Am I crazy? Silicone amorphous waterblock

nicozeg · 03-25-2004, 01:23 PM

I was looking for a solution to cool the power regulator circuits of my Abit NF7-S 2.0 that completely avoid the need of forced air cooling, and cools the mosfets, coils and capacitors.

As a waterblock that conforms to all those shapes would be impossible, I developed a bright (or dumb?) idea.

Why not impermeabilize all mobo components in the zone, and allow water to move freely over them.

As this is an experiment with big chances of being a stupid thing, it must be undoable, so the first stage was oiling the mobo to make easier the silicone removal.

I'm making this by layers, and up to now have the first two drying, Ill keep you all posted on the development of this.

Tempus · 03-25-2004, 01:50 PM

Wow. I love it. Hope it works@!

MMZ_TimeLord · 03-25-2004, 03:09 PM

nicozeg,

Yep, you be crazy... but I like it!

I will monitor this thread and see how it turns out for you.

dima y · 03-25-2004, 03:29 PM

just when i thought direct die was craZy. You come out and blow me away.

wow I really want to see what happens with this.

its really a shame that mobo makers dont put those on the other side of the mobo

Pears0 · 03-25-2004, 03:48 PM

Bye bye perfectly good mobo.

U will probly cause a power cut for ur area when u blow the power transformer on ur part of the power grid!

Hope u dont mind £1million+ bills!

Jason_The_Angry · 03-25-2004, 03:50 PM

just wondering about the heat transfer characteristics of silicone. would that not insulate the components?

Pears0 · 03-25-2004, 04:08 PM

If it dosnt insualte, it will conduct, if it conducts heat, it will in the same way as a metal dose, elctron motively, which means a currnt, which means a short on high current components, bye bye mobo! lol!

MadHacker · 03-25-2004, 04:35 PM

I think this is pretty cool...(pun intended)
if it is strong enough you could make a resevoir out of it...

|kbn| · 03-25-2004, 04:51 PM

Silicone is used for thermal pastes and doesnt cunduct... however I dont think sealant silicone will work well for this atall. Also cooling the mosfets will u sually cool all the other components in the area, as they dont output that much heat and but the mobo's traces do cunduct heat which takes heat away form surrounding areas.
The TO-220 package (which mosfeets use) is disigned so that the part on mosfets which is soldered to the PCB is suppost to be mounted to a heatsink. This means the mobo acting as the heatsink and disiapating the heat more than the small plastic surfact that people usually assume disipates the heat...

MadHacker · 03-25-2004, 05:13 PM

Quote:

Originally Posted by |kbn|

Silicone is used for thermal pastes and doesnt cunduct... however I dont think sealant silicone will work well for this atall. Also cooling the mosfets will u sually cool all the other components in the area, as they dont output that much heat and but the mobo's traces do cunduct heat which takes heat away form surrounding areas.
The TO-220 package (which mosfeets use) is disigned so that the part on mosfets which is soldered to the PCB is suppost to be mounted to a heatsink. This means the mobo acting as the heatsink and disiapating the heat more than the small plastic surfact that people usually assume disipates the heat...

that makes sense...

but one idea would be when he is all finnished he can scratch off the silicon off the plastic part of the mosfets. as long as he doesn't scratch off to much he should be ok...

nicozeg · 03-25-2004, 05:23 PM

I don't know silicone's thermal transfer properties, but sure are horrible.

But very few pople remember that mosfets already have a plastic package, with a bad heat conduction, so theres no point in attaching some high performance material on top of it. The ideal surface for cooling them is in contact with the mobo, wich is also made with plastic and some thin metal that improve thing a bit.

That means the area available for cooling is mostly plastic, so using silicone should not be so terrible. My idea is that the huge contact surface of this "block" will compensate for the low conductivity of the material.

Stay tuned...

Butcher · 03-25-2004, 05:24 PM

Quote:

Originally Posted by Pears0

Bye bye perfectly good mobo.

U will probly cause a power cut for ur area when u blow the power transformer on ur part of the power grid!

Hope u dont mind £1million+ bills!

Unlikely to blow more than the house power, in fact unlikely to do more that trash his motherboard.

Incoherent · 03-25-2004, 05:41 PM

Quote:

Originally Posted by Pears0

If it dosnt insualte, it will conduct, if it conducts heat, it will in the same way as a metal dose, elctron motively, which means a currnt, which means a short on high current components, bye bye mobo! lol!

Is this supposed to be science? I want to see the text book. Interesting interpretation of the problem.

Most PSU's protect themselves from a direct short. Water is not a great conductor, the voltages around the regulators are not high.
My money's on a blank screen, a wet motherboard and a PSU in need of a reset.
The problem I see is getting the water to the channel without getting it everywhere else.

But I like it.

Cheers

Incoherent

Butcher · 03-25-2004, 05:48 PM

Yeah I've shorted my psu a few times (luckily only across the hdd power connectors so no hardware blown) and all that happens it the box powers down and you have to power off for 30 seconds to reset it.

Joe · 03-25-2004, 05:58 PM

Quote:

Originally Posted by Pears0

Bye bye perfectly good mobo.

U will probly cause a power cut for ur area when u blow the power transformer on ur part of the power grid!

Hope u dont mind £1million+ bills!

Thats so retarded that its almost funny

trit187 · 03-25-2004, 09:01 PM

why not try on an old mb first?? might save you a good mb...

nicozeg · 03-25-2004, 09:46 PM

You may ask how I'm going to put barbs on this?

That's easy, it dont need it.

I'm integrating a silicone hose directly in the box, it should merge with the fresh silicone in one solid piece.

I'm almost done with it, tomorrow morning I'll bond the cap, and should be ready for a test run on the night

About the mobo: I bought it used just for experimentation.

DrCooling™ · 03-25-2004, 09:47 PM

Cool.

Como lo taparas?

Incoherent · 03-26-2004, 01:28 AM

Quote:

Originally Posted by nicozeg

You may ask how I'm going to put barbs on this?

That's easy, it dont need it.

I'm integrating a silicone hose directly in the box, it should merge with the fresh silicone in one solid piece.

I'm almost done with it, tomorrow morning I'll bond the cap, and should be ready for a test run on the night

About the mobo: I bought it used just for experimentation.

Looking forward with great interest to your results. I hope your feed solution is robust enough for it to work well.

Cheers

Incoherent

Cptn. Foo Foo · 03-26-2004, 03:48 AM

I dont want to hijack this thread but this is how I cooled the MOSFETs in my system...

http://www.overclockers.com/tips1104/ (waterblock on the front of the motherboard)

http://www.overclockers.com/tips1115/ (waterblock on the back of the motherboard)

The back-side block is much more effective than the top-side block I believe since it is cooling the motherboard directly (the motherboard is used as a big heatsink for the MOSFETs).

The capacitors and other components around that area are also significantly cooler with the back-side block in place. Cool enough to be able to leave your finger tip on them indefinately. Before the blocks touching those small "donuts" resulted in instant pain.

G33k · 03-26-2004, 04:38 AM

Don't want to put a downer on this (I love seeing people try stupid/crazy things like I do

), but I'd put money on those hoses leaking. If itdoes work, I'd expect it to stop working within a week or two - I've used silicon sealant to seal blocks before and even in a water-block which was BOLTED TOGETHER, the selant eventually gave and the block leaked.

Still, always interesting to see new ideas being tried - just glad it's not my mobo

Butcher · 03-26-2004, 07:13 AM

How much difference does cooling the mosfets actually make to stability / max overclock anyway?

BTW the donuts are inductors.

|kbn| · 03-26-2004, 07:20 AM

Ive used sillicone sealant before in blocks and never had problems.... I wouldnt trust sillicone for the WHOLE block like you are though, but as its one-peice you shouldnt have many problems if it stays together. How strong are the sides?
Another way you could have done this is using resin/body filler for the sides and top, and not having anything over the chips except for a non-conducting layer of paint or something...?
I still think just cooling the MOSFETs would have been enough, the capacitors and other components dont make any heat themselves but they are absorbing it from the mosfets, as Cptn. Foo Foo said.

j813 · 03-26-2004, 07:29 AM

yes a bit out of topic, what are the real benefits of cooling this part? Aside that I know this is another hot part of this NF2 boards.
I notice when I upgraded my board to a new Epox NF2 version, the reading on the 5volts and most of the railings was higher than my old overclocked EpoxNF2.
hoping I generated an interest on cooling this part of a board.

|kbn| · 03-26-2004, 07:51 AM

Ill try to explain but some of it might be wrong, Ive not studied electronics!
That part of the board is where the 3.3, 5 or 12v? rails get stepped down to what your Vcore is set to. Its done using the switching method - DC power can not be stepped down, so instead its switched on and off very quickly to simulate AC. The main components are a PWM controler to do the high frequency switching, MOSFETs, capacitors, donut shaped inductors...
The MOSFETs do the power conversion, and becuase of the type of chip they are, they only consume power (heat) when they change state. The frequency of switching is very high, 400khz if i remember, which means the MOSFETs are busy, and make lots of heat.
The capacitors are there to smooth out any ripple in the outut voltage and make almost no heat.
MOSFETs can only switch as fast as there rated for and when they get hot there ability to switch is reduced so the output is less stable.

My nf7-s has a cut up heatsink on the mosfets, which is passive and helps a lot. At 1.95v it used to varie between 1.75 and 1.9v!! since I added the heatsink I can go al the way up to 2.10v with 2.03-6 showing in mbm5. The cpu is more stable at higher overclocks.

03-25-2004, 01:50 PM	#2
Tempus Cooling Savant Join Date: Jul 2003 Location: Michigan, USA Posts: 456	Wow. I love it. Hope it works@! __________________ Thou art God.

03-25-2004, 03:09 PM	#3
MMZ_TimeLord Cooling Savant Join Date: Apr 2003 Location: San Mateo, CA, USA, Earth Posts: 433	nicozeg, Yep, you be crazy... but I like it! I will monitor this thread and see how it turns out for you. __________________ MMZ>TimeLord "Oooooooooh... that's gonna leave a mark!"

03-25-2004, 03:48 PM	#5
Pears0 Cooling Neophyte Join Date: Nov 2003 Location: Birmingham, england Posts: 61	Bye bye perfectly good mobo. U will probly cause a power cut for ur area when u blow the power transformer on ur part of the power grid! Hope u dont mind £1million+ bills! __________________ England rule @ rugby. Lookin to build custom water coolin rig in next few months with compleat new hardware!

03-25-2004, 03:50 PM	#6
Jason_The_Angry Cooling Neophyte Join Date: Feb 2004 Location: D.C. Posts: 55	just wondering about the heat transfer characteristics of silicone. would that not insulate the components? __________________ "You can get much further with a kind word and a gun than you can with a kind word alone." - Al Capone

03-25-2004, 04:08 PM	#7
Pears0 Cooling Neophyte Join Date: Nov 2003 Location: Birmingham, england Posts: 61	If it dosnt insualte, it will conduct, if it conducts heat, it will in the same way as a metal dose, elctron motively, which means a currnt, which means a short on high current components, bye bye mobo! lol! __________________ England rule @ rugby. Lookin to build custom water coolin rig in next few months with compleat new hardware!

03-25-2004, 03:29 PM	#4
dima y Cooling Savant Join Date: Nov 2002 Location: palo alto, CA Posts: 164	just when i thought direct die was craZy. You come out and blow me away. wow I really want to see what happens with this. its really a shame that mobo makers dont put those on the other side of the mobo

03-25-2004, 04:35 PM	#8
MadHacker Cooling Savant Join Date: May 2003 Location: Okotoks, A.B. Canada Posts: 726	I think this is pretty cool...(pun intended) if it is strong enough you could make a resevoir out of it... __________________ "Great spirits have always encountered violent opposition from mediocre minds" - (Einstein)

03-25-2004, 04:51 PM	#9
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	Silicone is used for thermal pastes and doesnt cunduct... however I dont think sealant silicone will work well for this atall. Also cooling the mosfets will u sually cool all the other components in the area, as they dont output that much heat and but the mobo's traces do cunduct heat which takes heat away form surrounding areas. The TO-220 package (which mosfeets use) is disigned so that the part on mosfets which is soldered to the PCB is suppost to be mounted to a heatsink. This means the mobo acting as the heatsink and disiapating the heat more than the small plastic surfact that people usually assume disipates the heat...

03-25-2004, 05:23 PM	#11
nicozeg Cooling Savant Join Date: Sep 2002 Location: Santiago, Chile Posts: 403	I don't know silicone's thermal transfer properties, but sure are horrible. But very few pople remember that mosfets already have a plastic package, with a bad heat conduction, so theres no point in attaching some high performance material on top of it. The ideal surface for cooling them is in contact with the mobo, wich is also made with plastic and some thin metal that improve thing a bit. That means the area available for cooling is mostly plastic, so using silicone should not be so terrible. My idea is that the huge contact surface of this "block" will compensate for the low conductivity of the material. Stay tuned...

03-25-2004, 05:48 PM	#14
Butcher Thermophile Join Date: May 2001 Location: UK Posts: 1,064	Yeah I've shorted my psu a few times (luckily only across the hdd power connectors so no hardware blown) and all that happens it the box powers down and you have to power off for 30 seconds to reset it. __________________ Once upon a time, in a land far far away...

03-25-2004, 09:01 PM	#16
trit187 Cooling Savant Join Date: May 2003 Location: shanghai, china Posts: 200	why not try on an old mb first?? might save you a good mb...

03-25-2004, 09:47 PM	#18
DrCooling™ Cooling Neophyte Join Date: Mar 2004 Location: chile Posts: 20	Cool. Como lo taparas?

03-26-2004, 03:48 AM	#20
Cptn. Foo Foo Cooling Neophyte Join Date: Nov 2002 Location: Calgary, Alberta, Canada Posts: 95	I dont want to hijack this thread but this is how I cooled the MOSFETs in my system... http://www.overclockers.com/tips1104/ (waterblock on the front of the motherboard) http://www.overclockers.com/tips1115/ (waterblock on the back of the motherboard) The back-side block is much more effective than the top-side block I believe since it is cooling the motherboard directly (the motherboard is used as a big heatsink for the MOSFETs). The capacitors and other components around that area are also significantly cooler with the back-side block in place. Cool enough to be able to leave your finger tip on them indefinately. Before the blocks touching those small "donuts" resulted in instant pain.

03-26-2004, 04:38 AM	#21
G33k Cooling Neophyte Join Date: Sep 2002 Location: Bristol [UK] Posts: 73	Don't want to put a downer on this (I love seeing people try stupid/crazy things like I do ), but I'd put money on those hoses leaking. If itdoes work, I'd expect it to stop working within a week or two - I've used silicon sealant to seal blocks before and even in a water-block which was BOLTED TOGETHER, the selant eventually gave and the block leaked. Still, always interesting to see new ideas being tried - just glad it's not my mobo __________________ "Success is 99% Failure"

03-26-2004, 07:13 AM	#22
Butcher Thermophile Join Date: May 2001 Location: UK Posts: 1,064	How much difference does cooling the mosfets actually make to stability / max overclock anyway? BTW the donuts are inductors. __________________ Once upon a time, in a land far far away...

03-26-2004, 07:20 AM	#23
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	Ive used sillicone sealant before in blocks and never had problems.... I wouldnt trust sillicone for the WHOLE block like you are though, but as its one-peice you shouldnt have many problems if it stays together. How strong are the sides? Another way you could have done this is using resin/body filler for the sides and top, and not having anything over the chips except for a non-conducting layer of paint or something...? I still think just cooling the MOSFETs would have been enough, the capacitors and other components dont make any heat themselves but they are absorbing it from the mosfets, as Cptn. Foo Foo said.

03-26-2004, 07:29 AM	#24
j813 Cooling Savant Join Date: Mar 2003 Location: southeast asia Posts: 164	yes a bit out of topic, what are the real benefits of cooling this part? Aside that I know this is another hot part of this NF2 boards. I notice when I upgraded my board to a new Epox NF2 version, the reading on the 5volts and most of the railings was higher than my old overclocked EpoxNF2. hoping I generated an interest on cooling this part of a board.

03-26-2004, 07:51 AM	#25
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	Ill try to explain but some of it might be wrong, Ive not studied electronics! That part of the board is where the 3.3, 5 or 12v? rails get stepped down to what your Vcore is set to. Its done using the switching method - DC power can not be stepped down, so instead its switched on and off very quickly to simulate AC. The main components are a PWM controler to do the high frequency switching, MOSFETs, capacitors, donut shaped inductors... The MOSFETs do the power conversion, and becuase of the type of chip they are, they only consume power (heat) when they change state. The frequency of switching is very high, 400khz if i remember, which means the MOSFETs are busy, and make lots of heat. The capacitors are there to smooth out any ripple in the outut voltage and make almost no heat. MOSFETs can only switch as fast as there rated for and when they get hot there ability to switch is reduced so the output is less stable. My nf7-s has a cut up heatsink on the mosfets, which is passive and helps a lot. At 1.95v it used to varie between 1.75 and 1.9v!! since I added the heatsink I can go al the way up to 2.10v with 2.03-6 showing in mbm5. The cpu is more stable at higher overclocks.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)