Another theoretical idea
 

k, I'm fairly new here so hi everyone.
It'd be nice if you could cool a whole bunch of components in your computer instead of just CPU/VPU. I'm working on a simple rig right now that is going to water cool CPU/VPU/N&S bridge/RAM/HD/whatever the hell else I find getting hot in there. Me and a buddie nicked it project "stealth" because the whole point is to have a nice semi-fast computer that I can run a few feet from my head while I'm sleeping.
But what about below ambient? I've never built a pelt cooling system yet, and I came up with an idea. Now bear with me, like I said I'm fairly new to this whole thing...
What if I liquidcooled the rig the same way I've got it planned right now (basically just an average watercooled setup with a pressure head and a lot of little tubing/valving to get to all the places I need), except with a pelt cooled resivoir? This of course means it would need a second water system to cool the pelts, but there would be no need for any radiator for the cooled jacket.
I noticed two major things here from a simple engineering standpoint: 1) My pressure head resivoir is going to WAY more surface to slap pelts on than any CPU, thus allowing me to basically get my cold system as cold as I want. 2) Air cooling a radiator gets more efficient with a bigger difference in fluid/ambient temps. I can keep a 20 Hp Johnson below 105 C with a mustang heater core and a tiny pump, so I'm assuming I should be able to keep a few pelts down under burnout with a radiator or two and a 160 Gph pump.
I haven't seen a setup like this yet, and I haven't started working on any designs yet because I wanted to get some input from people who've dealt with pelt cooling before.
Like johnny 5 said - I need input!

(a little clean up was needed, sorry)

1) I think you're oversimplifying the issue. There's flow, but also the powers involved: do you know how much power those pelts are going to use? Do you understand that they add heat, on top of the heat that they transfer?

2) Again, oversimplified. It's not just the delta T, it's also the whole temperature gradient. Try to estimate all the temperatures, from the hottest to the coolest: it'll make more sense.


Your mama was a snowblower. ;)

sorry 'bout my writing, yeah it's a mess.

Yeah, I was looking around at operating temps of pelts and whatnot. I do understand that the hot side of such a system would need one hell of a lot of cooling, but I do think it's possible with a big enough pump. Yes, putting 1000 watts inside a case is another thing, but isn't that just a money issue?

As for the cold side I'll take your word, I don't understand much about pelts yet.
I'm still insanely curious about this though, even outside of computing. What's the coldest you could make liquid with a huge pelt setup?

I have another question though, sorry if it's posted somewhere else in the forums and I couldn't find it: What's the best way to avoid moisture in systems operating below the dew point?

Pelts are very ineffecient, and they get worse the colder you run them. I don't know what the practical limit is. There are charts on google for common TEC elements. Look up one around your size and look where the dissipation you're interested in (plus a lot for leaked heat and divided by the number of elements you'll be useing) intersects temperature.

Insulation. But its not a big concern since you'll have to insulate the hell out of your loop before you'll even get to the point where its condenseing anything :)

Dan's got a great article on pelts; it really helped me pass the stage to where I'd consider using them. His site is: www.dansdata.com . The article may be hard to find, but is an excellent read.

Wicked site, thanx. I AM getting what you're saying now, and I managed to get a professor at the local university to think about the idea too. He explained to me that there'd be very little I could do under a certain temp, and even then the system would have to be pretty much static (no changes in temps of anything) to work at all. Oh well.

It would be far more effective to use immersion cooling with a de-humidifier coil in the case to get sub ambient cooling with low noise.

as for keeping the system static... all you would need is something that would monitor the temperature and turn on or off the peitlers as needed...
my 2¢