separate coolers for CPU and VRM area?

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
marcus_helsinki
Posts: 15
Joined: Mon Aug 09, 2010 1:23 pm
Location: Finland

separate coolers for CPU and VRM area?

Post by marcus_helsinki » Fri Aug 13, 2010 12:58 pm

BACKGROUND:
With many boards, if we think about long term component life, the VRM area gets very hot while the CPUs are cooled with modern tower heatsinks very effectively.

PROBLEM:
With a tower CPU cooler, the VRM area may get too hot.
With a top-down CPU cooler, effective coooling of VRM area with an airflow THROUGH a dense heat sinks requires a high pressure capacity of the fan, which means high RPM and high noise. I have compared the airflow from a freely blowing fan with the air fow from a fan blowing through a heatsink, and the result was that the heatsink disperses a part of the flow to sides but also weakens the flow very much.

SUGGESTION:
using a fan blowing freely at VRM area, and a tower heatsink for CPU without a fan or with a very low RPM fan
BENEFITS:
no fans blowing through the fins of heatsinks causing noise; only free flowing very silent fans with low RPM; an effective cooling of VRM area and CPU
REQUIREMENTS:
space to install the fan for VRM area. The fan doesnt have to be installed at ready-made places in the chassis but usually has to be installed, and possibly also usually can be installed with some self-made solution.

OTHER SOLUTIONS OF THE SAME TYPE:
Some of this type of solution can be seen in Noctua's top-down blower, which allows a part of the air flow pass the finned area of the heatsink and hit the PCB directly, but, I think, possibly from too high a distance from it, if low RPM is going to be used.

faugusztin
Posts: 450
Joined: Mon Mar 29, 2010 2:47 am
Location: Bratislava, Slovak Republic

Post by faugusztin » Fri Aug 13, 2010 1:19 pm

SUGGESTION: do not buy boards which don't give a damn about VRM cooling (lowend boards and Intel boards).

Seriously, i have hard time figuring out why does anyone buy Intel boards - they usually have bad layout, bad cooling, zero OC support...

marcus_helsinki
Posts: 15
Joined: Mon Aug 09, 2010 1:23 pm
Location: Finland

Post by marcus_helsinki » Fri Aug 13, 2010 1:35 pm

DQ57TM has double dual DVI and low power consumption, and excellent documentation. Installation was very easy and works fine. There may be better boards too but I don't have the time to test them.
Intel boards may be planned so that they work fine with the stock cooler while other boards are more for private consumers?
Another reason to buy Intel board is that because Intel SSDs have the reputation being the most reliable and the processor is an excellent Intel's processor (office / web use / low power)m so having also the motherboard from Intel may produce asystem that fits together very well.

ntavlas
Posts: 811
Joined: Mon Jul 16, 2007 2:35 pm
Location: Greece
Contact:

Post by ntavlas » Fri Aug 13, 2010 1:39 pm

If you don`t overclock the vrms are not going to get that hot. I`ve touched the power mosfets on a couple of my boards (p45 and 780g, no heatsinks attached) and while hot, they didn`t burn my fingers. Even with tower coolers they do get some airflow, especially if the cooler fan is relatively close to the motherboard.

lodestar
Posts: 1683
Joined: Fri Aug 05, 2005 3:29 am
Location: UK

Post by lodestar » Fri Aug 13, 2010 1:59 pm

The practice of overclocking CPUs, particularly Core i7s can cause a significant rise in power draw and VRM temps. But then again most people doing this are gamers and are likely to have high air flow cases, and performance motherboards with multiple power phases and VRM heatsinks. If you look at a gaming case like the Antec 902 it has an optional 120mm side fan. This is positioned to blow towards the the processor/VRM area of the motherboard, and can be made more effective by running it in sync with the CPU fan.

So overall I don't think this is going to be a problem in practice.

faugusztin
Posts: 450
Joined: Mon Mar 29, 2010 2:47 am
Location: Bratislava, Slovak Republic

Post by faugusztin » Fri Aug 13, 2010 2:40 pm

marcus_helsinki wrote:Another reason to buy Intel board is that because Intel SSDs have the reputation being the most reliable and the processor is an excellent Intel's processor (office / web use / low power)m so having also the motherboard from Intel may produce asystem that fits together very well.
This is a sort of logic i seen before which led to questions "Will i have problems having a NVIDIA card in AMD board with AMD chipset? Will i have problems having ATI card with Socket 775 board with NVIDIA chipset?".

Intel SSDs work good on any board, so do Intel CPUs work good on any compatible board from any manufacturer. Intel boards are not better, more like contrary, they are actually the lowend boards for specific chipset.

You can continue buying Intel boards, but they aren't the best boards.

marcus_helsinki
Posts: 15
Joined: Mon Aug 09, 2010 1:23 pm
Location: Finland

Post by marcus_helsinki » Sat Aug 14, 2010 2:03 am

Intel boards might be meant for office computers, and for commercial, larger builders.

Intel has many excellent technical documents about each their board, but I have noticed, other manufacturers have only one poor...medium quality document (not always real english, unclearly, not precisely written, much things missing). A builder may want to know the properties of the board.

Is this VRM temp thing really only Intel's problem? SPCR reviewed a Gigabyte board and others and the VRMs reached even 100C!

At the Intel board, there is temperature sensor in the VRM area, is this found in boards from other manufacturers?

So, is this problem totally neglected by other manufacturers and high temperatures are not shown?? (this is only a question, are there temperature sensors in other board's VRM area?)

Or, does Intel take the heating up (long term durability) of VRMs more seriously than other manufacturers and give lower limits?

On the other hand, ASUS boards have many stages in their VR and fine looking heatsinks on them, which would mean a lot better design?
Or, are all parts in VRMs really cooled enough with heatsinks only on transistors?

One ASUS board consumed a mystical 5 W more in a SPCR review, which might warm up the components.

So, has anyone done a good comparison/measurements of temps / assesment of quality? I haven't. So, what I wrote is questions or assumptions which should be commented / rejected / accepted.

yuu
Posts: 132
Joined: Sun Aug 26, 2007 5:03 pm
Location: eu

Post by yuu » Mon Aug 16, 2010 6:36 pm

The northbridge part of the cpu is powered by single VRM phase. so it could be getting hot, according some reviews <2A, and another ~2A the CPU part while burntest running, so that is 48 watt total and 80-90% efficiency included, 5-10 watts to dissipate by the VRM. That is less than 1W per each mosfet.

With mosfets pointing up and little contact between them and the PCB, intel must be having some pretty efficient design >90%, or is intel crasy. Mosfets are supposed to be cooled by the motherboard, seen in gigabyte 2oz deisgn or it helps.

Adding fan doesn't mean you are cooling the mosfets, you could be cooling mostly the thermistor reporting temperature.

I have the E8400 that lacks the HT and 32nm improvements of i5, and OC'd 3.6Ghz is barely 26.5 Watts Cinebench 11.5, and 17Watts in Winrar Bench, so that includes in theory a maximum of 2-4 watts dissipated by VRM at any given time. Worrying about cooling that is just ridiculous.

marcus_helsinki
Posts: 15
Joined: Mon Aug 09, 2010 1:23 pm
Location: Finland

Post by marcus_helsinki » Wed Aug 18, 2010 12:18 pm

Which alternative would You consider is better when taking sound pressure level into consideration:
1) passive cooling of CPU with a large tower heatsink + a nearly silent fan blowing at components
2) a fan blowing rather fast through a heatsink

Post Reply