Peltier Questions

[OptyTrooper]

Don't really need a second PSU, just a big one.
400w of pelts is nothing to a modern beefy PSU.
Now if you're using 24v pelts that's a different story of course!

Condensation issues make any subambient cooling a dubious idea in my book, but if you know what you're up to I certainly don't care what you do with your time/money

Plus, I'm curious to see how this works out.

No need for curiosity Bob it will be fail.

@ OP: How do plan on combating condensation? Perhaps you can magically lower the dew point...

 

[Winflakes]

Look if you had done the research you would know that what you a proposing is impractical to say the least. It has been explored over and over. Don't get pissy b/c I am being straight with you.

Your right though if you want to dump large amounts of cash on a set-up that will ultimately fail that's your business. After all I am only a top notch bencher so what do I know?

I'm getting "pissy" because you are being obtuse.

Yes I know that phase change would be a hell of a lot better, performance-wise, than Pelts or water. But I also know that those are loud... Maybe I don't want a phase change machine running. Also maybe I dont want an ugly box next to my case feeding in cooling for my CPU. i have done my research, and I understand the arguments against pelts. They are all valid, from certain perspectives. If my goal was max OC for benching, I would not even consider pelts. I am entertaining the idea of the possibility of a system that is closed, can run 24/7, and has the ability to go sub ambient or su-zero with the turn of a dial.

You can have your opinion of pelts, its all fine and well, and I won't be hurt at all if you think they are a waste of time and money. But why do you need to come in here and rant on a thread whose objective is learning. You added absolutely 0 value to this thread.


Thank you for the vote of support Bob. I am still trying to fill my head with as much info as I can on the matter before I spend cash and take a hands on approach. And if I do end up doing anything, I'll be sure to make a thread and post what I do and find.

 

[Winflakes]

No need for curiosity Bob it will be fail.

@ OP: How do plan on combating condensation? Perhaps you can magically lower the dew point...

/sigh. Based on your logic, all phase change machines are terrible ideas. Please tell me, with a phase change machine... do they magically lower the dew point? Because by your logic if something cools a component past the dew point, it will not work and therefore is not worth the time entertaining the idea.

 

[OptyTrooper]

No value? Really? It will not work period. What you call a rant is what I call years of knowledge and experience. If the Opteron 1212/1210/X2 5000 Brizzy I killed doing just what you propose could talk it would tell you the same.

Please don't think I am trying to be rude as I tend to not mince words. I applaud your curiosity sir as knowledge is power but again it simply will not work as the laws of physics can not be circumvented.

Phase change does not create condensation only ice which can be combated under a benching session by the use of various insulation techniques.

 

[Winflakes]

No value? Really? It will not work period. What you call a rant is what I call years of knowledge and experience. If the Opteron 1212/1210/X2 5000 Brizzy I killed doing just what you propose could talk it would tell you the same.

Please don't think I am trying to be rude as I tend to not mince words. I applaud your curiosity sir as knowledge is power but again it simply will not work as the laws of physics can not be circumvented.

Up until now, you had added no value. I'm sorry but when person A is thinking of an idea and person B comes in and says "nope won't work" that is valueless. Now when person B backs that statement up with something, it is a different story.

However, I have read too many forum posts by people for Tec's and operating on successful Tec systems to yield that it will fail, black and white.

You talk about laws being uncircumventable, please elaborate because what I have found thus far says that it is possible. It has to be done right but it is possible.

 

[ssjwizard]

Does it work, well yes actually pelts are used in alot of laboratory equipment. The fact though is that pelts are horrifically inefficient and a setup like this is going to draw a lot of power, cost alot, and add multiple new points of failure to your setup. If those are worth it to you then theres no reason not to do it.

I considered a similar setup as I have a seriously oversized radiator to use but elected that the electrical draw and added complexity wasnt worth it to me. Only you know when to draw the line.

 

[OptyTrooper]

Up until now, you had added no value. I'm sorry but when person A is thinking of an idea and person B comes in and says "nope won't work" that is valueless. Now when person B backs that statement up with something, it is a different story.

However, I have read too many forum posts by people for Tec's and operating on successful Tec systems to yield that it will fail, black and white.

You talk about laws being uncircumventable, please elaborate because what I have found thus far says that it is possible. It has to be done right but it is possible.

Okay just one problem then I'm out.

Let's say that your pelt set-up emits X amount of heat and your dew point is Y. Now unless your in a extremely tuned climate control situation Y is in flux X will condensate thereby shorting components.

There are TEC success stories but I'd be willing to bet they were on old systems such as Athlon XP's or PIII's, I.E. low wattage.

 

[ssjwizard]

There are insulation techniques that could be employed even in a condensation situation with the pelts. Pack the socket with grease, pack around the WB with grease and contain with tape(I would use the 10mil underground insulation tape used on electrical conduit). Foam insulate the hoses. Of course this is all a nightmare of mess to fight with for small gains. However, if the OP wants to do it after we all told him the drawbacks, I want to see the finished product. We are a collective of tinkerers and engineers around here after all.

 

[gregmacknass]

Ok, I'll give you a little insight into TEC's that I have learned from 1 of my personal setups. The setup that I have uses 7 TEC's overall. Some are run at 5v (higher wattage ones) some are run at 12v (lower wattage ones). A single TEC will not make very much difference in cooling performance, and can just create more of a headache than watercooling. You will need at least 4-5 TEC's to make a substantial difference and get sub 0 temps. The thing that you have to take into account is the delta of the TEC's. So if it has a delta of 65, then if you keep the hot side of the TEC at 65 the cold side will be at 0. The delta ratings are given for TEC's for their maximum specified operational voltage, and if they are run with less voltage the delta will change. One of the big challenges (as mentioned by others) is finding a power supply that is capable of supplying the TEC's with enough power and amperage, while not running at it's limits. I personally use a 1250W power supply, because the TEC's alone are pulling around 50A. I was running this setup with 2 less TEC's on a 650W (good quality) and it killed it within 3 months because it was running on the ragged edge.

I'll go ahead and get into how I went about my setup, and maybe it will give you some ideas as well. My adventure started out with a coolit eliminator. It was good, but I wanted colder. So I took it apart to see what I could improve. I first got a 92-120mm fan adapter and a delta 160CFM fan to cool the heatsink better. I wired the TEC's in parallel because they were originally in series (to operate them at 12v). I then took 2 Maze 5 GPU blocks (I bought from a fellow user here), and used a 775 backplate, a 775 intel heatsink, a copper cold plate, a 200w tec, a vantec tornado 80mm fan, and some 3.5" screws to make my own coolers. I decided to run these at 6v (in series), because at 12v the copper cold plate was hitting 117F. I used a 200w TEC under the waterblock for the cpu, so in effect the 5 TEC's are cooling the water that cools the hot side of the TEC on the processor. This keeps the CPU TEC so cold that it will literally freeze water on contact (which is at least -30). The bios screen reports CPU temp as -32C. I run roughly a 60/40 mixture of coolant/distilled and never had any troubles with freezing. I'll go ahead and throw you some pics I have of the setup here. If you have any questions, I will do my best to answer them.

I figured that I would add that the cpu socket is packed with dielectric grease, and the motherboard had all the heatsinks removed and all the important bits taped off and was coated with several layers of automotive clear coat. The video card was as well also. The PCIE slot was also packed with dielectric grease as well. This setup will produce quite a bit of moisture on the board and block, so some form of insulation is absolutely necessary period.

 

[Winflakes]

Okay just one problem then I'm out.

Let's say that your pelt set-up emits X amount of heat and your dew point is Y. Now unless your in a extremely tuned climate control situation Y is in flux X will condensate thereby shorting components.

There are TEC success stories but I'd be willing to bet they were on old systems such as Athlon XP's or PIII's, I.E. low wattage.

As Wizard said, techniques used for LN2 or phase can be used here. I would just have to make sure to apply that to the hoses and the plate attached to the cold side of the TECs as well.

But the project log here, http://www.overclock.net/t/1094151/update-on-4-tec-chiller-and-controller, is exactly what I wanted to do when I first read about the problems with TECs. This would allow me to Set my target temp at dew point if I chose, or to set target temp at X celcius.

Either way, I would take plenty of time prepping everything for the condensation that would come.

 

[Winflakes]

Ok, I'll give you a little insight into TEC's that I have learned from 1 of my personal setups. The setup that I have uses 7 TEC's overall. Some are run at 5v (higher wattage ones) some are run at 12v (lower wattage ones). A single TEC will not make very much difference in cooling performance, and can just create more of a headache than watercooling. You will need at least 4-5 TEC's to make a substantial difference and get sub 0 temps. The thing that you have to take into account is the delta of the TEC's. So if it has a delta of 65, then if you keep the hot side of the TEC at 65 the cold side will be at 0. The delta ratings are given for TEC's for their maximum specified operational voltage, and if they are run with less voltage the delta will change. One of the big challenges (as mentioned by others) is finding a power supply that is capable of supplying the TEC's with enough power and amperage, while not running at it's limits. I personally use a 1250W power supply, because the TEC's alone are pulling around 50A. I was running this setup with 2 less TEC's on a 650W (good quality) and it killed it within 3 months because it was running on the ragged edge.

I'll go ahead and get into how I went about my setup, and maybe it will give you some ideas as well. My adventure started out with a coolit eliminator. It was good, but I wanted colder. So I took it apart to see what I could improve. I first got a 92-120mm fan adapter and a delta 160CFM fan to cool the heatsink better. I wired the TEC's in parallel because they were originally in series (to operate them at 12v). I then took 2 Maze 5 GPU blocks (I bought from a fellow user here), and used a 775 backplate, a 775 intel heatsink, a copper cold plate, a 200w tec, a vantec tornado 80mm fan, and some 3.5" screws to make my own coolers. I decided to run these at 6v (in series), because at 12v the copper cold plate was hitting 117F. I used a 200w TEC under the waterblock for the cpu, so in effect the 5 TEC's are cooling the water that cools the hot side of the TEC on the processor. This keeps the CPU TEC so cold that it will literally freeze water on contact (which is at least -30). The bios screen reports CPU temp as -32C. I run roughly a 60/40 mixture of coolant/distilled and never had any troubles with freezing. I'll go ahead and throw you some pics I have of the setup here. If you have any questions, I will do my best to answer them.

I figured that I would add that the cpu socket is packed with dielectric grease, and the motherboard had all the heatsinks removed and all the important bits taped off and was coated with several layers of automotive clear coat. The video card was as well also. The PCIE slot was also packed with dielectric grease as well. This setup will produce quite a bit of moisture on the board and block, so some form of insulation is absolutely necessary period.

So, your 5 TECs, are they all attached to a single block that just pumps water to the CPU TEC? What are the specs of these TECs?

How is this more beneficial than just running all 6 TECs to cool water pumped to the CPU? If you could maybe use an equation, I might be able to wrap my head around it.

Do you have any pics of the full PC? I'd like to see how it all looks when it is hooked up.

Thanks for the info man, Sounds like you are doing some cool things with cool TEChnology

 

[gregmacknass]

All of the 5 TEC's are cooled on the hot side by heatsinks, and each transfer cold to an individual waterblock for each TEC. The cooled water then cools the waterblock that is on the hot side of the TEC that is on the processor. The reason that I have it set up this way instead of all the TEC's cooling the water that cools the processor is because this way let's me get the processor colder. The water that is cooled by the TEC's is around 30F, so 30F on the hot side of the CPU TEC with a delta of 65 gives me -35. So obviously -35 is better than +30 right. That is why I have it set up this way. I don't know what the actual delta's of the TEC's I have are off hand, so I was just throwing some numbers up there. I know that 3 of them are 195W TEC's, and 4 are 70W or so. I hope this explains it well enough for you. I don't have any pictures of the rig all put together, but I will try to get some sometime here soon, as I plan on doing a writeup of the whole build.

 

[djscrew]

Okay just one problem then I'm out.

Let's say that your pelt set-up emits X amount of heat and your dew point is Y. Now unless your in a extremely tuned climate control situation Y is in flux X will condensate thereby shorting components.

There are TEC success stories but I'd be willing to bet they were on old systems such as Athlon XP's or PIII's, I.E. low wattage.

Seems to me that as long as you don't get careless with the pelts and lower temps more then necessary you should be able to avoid frost altogether.

CPU & Waterblock #1 -> Waterblock #2 with Pelts applied -> final cooling solution, pelt power is throttled as necessary based on load to keep the temperature of the CPU waterblock above dew point.

If your 2nd water block frosts over its fine because its not sitting on your components.

I don't pretend to know crap about the specifics here but what you are saying that somehow using a peltier cooler for any PC application will end the universe seems rather silly to me. There is absolutly a way to do this and make it work, can you just slap some pelts on water block and expect it to give you sub-ambient temps without the risk of blowing up the universe? No. Is it more efficient then LN/2? No. Could it be made to work 24/7 though with far superior cooling to a h100? Certainly.

I am also very interested to see where this project takes you

 

[Bobnova]

You need active control.
A voltage/current to the pelts that gives a cooling effect at full load will take things cold enough for condensation at idle.
A voltage/current that works well at idle will fry under load.

 

[Winflakes]

You need active control.
A voltage/current to the pelts that gives a cooling effect at full load will take things cold enough for condensation at idle.
A voltage/current that works well at idle will fry under load.

Yes! And that is exactly why my first project, in this long haul of projects, will be to make my newly acquired Arduino board be a powerful case controller. The plan atm is to have hook up a Temp Sensor and Humidity sensor to monitor ambient temps inside the case. Then run a small calculation every 5-10 seconds to calculate current dew point levels. Then use that in tandem with some Temp Sensors throughout the pelt loop to tone up or down the cooling throttle. I will also have some fail-safes put in place, like buzzers or blinking LED's to tell me if/when a sensor went out.

The Idea is to be able to tell my board, "Hey, never go below 1.5-2C above ambient." Then the next project will be to see if I can adapt that for a "manual setting" button, so if I feel daring I can crank the thing to full tilt and watch the frost form!

Anywho, if you have any advice on the matter I would love to hear it. This project is still a ways off as before I get too serious into this, I need to finish the house buying and moving process.

 

[djscrew]

One thing to consider: the more liquid in the system and higher flow of the pumps, the less pronounced spikes the temperature will be. I wonder if its worth it to have an insulated reservoir to spread temperature and reduce spikes.

 

[djscrew]

I will also have some fail-safes put in place, like buzzers or blinking LED's to tell me if/when a sensor went out.

It's called TJMAX