Peltier cooling and sililcone oil

[FishTankX]

I was thinking of trying an extreme over clocking project, but would like opinions on the success or failure here first.

My basic idea was to use water cooling to cool a bank of peltiers, powered by a dedicated power supply. This bank of peltiers would in turn cool a heat exchanger pumping silicone oil around a submersion case that would contain the motherboard and other components. in addition another pump would run the silicone oil through water blocks on the GPU and CPU.

My theory is that with this system I can get extremely low temperatures (This stuff is pump able down to -50C) and avoid all threats of condensation, in addition to making my space heater obsolete.

Comments?

 

[Idontcare]

If you want -50C why not go with a commercial vaporphase cooling rig like the VapoChill LS?

 

[FishTankX]

It costs significantly more than this would. And the cool factor is much less.

In addition to the fact that this system would be dust immune.

 

[PCTC2]

Well, the problems I see with this project is 1) the cost and 2) having enough radiators to cool the peltiers, because the heat output would be CPU + GPU + wattage ratings for the peltiers. The best radiators today are able to dissipate 400-500W for a 120x3 radiator. Plus having a pump to offset all the restriction and keep a good flow rate would require an Iwaki RD-30. Plus the power. A bank of peltiers capable of bringing a GPU and CPU to -50C is power hungry. I would recommend vaporphase instead.

 

[Idontcare]

Originally posted by: FishTankX
It costs significantly more than this would. And the cool factor is much less.

In addition to the fact that this system would be dust immune.

Do you have a cost workup on that? Considering the cost of *good* water-cooling setup (pumps, radiators, etc) gets up above $300-$400 I am interested to see how you envision throwing in multiple peltiers, another pump or two, plus the encasement of the secondary silicone system is going to net all that much less than a commercial vapor cooled system.

Cool factor?

Dust immune? You are still discharging the heat into the air by way of thermal exchanger at some point in your system. Sounds like you intend to do this with the water-cooling radiators. They will not be immune to dust.

I like the idea of water-cooling peltiers and then using peltier cooling directly on the IHS. But this added loop of silicone...sounds like an engineering nightmare to be honest. Keep us posted if you go thru with it. You'll definitely be in uncharted territory.

 

[PCTC2]

Originally posted by: Idontcare

Originally posted by: FishTankX
It costs significantly more than this would. And the cool factor is much less.

In addition to the fact that this system would be dust immune.

Do you have a cost workup on that? Considering the cost of *good* water-cooling setup (pumps, radiators, etc) gets up above $300-$400 I am interested to see how you envision throwing in multiple peltiers, another pump or two, plus the encasement of the secondary silicone system is going to net all that much less than a commercial vapor cooled system.

Cool factor?

Dust immune? You are still discharging the heat into the air by way of thermal exchanger at some point in your system. Sounds like you intend to do this with the water-cooling radiators. They will not be immune to dust.

I like the idea of water-cooling peltiers and then using peltier cooling directly on the IHS. But this added loop of silicone...sounds like an engineering nightmare to be honest. Keep us posted if you go thru with it. You'll definitely be in uncharted territory.

QFT. My radiators fill with dust all the time. The cost of a watercooling system would be about $400-800. The power costs would probably be similar to a vaporchill system. Go for vapor. Buy a custom one. I'm sure there's someone on XS that will build one for you.

 

[DrMrLordX]

I dunno . . . it could work. The truly cool part about it would be that he could cool the entire system with silicone oil rather than just one or more hotspots. The CPU, RAM, PWM/mosfets, vid card, etc. would all get some love provided that the oil circulated quickly and thoroughly.

That being said, why use a bank of thermoelectric coolers to cool the silicone oil? They are not efficient, and they certainly aren't the only way to reach sub-zero temperatures. Arranging them properly and/or setting up a control mechanism to prevent temperatures from dropping below -50C would also be a headache (otherwise your silicone oil might gel up on you). I have a feeling you might have to stack your TEC coolers just to get down to -50C anyway, and the power consumption involved with stacked TECs when you're trying to remove the heat dissipated by your entire system would be astronomical.

You'd be better off cooling the heat exchanger with a high heat capacity single-stage phase system. Were the system well-built, I'm sure you could control temperature limits more easily than you could with a mess of stacked TECs and reach -50C with no hiccups.

 

[BTRY B 529th FA BN]

Could of posted this in the Case & Cooling section and got some advice from Aigo.

 

[PCTC2]

Originally posted by: BTRY B 529th FA BN
Could of posted this in the Case & Cooling section and got some advice from Aigo.

He'd probably just state the astronomical costs and headaches of such a system and say to go phase. Could we get a mod to move this thread to cases and cooling?

 

[Idontcare]

Originally posted by: BTRY B 529th FA BN
Could of posted this in the Case & Cooling section and got some advice from Aigo.

It's highly unlikely Aigo has not seen this thread already, unless he simply hasn't visited the AT forums at all recently.

He's fairly liberal with his advice on any and all things discussed in the CPU forum in nearly the same realtime fashion as he does in the Cases and Cooling forum.

Like the eye of sauron that Aigo is.

 

[BTRY B 529th FA BN]

Originally posted by: Idontcare

Originally posted by: BTRY B 529th FA BN
Could of posted this in the Case & Cooling section and got some advice from Aigo.

It's highly unlikely Aigo has not seen this thread already, unless he simply hasn't visited the AT forums at all recently.

He's fairly liberal with his advice on any and all things discussed in the CPU forum in nearly the same realtime fashion as he does in the Cases and Cooling forum.

Like the eye of sauron that Aigo is.

LOL

yeah, i was kinda wondering where is voice was

 

[Concillian]

I think phase would actually end up costing less than a peltier system that could do anywhere near the same performance. There's a reason people don't use peltiers anymore. They work great when you need to cool something with a max output of 20-50W. When you start talking about performance systems of today which are dissipating upwards of 300W between the CPU and GPU(s), you start running into very serious limitations of peltiers.

 

[aigomorla]

Originally posted by: Idontcare
Like the eye of sauron that Aigo is.

LOL...

first off you cant use mineral oil if you want to take it to sub degree's.

Its only useful to around 10C, and anything lower it strarts to clump up pretty bad.

Second, u are aware the cost of electricity for such a unit will bit you in your butt.

Pelliters arent something magic that gobbles heat away. There terribly inefficient, and within a matter of 2 months, you'll probably recoup the cost of a phse unit in electricity bills.

If your after sub degree temps, get a phase unit.

 

[Sentry2]

Yeah dude, I've always been curious about running a few 220W pelts(or better) but if you think about it the cool factor isn't really there. Phase is the way to go. I'm gonna grab a phase unit this year I think...finally. I'll probably see if Jin is still pumping out custom units. With the money you'd have to spend on the different parts for that setup I'd just go vapor. If you want even half decent temps you'll need to use some pretty stout pelts...and a pretty expensive dedicated psu to run them properly.

Interesting idea though.


:beer:


If you do try it be sure to post some pics!

 

[Idontcare]

Originally posted by: Sentry2
Yeah dude, I've always been curious about running a few 220W pelts(or better) but if you think about it the cool factor isn't really there. Phase is the way to go. I'm gonna grab a phase unit this year I think...finally. I'll probably see if Jin is still pumping out custom units. With the money you'd have to spend on the different parts for that setup I'd just go vapor. If you want even half decent temps you'll need to use some pretty stout pelts...and a pretty expensive dedicated psu to run them properly.

Interesting idea though.


:beer:


If you do try it be sure to post some pics!

Having been a phase owner for >2yrs now I can say the cool factor of super GHz speeds and sub-zero cooling is pretty much over-rated.

The question is what do you do with the extra GHz you get after going thru all the expense and hassle of getting a phase rig setup?

For me it was an obvious answer as my computer(s) are put to work making me money, so everything is assessed in a ROI viewpoint. In 2006 the ROI for a phase rig justified the investment. Not so much nowadays in the land of i7's and easy 4Ghz clocks.

So what will a 5GHz i7 or yorkfield do for you that a 4.1GHz equivalent system won't? And is that delta in performance worth ~$1k upfront plus another $200 extra annual electrical expenses and the unavoidable higher frequency of mobo replacements?

Back in the day of 800MHz Athlon's and P3's using a phase rig to overclock another 20% and hit the 1GHz "barrier" was a cool-factor and also gave the user a noticeable improvement in your system response time when you used the computer.

A 20% overclock above and beyond a 4GHz clocked quad-core rig is not really going to provide a noticeable change in the user's experience with their rig. The cool factor is gone, replaced with the question of what do you do with an extra 20% CPU power?

 

[aigomorla]

im in a middle of a petlier project.

it uses 4 x 537W TEC's.

The hotside requires 2 fesser 480's.

im running 2 tec's in serial, and 2 pairs in paralell.

The math comes down to i should be able to pull 320W of heat with only using 240W of power.

However that means the hotside will be about 560W. You see, you need to make sure your math is down with TEC's b4 you try. You goof on TEC's and it spells uber accident.