DFI using 5v rail for ram not 24/7 stable?

**Ailleur** · 03-22-2005, 07:04 PM

Well, i found a purpose for the big wire clutter. This 80mm fan keeps the mosfet cool enough to touch and hold after an hour of half life 2, which is all that matters to me!

http://www.alienhax.us/uploads/Picture 006.jpg

I would definetly say a fan is a must. With one, im pretty confident about runing the 5v jumper 24/7, especially after reading the caps were stated to live through 105C. Id say the temp is about half that now.

**EMC2** · 03-22-2005, 10:47 PM

Originally Posted by cantankerous

I wish there was a way to get it working from the 3.3v rail of the psu.

There is... There have been posts in threads around here about one way to do it. Other ways will be forthcoming at a future time.

And with cooling on the area, it is safe to use the 5V 24/7 if you're using 3.x for Vmem... wouldn't do it with Vmem set to 2.6 or such though.

**HKPolice** · 03-22-2005, 11:41 PM

Repost: What about the 3.3v rail = vdimm mod? Can it be safely done when set to use 3.3vrail for vdimm instead of 5v rail?

This is the most stable way of getting voltage to the dimms IMO.

**Eversor** · 03-23-2005, 03:28 AM

Originally Posted by HKPolice

Repost: What about the 3.3v rail = vdimm mod? Can it be safely done when set to use 3.3vrail for vdimm instead of 5v rail?

This is the most stable way of getting voltage to the dimms IMO.

Yup it can. Tough the bios will not read the dimm voltage right.
Like from 3.3v at 3.2v i get +-3.15v. with vdimm = 3.3v i get 3.25, tough the 3.3v rail reads 3.3v

But it works. Just don't use the 5v rail after u do the mod.

**HKPolice** · 03-23-2005, 10:39 AM

Originally Posted by Eversor

Yup it can. Tough the bios will not read the dimm voltage right.
Like from 3.3v at 3.2v i get +-3.15v. with vdimm = 3.3v i get 3.25, tough the 3.3v rail reads 3.3v

But it works. Just don't use the 5v rail after u do the mod.

Great, then what's the fuss about? Just do this mod and crank up the 3.3v rail to 3.6v for 24/7 usage

**S0nic** · 03-23-2005, 11:46 AM

so how would I get 3.3V vdimm using the 3.3V rail? What vdimm do I chose in the BIOS?

**mdzcpa** · 03-23-2005, 05:44 PM

Originally Posted by HKPolice

Great, then what's the fuss about? Just do this mod and crank up the 3.3v rail to 3.6v for 24/7 usage

But the real question is why bother?

With the 5v jumper right there, the 3.3 mod is a waste of time.

**S0nic** · 03-23-2005, 07:27 PM

Originally Posted by Eversor

Yup it can. Tough the bios will not read the dimm voltage right.
Like from 3.3v at 3.2v i get +-3.15v. with vdimm = 3.3v i get 3.25, tough the 3.3v rail reads 3.3v

But it works. Just don't use the 5v rail after u do the mod.

aren't those options disabled if u have the jumper in the 3.3V position? isn't 3.1 max?

**Jasonxxx** · 03-23-2005, 08:29 PM

Can you turn your 3.3v line up to say 3.5 or 3.6v and be able to obtain say 3.4 or 3.5v w/o using the 5v jumper?

**Revv23** · 03-23-2005, 08:30 PM

nope the options are still there, just wont work without a higher 3.3v rail.

**S0nic** · 03-23-2005, 08:35 PM

Originally Posted by Revv23

nope the options are still there, just wont work without a higher 3.3v rail.

sounds good (can't test now because I'm still waiting for my VX), is there any disadvantages to doing it this way? if we can get 3.4-3.5 vdimm from our (adjusted) 3.3V rail then why would anyone want to use the 5V rail?

**Revv23** · 03-23-2005, 08:42 PM

because who wants to take the time to mod thier psu to run thier 3.3v rail out of spec when you can just move a jumper and get same results if not better?

**Jasonxxx** · 03-23-2005, 08:50 PM

The only reason I thought of that is because it causes more heat to drop the voltage from 5v to ~3.5v than to drop .2 or .1 volts....

I'm assumming that you will only be able to reach with in .1 to .15 volts of your 3.3v line voltage, ie... if you adjusted you 3.3v line to 3.6v then you could ony get 3.4 or 3.5v max...Any able to confirm this?

**S0nic** · 03-23-2005, 08:55 PM

my PSU has adjustable rails (OCZ PowerStream 520W), does that mean it's pointless for me to use the 4V rail even with cooling?

**Revv23** · 03-23-2005, 09:03 PM

Originally Posted by Jasonxxx

The only reason I thought of that is because it causes more heat to drop the voltage from 5v to ~3.5v than to drop .2 or .1 volts....

I'm assumming that you will only be able to reach with in .1 to .15 volts of your 3.3v line voltage, ie... if you adjusted you 3.3v line to 3.6v then you could ony get 3.4 or 3.5v max...Any able to confirm this?

I cannot confirm 100% but i am fairly sure you are right, but again, with the rail out of spec you will need to be cooling the fets anyways, so i dont understand why you wouldnt just use the 5v jumper, and save yourself the hassle of running your PSU out of spec.

**S0nic** · 03-23-2005, 09:16 PM

Originally Posted by Revv23

I cannot confirm 100% but i am fairly sure you are right, but again, with the rail out of spec you will need to be cooling the fets anyways, so i dont understand why you wouldnt just use the 5v jumper, and save yourself the hassle of running your PSU out of spec.

Heat is why I'm considering using the 3.3V rail, if it's gonna generate the same amount of heat then obviously I'll just use the 5V rail..

**DarkManX_BG** · 03-24-2005, 04:34 AM

Originally Posted by Revv23

nope the options are still there, just wont work without a higher 3.3v rail.

They aren't there for me

**HKPolice** · 03-24-2005, 03:46 PM

Nope... 3.3v rail mod is the most stable source of power. Measure the vdimm yourself with a multimeter, even when using the 5v rail to supply 3.3vdimm, it will sag during memtest.

Using the OCZ PSU, you can adjust the 3.3vrail on the fly from outside the case, which makes this mod even easier to adjust.

Also, you don't have to worry about cooling those ram mosfets anymore, which pumps more heat into your case anyways.

Originally Posted by mdzcpa

But the real question is why bother?

With the 5v jumper right there, the 3.3 mod is a waste of time.

**mdzcpa** · 03-24-2005, 04:06 PM

Originally Posted by HKPolice

Nope... 3.3v rail mod is the most stable source of power. Measure the vdimm yourself with a multimeter, even when using the 5v rail to supply 3.3vdimm, it will sag during memtest.

Okay, I just tried this and have no sagging at all running memtest using the 5v jumper

I'm not sure your generalization fits here.

**mdzcpa** · 03-24-2005, 04:11 PM

Update - Okay...there is some sagging. After 30 minutes I am getting a sag from 3.4v to 3.39v using a multimeter.

If that's the sag you are referring to, .001v isn't worth the time to do the mod.

**S0nic** · 03-24-2005, 04:20 PM

I guess I'll try doing it with the 3.3V rail first since I don't have to mod my PSU, if it doesn't work or if it generates the same amount of heat then I'll just use the 5V rail..

**Eversor** · 03-24-2005, 04:40 PM

Originally Posted by S0nic

I guess I'll try doing it with the 3.3V rail first since I don't have to mod my PSU, if it doesn't work or if it generates the same amount of heat then I'll just use the 5V rail..

The mod will produce ZERO heat from the mosfets, since they won't be transforming voltage anymore. In fact i think the vdimm vreg will be disabled, since the vsense it's receiving is higher than it's supposedly supplying thus lowering the voltage till 0.

The only thing that can heat up is the vtt vreg wich will still produce vtt voltage based on the 3.3v it is receiving.

At least the last time i did a mod on a nforce 2 it worked in this exact way.

@mdzcpa

You got me wrong, using 5v vdimm even at 3.6v, during memtest, and WITH a fan, the mosfets don't get so hot, but the pcboard burns my fingers. So I'm not confident to use it 24/7.

Don't want my DFI Ultra @ SLI to die on me prematurely

Regards

**Formann** · 03-24-2005, 04:55 PM

Soldering the 3.3v rail directly to the vdimm circut will never provide RAM with the clean voltage that the vdimm circut on the MOBO does. Fluctuations on the 3.3v rail is normal, and this can affect the stability of the RAM.

The voltages on the DFI nf4 isnt fluctuating at all. Im really impressed by the voltage regulation on this MOBO. Its perfect. No fluctuations what so ever. I dont get why anyone would want use a 3.3v mod on this DFI.

When im running 3.3v the mosfets are about 50-55c. This temp is no match for them as they are made to run twice and over three times as hot without any problems.

**mdzcpa** · 03-24-2005, 06:51 PM

Originally Posted by Eversor

@mdzcpa

You got me wrong, using 5v vdimm even at 3.6v, during memtest, and WITH a fan, the mosfets don't get so hot, but the pcboard burns my fingers. So I'm not confident to use it 24/7.

I understand your concern. Mine just doesn't get quite that hot....and if it did, I'd be inclined to go the same direction as you have

However, that said, the point is now MOOT for me. I'm not going to continue with the OCZ VX anymore. I'm going to sell it. After trying the Gskill LA, some older Corsair BH5, and then the OCZ VX with some decent results, I decided to try a pair of trusty Gskill LE. BAM! Right to 295 2.5-3-3-7 at only 2.8v rock stable. It would do way more but my 3500+ chip is maxxed at 295 x 9 for 2650. I was running the VX at 260-265...but it took 3.5v +/- to get there. And even at 2-2-2-8 it could not compete with the Gskill LE at 295 2.5-3-3-7.

**EMC2** · 03-24-2005, 07:55 PM

--- Eversor ---

Where specifically are you getting this hot spot on your MB, in the area of Q60 or over by D8 or ???

Thread: DFI using 5v rail for ram not 24/7 stable?

Thread Tools

Search Thread

Rate This Thread

Display

Bookmarks

Bookmarks

Posting Permissions