Supermicro Quad G34 VRM temperatures

[sfield]

Readings taken using a Fluke IR thermometer while running sustained folding@home load @ 2.7ghz. No fan pointed at VRMs during readings. With a (loud) panaflo 120mm fan pointed at the VRMs, temperatures are 15-20C lower.

These readings were from CPU4. Other CPUs have similiar readings, including the 2 CPUs which have small VRM heatsinks -- temperatures of the sinks as well as the stray uncovered VRMs are similiar to those pictured. CPU (watercooled) load temperatures are 47-55C.

Working on blocks for cooling these, more on that later.

 

[nwrtarget]

With their location could this be caused by the designers of the board counting on air from the CPU Fan to keep them cool?

 

[Kendrak]

With their location could this be caused by the designers of the board counting on air from the CPU Fan to keep them cool?

I'm quite sure that is the case.

 

[sfield]

The SM chassis tend to have several (very loud) intake fans near the front of the case -- would be mounted near the right of the pictured CPU4 (and CPU2), and would blow across the VRMs and into the CPU cooler. CPU1, 3 typically exhaust out the back of the case, so there would be some additional airflow, but CPU1+3 would likely end up being warmer due to blocked airflow and lack of additional fans in the back of the case.

Something to watch for in any case.

With their location could this be caused by the designers of the board counting on air from the CPU Fan to keep them cool?

 

[jebo_4jc]

Hmm.....this is something to consider as I am swapping my As for noctua coolers.

 

[Red Falcon]

I'm quite sure that is the case.

+1

OP, is there any way to add some heat-syncs to these?

 

[sfield]

I'm thinking a structural aluminum angle or channel segment, mounted to the case, which the copper VRM blocks are in turn mounted to -- would lock in place from the side, or typical downward pressure mount respectively. Given the weight of the copper, and the lack of support close to CPU2+CPU4, this seems like a workable option.

I have some Koolance MVR-40 water blocks, which would be connected to the copper VRM sinks. Alternately, the copper VRM sink blocks could be connected to a larger copper passive cooling scheme.

I looked for longer water blocks suitable to cover 2 CPUs worth of VRMs, but the best I could come up with was 105mm long, whereas ~185mm would be close to ideal.

Raw copper VRM blocks as such:

I also have a copper plate which would take care of the SR5690, SR5670, and 5100 chips -- the main issue here being the noise associated with the 2 chipset fans running at 6000+ RPM.

I'm waiting on a drill press attachment which will allow for sufficiently fine grained and repeatable drilling to machine down the blocks. CNC not needed for this, and not something I have time for currently... Current plan is to cover the VRMs and the inductors (6mm height offset between VRMs and inductors, if calculations are correct).

More on this in a week or two -- this is a low priority background project...

+1

OP, is there any way to add some heat-syncs to these?

 

[Red Falcon]

Sweet, well good luck and let us know how it turns out.

 

[sirmonkey1985]

+1

OP, is there any way to add some heat-syncs to these?

blocks like sfield said or you can go the ultra cheap ghetto way and get those blue vram heatsinks they sell for 10 bucks at fry's. those things work wonders on motherboard VRM's. used them on my biostar 780G board so i could run my phenom II x4 940 on it.

 

[jebo_4jc]

Yeah I'm probably going with some small sinks myself.

 

[Biffa]

Anyone else had issues with this? I'm not sure if it is the VRMs overheating but my box was very unstable over the last few months, and is only now getting back to normal 24/7 stability.

By unstable I mean the entire system would just shut down, thank god for IPMI

I've put heatsinks on the VRM's and upgraded my PSU but it must have been ambient temp and heat related because now its getting a little bit cooler its become more stable.

 

[tjmagneto]

Hmm.....this is something to consider as I am swapping my As for noctua coolers.

Which model of Noctua?

 

[jebo_4jc]

Which model of Noctua?

The 92mm version. I think its the NH-U9DO

 

[lehmann]

Tell me what you guys think of this but i use my finger as a heat tester all over the motherboard. warm is OK uncomfortable warm should be adressed for long term. phucking ouch needs fixed now!

 

[tjmagneto]

Tell me what you guys think of this but i use my finger as a heat tester all over the motherboard. warm is OK uncomfortable warm should be adressed for long term. phucking ouch needs fixed now!

I'm not really sure but don't use your tongue, that's for testing cold surfaces.

 

[Tobit]

Tell me what you guys think of this but i use my finger as a heat tester all over the motherboard. warm is OK uncomfortable warm should be adressed for long term. phucking ouch needs fixed now!

I use another extremity.

If it leaves a mark, it's too hot.
It it stays limp, it's not OC'd fast enough,
If it makes it hard, it's just right.

(Sorry, couldn't resist and no, I'm sober (for a change))

 

[R-Type]

I'm surprised your temps are that high sfield, it definitely seems like the lack of airflow associated with the watercooling is to blame here. With my 2p dodeca system folding at 3.2ghz, the vrm temps max out at about 78C measured with an IR thermometer.

At least on the supermicro boards they line your VRM's up nicely for sinking, the Tyan's have them interspersed with the chokes and would be a huge pain.

 

[tjmagneto]

Tell me what you guys think of this but i use my finger as a heat tester all over the motherboard. warm is OK uncomfortable warm should be adressed for long term. phucking ouch needs fixed now!

Seriously, I'm more into using an IR thermometer but that's pretty close,

 

[Untitledone]

Seriously, I'm more into using an IR thermometer but that's pretty close,

Some of us are not that fancy

 

[R-Type]

Some of us are not that fancy

You can get pretty fancy (and within a couple C accuracy) for next to nothing.
http://www.amazon.com/HDE-Temperatu...3FS4/ref=sr_1_1?ie=UTF8&qid=1318597052&sr=8-1

Its no Fluke but I'd wager its a damn sight more accurate than your finger (or other appendages in Tobit's case).

 

[Untitledone]

If it was not amazon I would have totally picked that up .

 

[firedfly]

If it was not amazon I would have totally picked that up .

You have something against amazon?

 

[R-Type]

Yeah I don't get it, why would you not want to order from what is easily the most consistently loved major etailer around?

 

[jarhead]

He's sticking it to the man.

 

[Untitledone]

No, someone within amazon sold my account and I found out I had $600 worth of battlefield bad company 2 digital downloads. They would not tell me anything, and closed my account. I had to go through my bank and do a charge back to get my money back. I used their service once in august of 2010 to purchace a $20 roll of 800' of 24 gage manget wire. Five months later in February of 2011 I get $600 with of fraudulent charges. I had only logged into my account on one or two occasions. I will not use amazon again. I thought it was kind of odd that I had 60 emails from amazon in one day, and that is what alerted me to the problem. Obviously my situtation is an isolated incident, and not true for everyone, but I personally will not use them because of it. I had no attachments to them to begin with because I had only used them once. This is the only time I have had my credit card or account information stolen from an online retailer. My bank informed me that nearly half of their fraud reports are from amazon and their affiliates, and this is a very common occurance.