Question Peltier CPU Cooling

[Carl2]

I've been playing with Peltiers for quite some time, experimenting and learning from mistakes, I started out with the Tec 12706 which took 6 amps and have just begun using the Tec1 12709 that takes 9 amps which is much better. I have a cooper radiator used to cool the hot side of the Peltier, 2 pumps to move water for water blocks, one for the warm water generated and one for cool water going to the CPU. So the 12709 allows me to go from 2 tanks with Peltiers to one smaller tank and my next test will be to just cool the water block that will be on a CPU. In the back of my mind I keep thinking about just mounting the Peltier on the CPU itself with a water block on the hot side. I've worked in electronics and already have a transistorized regulator for temps where I can control the temp the Peltier gets to. My biggest concern is damaging the CPU. Any thoughts on this would be appreciated.

 

[LORDPrometheus]

it will all depend on how cold the surface of the CPU die gets. id you are going below ambient then you need to read up on sub zero cooling and ensure you take the proper precautions against condensation. Otherwise just make sure you use a mount that fits properly and applies the correct amount of pressure. The biggest concern with a Peltier may be startup time. In my experience they can take awhile to reach operating temp so either turn it on beforehand or make sure your loop has enough thermal capacity to keep things under control as it cools down

 

[tennis2]

Thoughts - Fun experiment. Impractical.

 

[Carl2]

I ran across an article ages ago where the Peltier is insulated between 2 copper plates, foam insulation is placed around the Peltier, this eliminates any condensation and has been working for ages. I also have begun working on a silicon solution that requires less work but is still work in progress.
I did test out the 2 water blocks, I used the aluminum water blocks because they hold 2 table spoons of water where the cooper only holds one table spoon. So I no longer need the cooper tank and it reduces the size. Water holds a lot of heat which is why it is used for cooling, I originally had a tank holding 1 cup of water that took quite a while to cool down, making a tank holding 1/4 cup of water with higher current made the cool down much more reasonable.
The 2 water blocks holding 2 table spoons of water Is pretty quick. A direct link to the CPU means no water cooling and response should be very quick.
I guess my next step is to try to control the temp of a cooper sheet using my transistorized controller.

 

[Spaceghaze]

I remember from back in the days, these where quite popular. I think i had a Athlon 1200 mhz at the time. Never got around to try one of those though.

What i do remember that people tended to stick one between the CPU cooler(just a normal one with heat sink and a fan). Dont remember though what the results of that was and how effectiv it was, also how it was with condensation.

What kind of CPU are you cooling with this?

 

[tennis2]

Linus Tech Tips have done some fun experiments with Peltiers. Including a 545W one a month or two ago. IIRC it was bested by a bog standard 240mm AIO in cooling performance. And they didn't have to deal with that extra 545W of waste heat from the Peltier inside a case since they did everything on the test bench.

 

[Carl2]

While looking for something to cool I came across a cooper piece I used in a plastic tank, cooper rods and fins. I decided to try this for cooling the hot side water block and it failed real bad. Never realized how much heat the radiator is removing till I tried the cooper piece. So for now the cooper piece will be used in place of the CPU for testing purposes. I have found that you tube has videos on making your own heat pipes.
For now I think I'll try getting the transistorized Peltier controller working
Thanks for the input.

 

[Carl2]

I'm using a 6 core i 7 Coffee Lake 95 W, built the computer a few years ago. I began thinking I had tested the Peltier with no load and actually should have applied a load so it has to do some work. I just checked the data sheet for the 12709 Peltier and it is rated for 77 W but the CPU shouldn't be maxed out at 95 watts so I don't think that will be a problem, I'll just have it cool down with some copper at a warm temp and see how it does. I seem to have gotten a cold yesterday, have to take it easy for a few days. I'm a fan of You tube, Linus is okay, watched the video and it seems the Peltier failed, I'll just say I used LED displays in the 70's and it took 40 to 50 years to get them to home and automotive lighting. I think they have potential and just keep trying.

 

[Carl2]

I watched some videos on thermal physics, it is some what similar to some electrical laws so not to difficult to understand and should be helpful. I have a few things to get done and then I can move forward.
Carl2

 

[Carl2]

After looking at the data sheet for the 12709 again I decided the hot side should have better cooling, never like the commonly sold water blocks the copper ones only hold less than a table spoon of water and the aluminum one hold about 2 table spoons of water. I'm also going to include a cool water system going through copper tubing to reduce the hot side temp and see if that will help. The hot side has the highest temps which degrades performance and providing lower temps there should be helpful.

 

[Carl2]

I just about have the new water block for the hot side finished, I just have to put on the bottom plate. Probably the most complex thing I've built, it contains a top level that inputs and exits some of the cooled water, the bottom level is for water to remove heat from the hot side and has inlets and outlets going to the radiator. Experimental, maybe a waste of time but I don't think it has been tried before.

 

[tennis2]

Pictures, or it never happened

 

[rubix_1011]

A lot of people have used Peltiers to cool CPUs using a water block to cool the hot side.

This was one of the more advanced cooling solutions of the early 2000's before you went to liquid nitrogen.

There are plenty of reasons why you don't see them much today, mostly because they require their own, independent power source which draws a lot of energy and the fact that there are other solutions that don't require this much complexity and allow the CPU to run at nearly the same config.

 

[Carl2]

Well my latest experiment was a failure, the cool water coming from the cool water block was heated in the chamber to cool the hot side and this caused a reduction in the low temp of the water coming from the cool side water. A lot of time, effort, copper soldering gone to waste but it is something I had wanted to try. Not much on You tube about Peltiers lately it seems most people consider it a waste of time but I do like the 12709 Peltier that I began using.
Back In the 1960's we were using liquid nitrogen to cool inferred detectors to lower the noise levels of the detectors and we tried the Peltiers but stayed with liquid nitrogen.

 

[Carl2]

I looked through my notes and found a tank that had a quick cool down time with a low temp and decided to use that tank as a cool down tank. I modified the hot side tank by connecting the chamber and the lower chamber to the radiator for cooling which removes the warmer water from combining with the cooled water. So far this seemed to work pretty well, I'll recalibrate the temp probe and test again when everything is at room temp. I have an inexpensive infrared thermometer but find some readings are off and the small probe I use to be more accurate. At this time the cool water tank is pretty large and I'll probably make a smaller version so I can clamp the hot side water block, Peltier and cool side water block together. Lots of copper fins needed to cool down the water in the cool side tank.

 

[Carl2]

I had to clear the kitchen table so it could be used for Thanksgiving so the Peltier project has been removed. A few days of snow kept me outside clearing walks and the driveway. Guess I'll be back on the Peltiers soon. Not much going on with Peltiers, I feel like I'm the only person trying to get one going.

 

[rubix_1011]

If you have photos to share, I would be interested to see what you have built. I'm always curious as to how people are trying to utilize different methods of cooling and see what they discover.

 

[Carl2]

I woke up this morning and my mind was thinking about Peltiers, for a long time I had been thinking of just connecting the Peltier directly to the CPU rather than cooling water pumping it to a water block for cooing the CPU. Do away with that loop and the associated losses and reduce the number of components used. A big issue I've found is condensation but ages ago I found the insulating the Peltier form outside moist air removes that problem and never had a problem with condensation. Since liquid nitrogen has been used to cool CPU's to low a temperature would not seem to be a problem but it seems sense less to go to the lowest temperature possible I believe it would be a waste of power and increase temps inside the computer for now a rough guess of 40 to 60 F would be acceptable. I think the people of today have been used to thinking digital where analog control of the Peltier is recommended. I am fortunate that a lot of my work had been done in the analog world and operational amps and transistors were used. I've found the temp sensors being sold are really to large with to great a mass to be useless since there reaction time would lag to much. A small pin sized thermistor would be needed to control the Peltier. I wish I had a computer to test out my thoughts and I guess this is why I started out going the more conventicle way. I do have a 85 W and 30 W soldering irons that may be useful for testing.