All these vertical tower heatsinks, what about the mosfets?

Shaman · Post by **Shaman** » Wed Jan 31, 2007 12:21 pm

Mosfets and voltage regulators, don't they need as much cooling nowadays? Or does everyone running tower coolers simply accept the fact that their motherboards will have a sorter life span as a result of insufficient cooling?

ryboto · Post by **ryboto** » Wed Jan 31, 2007 12:27 pm

what I read was that the VRM circuitry(MOSFETS) can safely operate above 100C, though I'd never be comfortable with that temperature, it lets me know I don't have to worry about a 50-60C PWM temp.

nici · Post by **nici** » Wed Jan 31, 2007 12:38 pm

VRM circuitry blowing up could be a problem with the incredibly hot P4 processors, but the three phase VRM on my DFI run at about 50c with no airflow whatsoever so i don't worry about them. I would imagine more recent mobos have more efficient four or eight phase VRMs and therefore run even cooler.

spaceman_spiff · Post by **spaceman_spiff** » Wed Jan 31, 2007 12:39 pm

Swiftech makes a mosfet heatsink now, similar to the ramsinks made for video cards. I'm thinking about getting a set for my Asus P5B Deluxe.

http://www.swiftech.com/products/mc21.asp

DE

aristide1 · Post by **aristide1** » Wed Jan 31, 2007 1:34 pm

Well that's why God invented this thingamajig:

http://www.svc.com/wpr.html

Put a fan at any location.

http://www.bigbruin.com/reviews05/where ... php?file=2

Bluefront · Post by **Bluefront** » Wed Jan 31, 2007 4:35 pm

The ultimate solution to this problem....blow your rear case fan inward, toward the tower heatsink. Construct a three-sided duct, the side toward the MB left open. Most of the air will be blown through the heatsink, but a portion of the airflow will hit the MB components directly......plenty of cool airflow over the board, like this.....

Shaman · Post by **Shaman** » Wed Jan 31, 2007 7:34 pm

aristide1 wrote:Well that's why God invented this thingamajig

I don't see how that would cool the mostefs with a huge tower heatsink in the way? And if you're gonna use an extra fan just to cool the mosfets, might as well use a HSF that blows down at the cpu and cools both and avoid using another fan = less noise.

Shaman · Post by **Shaman** » Wed Jan 31, 2007 7:37 pm

Bluefront wrote:The ultimate solution to this problem....blow your rear case fan inward, toward the tower heatsink. Construct a three-sided duct, the side toward the MB left open. Most of the air will be blown through the heatsink, but a portion of the airflow will hit the MB components directly......plenty of cool airflow over the board, like this.....

That's one option, but if the back case fan is blowing inward, how does the hot air leave the case? Do you put a fan at the bottom front or at the top of the case blowing out? Or do you rely on the PSU to do that?

aristide1 · Post by **aristide1** » Wed Jan 31, 2007 7:43 pm

Shaman wrote:
aristide1 wrote:Well that's why God invented this thingamajig
I don't see how that would cool the mostefs with a huge tower heatsink in the way? And if you're gonna use an extra fan just to cool the mosfets, might as well use a HSF that blows down at the cpu and cools both and avoid using another fan = less noise.

There are tiny heatsinks you can buy and attach to individual mosfets and any chip that runs hot. All that metal and surface area alone cools them significiantly.

Also - just because a cpu heatsink fan is blowing downwards doesn't mean the air is hitting the right places, hence the positioning aspects of the thingamajug.

Also - food for thought BF. Very interesting.

cmthomson · Post by **cmthomson** » Thu Feb 01, 2007 1:36 pm

MOSFETs are generally designed to withstand very high temperatures (many are spec'd at 165C). It isn't the MOSFETs that die on an overly-hot motherboard, it is the bulk capacitors beside them. Of course cooling the MOSFETs has the side-effect of cooling the caps.

I ran an overclocked Pentium D 830 on a garden-variety motherboard (P5LD2) for about a year. The caps bulged more and more over time until one day one of them shorted out.

Newer top-of-the-line motherboards (such as P5W or DQ6) have five advantages over my old setup: lower-power CPUs such as Conroe, multiphase VRMs (8 or even 12 phases), high-quality solid-state capacitors, stock radiators on the MOSFETs (although many people remove these to get better north bridge cooling), and lots of copper cladding on the motherboard to suck the heat away (StackCool, etc).

Bottom line: running a Conroe with a tower heat sink, even overclocked, will not overheat a multiphase VRM.

Bluefront · Post by **Bluefront** » Thu Feb 01, 2007 2:27 pm

Shaman.....In my setup there is another fan blowing inward from the bottom. This is a totally positive pressure case. The heat goes out a vented top. The PSU is a fanless Zen.

Aris · Post by **Aris** » Thu Feb 01, 2007 8:37 pm

Shaman wrote:Mosfets and voltage regulators, don't they need as much cooling nowadays? Or does everyone running tower coolers simply accept the fact that their motherboards will have a sorter life span as a result of insufficient cooling?

Shorter? seems a bit ambiguous.

IMO, "shorter" is probably in the neighborhood of: will only last 10-15 years instead of 15-20 years.

I dont have any motherboards that have been running for more than 5 years. So i really could care less if i'm shortening the life span of the motherboard or not. As long as its quiet and stable, thats all i care about.

cmthomson · Post by **cmthomson** » Fri Feb 02, 2007 6:48 am

If you let the capacitors of a cheap-o motherboard get hot, they will likely fail within 1-2 years, not 10.

dfrost · Post by **dfrost** » Fri Feb 02, 2007 4:10 pm

Also - just because a cpu heatsink fan is blowing downwards doesn't mean the air is hitting the right places, hence the positioning aspects of the thingamajug.

The VRM MOSFET's on my Socket 478 Abit AI7 indicate temperatures higher then the (3.0C Northwoods) CPU under load. They were even worse for the short time that I had a Prescott in there.

They may have been just fine for life at those temperatures, but it didn't seem like a good idea when there are easily available remedies.

I dropped the Abit indicated ("PWM" in uGuru-speak) load temp by 10-15C with the combination of directed air from the 92mm fan (at 5V, visble on-edge between PSU and optical drive in this photo), thermal-epoxied heatsinks, and the airflow from the Papst fan on the XP-120. Regardling the last, that temp is 7C cooler with the fan in "suck" mode compared to "blow."

aristide1 · Post by **aristide1** » Fri Feb 02, 2007 4:22 pm

Impressive, well thought out.

sjschwinn · Post by **sjschwinn** » Fri Feb 02, 2007 5:35 pm

Shaman wrote:Mosfets and voltage regulators, don't they need as much cooling nowadays? Or does everyone running tower coolers simply accept the fact that their motherboards will have a sorter life span as a result of insufficient cooling?

I put various heatsinks on them. I have some copper swiftechs and some generics from JabTech. Seems to work well...

dfrost · Post by **dfrost** » Fri Feb 02, 2007 10:54 pm

I put various heatsinks on them. I have some copper swiftechs and some generics from JabTech. Seems to work well...

I tried small copper heatsinks attached with thermal tape (like the one inside the CPU HSF frame in this shot) and saw no difference. The substantial improvement from heatsinks came when I put on those larger black aluminum ones (made from an old CPU passive HS) with Arctic Alumina epoxy.

Aris · Post by **Aris** » Sun Feb 11, 2007 7:41 pm

cmthomson wrote:If you let the capacitors of a cheap-o motherboard get hot, they will likely fail within 1-2 years, not 10.

thats what you get for buying cheep then isnt it?

aristide1 · Post by **aristide1** » Sun Feb 11, 2007 7:59 pm

The problem is sometimes you spend good money and get cheap anyway.

Biostar manages to put solid caps around the cpu socket even on the GForce 550, a $75 board.

sjschwinn · Post by **sjschwinn** » Sun Feb 11, 2007 10:17 pm

dfrost wrote:
I put various heatsinks on them. I have some copper swiftechs and some generics from JabTech. Seems to work well...
I tried small copper heatsinks attached with thermal tape (like the one inside the CPU HSF frame in this shot) and saw no difference. The substantial improvement from heatsinks came when I put on those larger black aluminum ones (made from an old CPU passive HS) with Arctic Alumina epoxy.

Using an adhesive TIM - like Arctic Alumina or Arctic Siver Adhesive - is a very good idea. I wonder why I never thought of it!

Ever tried putting it on the caps as well as the transistor packages? My caps get very hot also...

cmthomson · Post by **cmthomson** » Mon Feb 12, 2007 3:54 pm

Aris wrote:
cmthomson wrote:If you let the capacitors of a cheap-o motherboard get hot, they will likely fail within 1-2 years, not 10.
thats what you get for buying cheep then isnt it?

The P5LD2 is not cheap, but it uses liquid electrolytics. If you want peace of mind, look for the Sanyo capacitors. They are very distinctive: they are squat aluminum cylinders with a D-shaped patch of purple paint on top.

See for example the DQ6 or P5W boards.

All these vertical tower heatsinks, what about the mosfets?

All these vertical tower heatsinks, what about the mosfets?

Re: All these vertical tower heatsinks, what about the mosfe

Re: All these vertical tower heatsinks, what about the mosfe