Mosfets watercooling - anyone did it? :)

[trodas]

Hey, guys, ever considered radical method of cooling the mosfets like THIS...? :






Well, up to today, I gived up breaking the 200Mhz barrier (for me it IS a barrier ) and focused on enhancing cooling of my mosfets:

http://ax2.old-cans.com/show.php?p=badthin...52&c=8&d=1&v=v2

...with added heatpipe I ended up with stable 200x13 - errr, almost stable, when folding get replaced by prime torture CPU test, then machine hand hard after about 4 hours

So I quess is time to RADICAL cooling of CPU mosfets This should help cooling my CPU mosfets, but for the FSB, I need to cool the dimm mosfets as well. This heatsink is simply NOT ENOUGHT:

http://ax2.old-cans.com/show.php?p=badthin...13&c=8&d=1&v=v2

...so I quess I have to make watercooling block (not need to be very effective, about 60C is IMHO optimal mosfets temp) for them and then hope that this finally help me

Fo the record, I using now Antec Neo 480W PSU ( http://silentpcreview.com/article177-page1.html ) for their amazing 12V ratings and it should be able to give me enought power to go over 200Mhz FSB

What do you guys think?

[Chocobo]

I think that is pretty intense and at the same time over doing it. I think active cooling like a HSF would do just as good as water.

[amduser]

I think that is pretty intense and at the same time over doing it. I think active cooling like a HSF would do just as good as water.

no, even better, cause acutally not the mosfets get to hot but the (*spuheln, don't know english name, sorry*) get real real hot . mayby you get the point what i mean by this pic:

[-=DouglasteR=-]

I think that will solve the problem more because the copper plate itself will dissipate the heat, the water will not help.

Will may build a custom dissipator for all you motherboard, with flow inside it, not near

[trodas]

Chocobo - yep, but active cooling is noisy. I want stick with my passive watercooling, whatever this may sound extreme


amduser - yes, the toroids get very hot too, but the mosfets are the main source of heat - take the het "out of them" and all temps will drop considerably


-=DouglasteR=- - well, the block is designed not to be too much effective, so the target mosfets temp is about 60C That's because the more effective the block will be, then the higher my CPU/GPU temps will get, so...
Anyway, the cooper got excelent thermal conductivity, so it will transfer the heat to the water, that's for sure. Not extremly effectively, but effectively enought, so my machine stop crashing when being on 100% load for more that 8 hours I believe it. I just looking if someone else did it already

[MaxxxRacer]

um amd users, the white part in that pic (the hotest part) is the mosfets. They are the little black square things next to the cpu.

[amduser]

um amd users, the white part in that pic (the hotest part) is the mosfets. They are the little black square things next to the cpu.

believe me i know exactly what mosfets are and sorry i most correct you, that are no mosfets (*this lillte black thinks*) but phases (*yes, thats was the word i didn't rember earler*).

[MaxxxRacer]

sigh... those are mosfets. My dad builds electronics for a living and I know alot about this garbage and those are mosfets. If those are not mosfets, then there are alot of very knowledgeable ppl that need to be told otherwise.

[Wstinkbait]

These are made by HEAT-Trap and the design seems more efficient. I think a homemade version my be easy enough to make.

NF7


Generic

[MaxxxRacer]

here is some further proof.. just google mosfets and you will see how wrong you are

http://www.overclockers.com/tips1142/

EDIT

and some more mosfets

http://www.rojakpot.com/default.aspx...var1=71&var2=0

[amduser]

then check this out

Titel: Sind die Mosfets am wärmsten ?
Bildbeschreibung: Nein, die Spulen sind es.

the meaning on english is:

title: are see mosfetst teh hotest ?
picturediscribe: no, the phases are.

so what know specialist? you want to tell me a dothan produces more heat then a prescott .

eidt:

that your dad ist elecrotoc don'T interrest me so long you don't know major thinks he probely can't explain. only cause i don'T have that many post it and my english is probley realy bad, doesn't mean i don't know what i'm talking about .

[Aphex_Tom_9]

oh, i think i know what's going on. in that thermal pic, obviously the Chokes (toroids, whatever) are hotter because they are white, and the MOSFETS are pinkish because theyre about 50c. i bet that guy isnt using a very high Vcore. mosfets dont start really heating up until you give them a big load. raise the vcore, put the system under stress, and the mosfets will probably get hotter than the chokes.

[MaxxxRacer]

in that last pic there it looks like the capacitors are getting hot. is there anyway to show that pic in non thermal so i can see what all the parts are?

[amduser]

oh, i think i know what's going on. in that thermal pic, obviously the Chokes (toroids, whatever) are hotter because they are white, and the MOSFETS are pinkish because theyre about 50c. i bet that guy isnt using a very high Vcore. mosfets dont start really heating up until you give them a big load. raise the vcore, put the system under stress, and the mosfets will probably get hotter than the chokes.

that guy uses a 2,2volt barton, when thats not much vcore what then i don't know .

edit: it's a normal gigabyte nforce2, looks that way:



edit2: pic of the setup, can't find on without fan, bu it's really n ordinary gigabyte nforce2

[MaxxxRacer]

those coils are the hotest thing then.. that is wierd... like really freekign wierd.

normally those dont get nearly as hot as the mosfets. that is pretty wierd. but those are deffintely not mosfets.. lol..

aphex, you coudl be right that the pic wasnt taken at load, becuaes under load the mosfets get hotter than that i think.

[amduser]

those coils are the hotest thing then.. that is wierd... like really freekign wierd.

normally those dont get nearly as hot as the mosfets. that is pretty wierd. but those are deffintely not mosfets.. lol..

aphex, you coudl be right that the pic wasnt taken at load, becuaes under load the mosfets get hotter than that i think.

don't, but with my epox 9nda3+ even a load it looks the them. mosfets are colder then phases. was also really surprised but you learn every day somethink you whenn you overclock, at least me.

[Aphex_Tom_9]

yeah, typical mosfet temps are in the 60-70c region under heavy vcore/load.
in any case i just have a blower over all of that stuff so it stays cool

edit- you know what i'm thinking now?
in that pic, there's a fan right over that area. MOSFETs are smaller, and thus can dissiapte heat at a much faster rate, and have the added cooling spot on the motherboard. the chokes are larger and denser and can hold heat and dissipate it less efficiently. i bet if you took that fan out of the picture, the mosfets would get equally as warm as the chokes are there.

[MaxxxRacer]

well i would check the coils, but i dotn feel like getting electrocuted. i accidently did that a few days ago reaching into my case and touching one.

but aphex is right in that the mosfets do get up to 60-70C. they are always the thing to get the active cooling, even from the mobo manufactures. the coils dont need cooling as they will not get dammaged as long as the copper doesnt melt. they are near indestructable.

[amduser]

yeah, typical mosfet temps are in the 60-70c region under heavy vcore/load.
in any case i just have a blower over all of that stuff so it stays cool

edit- you know what i'm thinking now?
in that pic, there's a fan right over that area. MOSFETs are smaller, and thus can dissiapte heat at a much faster rate, and have the added cooling spot on the motherboard. the chokes are larger and denser and can hold heat and dissipate it less efficiently. i bet if you took that fan out of the picture, the mosfets would get equally as warm as the chokes are there.

no, that guy noramly uses a icewater watercooling and cool them under -10°C or so. he took that stuff just of for a presnetation to show us what he has discoverd with hie thermal cam. seems to be different on new motherbords.

[Revv23]

you cant exacly cool the coils, and like stated, it doesnt really matter how hot they get, mosfet cooling is whats important, as well as clockgens and ic's... etc

[trodas]

Wstinkbait - yes, these mosfets WC heatsinks looking far more effective that mine - however they are obviously restrictive = very bad for the rest of WC loop + I did not want to cool the mosfets very well anyway - they could work perfectly and optimaly at about 50-65C w/o a glitch Cool then down better will not gain me a thing and only ruin my CPU/GPU temps
Think about what I mean


Revv23 - well, all they can do is heat the surrounding components, like capacitors and mosfets. And increase the temperature in case. Cooling them will be wise, but sadly impossible...
Yep, they "theoreticaly" did not matter, but when I applied lot's of hi-temp silicon on them (in idea to stop the whine from them), my machine become unstable, as the cols overheat the rest of the components much faster
So, even they - teoreticaly when isolated - could get very hot and still did not have much effect on OC and stability - in reality we are bound to care, because other components are so dangerously close to them

[amduser]

nope, this cooler are acutally really simple to procduce, way more effetic and can get any fiiting you have. the innerpipe ost 10mm big, shouldn't kill thatmuch the flow.

[trodas]

well, it looked like much smaller (the ID of the thing) to me, I could be wrong, of course Mine is 10mm ID as well.

...but they, it's the effectivity I did not fully want - I want just a little cooling, so my CPU/GPU temps did not get hurt much

If I want to design effective mosfet cooler, I will go with the Swiftech MCW6000/6002 design - and this will be HELL effective cooling But mosfets not need that good cooling (the CPU did not need such good cooling too, it's just a "competition" who get's lover temps, no practical gain is there. No CPU I get in my hand in past 18 years was heat-limited in OC. All maintained their maximal OC at even crazy temps - latest try, 86C on my AXP-M 2500+ witch gas got a wall at 2550Mhz for 1.850Vcore
Still it was 100% stable at 86C (AMD says mobiles are good up to 100C, so, who cares about 86C )

[Revv23]

well it would have probably gotten further with a colder die...

i know from experience that better cooling= better oc on my mobile.

[amduser]

well, it looked like much smaller (the ID of the thing) to me, I could be wrong, of course Mine is 10mm ID as well.

nope it is exactly 11,8mm (*g 1/4*) so see .