~~ This is the only possible reason that I can think about ...

Well, I must add that I had the exact same problem most folks experienced here too. 1 stick of my Twinmos SP (UTT) mysteriously died after running 260Mhz at 3.4vdimm for just a couple of days. Got the 3 LED lights indicating memory not detected. Cleared CMOS, reflashed BIOS and others did not help. Decided to try just the faulty stick in orange slot 2 and somehow manage to boot. Didn't work anymore though after that.

Had my BH6 sticks in there since the incident. Several people mentioned flashing to new BIOS's and using yellow dimm slots have helped revived the 'dead' stick. Hopefully my Twinmos sticks aren't dead :(

Quote:

---

Originally Posted by Quanticles

The lovely moderator said there was a thread on the topic already, but failed to post which one it was. Very nice of him. Closed the thread too.

---

yea and I also pmed him as well, I got a read receipt but no response.
Also to all those that are having problems with their ram, what bioses are you guys using so we can narrow down what is causing this problem

i've tried every bios from 310 upwards.

no I mean with what bios did ur ram die on? I'm trying to see if anyones ram has died while using the new 414 or 510 bioses

My RAM was fine then I flashed to 414 and the sticks started having problems cold booting, but still I could get them booted if I rebooted a few times and swapped them around.

I flashed back to 310 and it was fine again until the cold boot problem came back and this time I had to boot using 1 stick. Now I've got 1 bad stick that doesn't want to boot at all on this mobo - no matter what BIOS I try. The other stick is still 100% fine.

so while on 414 there was cold boot issue but no ram killing correct?

Quote:

---

Originally Posted by P_1

so while on 414 there was cold boot issue but no ram killing correct?

---

That's correct, I may have had the cold boot issue before I flashed to that BIOS - just that was when I first stumbled across the cold boot problem.

I have this happeneing to me too. 2 sticks of TwinMOS SpeedPremiums and 2 sticks of TwinMOS regulr (non-SP). They are not actually dead, I can boot into windows fine, but they are very unstable and once in a while cause BSOD even at 200 MHz and 2.5-3-3-x timings. They both used to be P95 stable for 20+ hours at 260-263 MHz, 3.4V, 2-2-2-5-1T.

First one, TMSP ones, went bad when i flashed from the stock (1/25) BIOS to the 310 BIOS. I don't really know what happened (I followed the instuctions for the flash down to the letter), but after the flash it the sticks became unstable in windows.

Second one, TMII ones, worked fine after the flash (after I switched from the TMSP sticks). I actually scaled it down to 252 MHz and 3.2V, since I read about these TwinMOS sticks die from several people, and everyone was suggesting not to use over 3.2V. In any case it was stable at that speed and I used it for about a month, until I decided to try and push my CPU higher (it was at 9x280 with 9:10 RAM devider). After a couple of unsuccessful rans of P95 I decided to check my stability at the speed I was running and to my surprise P95 failed. Tried it several times, with the same result. So I decided to run Memtest86 again, and it failed. Then I tested each stick separately and one of the sticks was still fine (could still do 252 with 2.3V or 263 with 3.4), but the other one was erroring like crazy. I was getting errors in the very first test, as soon as Memtest86 starts. Unfortunately i didn't note which stick was good and which one was bad (meaning which DIMM slot) so I can't confirm or deny some views expressed here about 2nd DIMM.

Right now I am using Patriot XBLK TCCD, but I am still on +5V line and I am using 414-3 BIOS (I flashed the bios after switching over to TCCD). And to be honest I don't know what I should do next. Should I switche back to +3.3V line? I am kidn of afraid to do that after reading that that might actually damage RAM. I also want to flash to one of the newer BIOS (510-1, 2, 3) but I am not sure which one to try.

I would really like to get some feedback from DFI people. Once again I am not here to trash DFI. I think it is a great company and I think it is a great board. No doubt the cater overclocking community. But there are a lot of people out there having these issues, and you can't simply discard it.

Does anyone have a link to the thread in dfi-street that the moderator was referring to? I couldn't find anything.

Thanks.

Quote:

---

Originally Posted by hovo73

Does anyone have a link to the thread in dfi-street that the moderator was referring to? I couldn't find anything.

Thanks.

---

http://www.dfi-street.com/forum/showthread.php?t=10822

http://www.dfi-street.com/forum/showthread.php?t=11963

http://www.dfi-street.com/forum/showthread.php?t=11107


There is a few. :toast:
Thinking its a desing issue and we might just have to deal with it. As what they going to do? refund everyone a new board or something. :D right

Peace Nugz

Have all you guys with dead RAM been on the 5 volt line? and not using the 3.3 volt mod? I say this because I was on the 5 volt when I killed all my RAM!
A friend of mine has had 23 sticks of UTT back out of a batch of 100 Killed by the 5 volt line and DFI with most people using 3.2, 3.3 or 3.4 (BUT ALL 5 VOLT LINE!) I am not here to slag DFI my 250UT Board did great as did my NF2 Ultra I just want an answer to this problem! All posters tell us if you used the 5 volt line when your RAM Broke!

I was on the 5v line too.

i could swear my multimemter reading jumped 5.34v on the vdimm when i was checking to see if it flucuates while booting and during memtest and i think it was after a bios flash.

Hi Wingz,

Where you using the 5 Volt line for memory when you got this jump? Or was it from the 3.3 volt line?

I was on the +5V too.

I started testing with +3.3V line. I got up to around 220 to 230 MHz with 3.2V then I switched to +5V line. When I did, something interesting happened. With still 3.2V but now from the +5V line rather than +3.3V one, it went up to about 250 - 255 MHz. At the time I hypothisized that this is because at 3.2V the voltage is not very clean coming from the 3.3 line because it is at the very edge.

In any case, yes I was using +5V line.


Nugz, thanks for the links. It could be a BIOS issues rather than a design issue, and that can be fixed through BIOS updates. People can also RMA their boards, I am sure DFI doesn't want to have 1000 RMAs, so they should be looking into this problem. I wander if they are.

Quote:

---

Originally Posted by pazza316

Hi Wingz,

Where you using the 5 Volt line for memory when you got this jump? Or was it from the 3.3 volt line?

---

i been using the 5V.

I ask about the 5 volt because I tested with volt meter tonight and on boot got 5.2 going through the DIMMS these where old crappy ones for test purposes so didnt care if they got blasted but 5.2 the same! and this happened with both my DFI NF4 ULTRA D boards!!! I am on 3.3 volt rail form now on, well untill DFI have sorted the problem! :slapass: :slapass: :slapass:

Quote:

---

Originally Posted by pazza316

I ask about the 5 volt because I tested with volt meter tonight and on boot got 5.2 going through the DIMMS these where old crappy ones for test purposes so didnt care if they got blasted but 5.2 the same! and this happened with both my DFI NF4 ULTRA D boards!!! I am on 3.3 volt rail form now on, well untill DFI have sorted the problem! :slapass: :slapass: :slapass:

---

What do you mean, "5.2 going through the DIMMS" ??? Where did you happen to measure that?

Just for more input to the thread, I was using 2/18 bios and the 5v line running 3.4v when my stick died. Only flashed to 4/14 bios after I couldn't get it to boot. I am not even going to try to run any more UTT until there are some answers to this issue. Currently have my OZC plat. rev 2 at 290 2.5-3-3-8@2.8v. I think I'll stick with this instead.

I have an opinion to give...
Since bios 316 or 310, oscar changed something in the bios, that made my bh.5 ram to stop working on slots 2 and 4.
After i flashed the new bios, i coulnd't get into windows no way.
And could only boot after a cmos flash if i booted with some utt first and changed cas to 2, cus as you know bh.5 doesn't work with 3.

This was strange and after a few hours, when i was about to turn to 2xx bios or whatever i had, i tried the yellow slots, and bam! everything was back on track.
If i plug them again on orange slots, it won't boot to windows not even at 200 with relaxed timings.

Try and post back.

I´m using a direct 3.3v from psu rail, only used 3.5v for testing like 1 hour.

any short/long term side effects of having the 3.3v line set to 3.6-3.7v on the psu?