This project is all about to be cool - in a practical way. This design describes a compact heat exchanger based on a thermoelectric element, TEC - also called peltier. This kind of device acts like a heat pump, drawing heat from one side to the other. It is done when a DC current runs through it. This gives a difference between the cold and hot side at about 20°C (depending on type of TEC). By cooling the hot side, it is possible to get very low temperatures at the cold side.

This first prototype was built to test the TEC out - how cold could I go with rather simple equipment. These tests will be the base for a complete cooling system that could be used in a different ways. My primary objective is to cool down the air inside a computer case, lowering the overall temperature of the computer's CPU and graphic card. But it could be used for other purposes where cold air is needed.


* First - a TEC is needed. I bought mine when I was in California some years ago. I planned to use it for direct CPU cooling, but it wasn't enough powerful for that task and it went into the box, unused, for a long time. After beeing around in a swedish forum for PC overclockers, I saw another usage of a TEC - they called it a AirChiller. A closer exam of the idea revealed a very basic design that heve been used in portable refridgerators for a long time. Simple, I had the TEC and I wanted to give it a try. The TEC I'm using is rated 14.50V at 6A and measures 40 x 40 x 3.8mm.

* The second, and most important part was to find suitable heatsinks that wasn't too high. I wanted a low profile to be able to mount it at a side wall in a PC case. In a local electronics store, I found two of the heatsinks shown below. They measure 10 x 20 cm and the fin height is 20mm. They was rather cheap too.

* Third was the usage of fans. The most important is to cool the hot side - a more effective cooling will produce lower temperatures at the cold side. There is also a limit in how hot the TEC can be before it is destroyed. The selection of fans aren't completed yet - more testing have to be done to find a good match between coldness and noise levels.

To the construction of the cooling device

The basic idea was to put the TEC between the two heatsinks. The hot side was equipped with two 80mm fans for cooling and the cold side will have a 92mm fan for moving the chilly air from the device.
At the first prototype (not shown), was the cold and hot side heatsinks of same size. It worked, but it wasn't effective enough of two reasons: 1. the large heatsink took time to chill down and 2. the hot side heated up the cold side - the TEC is only 4mm thick and the space between the heatsinks aren't big. This made the unit not that effective as I wanted. A single, 120mm fan at the hot side wasn't enough either too cool down the TEC.
So, in prototype 2, I reduced the size of the cold side heatsink. This made a big difference - now I was able to reach sub-zero temperatures. Why? The smaller heatsink didn't heat up that much from the hot side. Two 80mm fans (I took something I found laying around - the performance of the fans are unknown) was attached to the hot side - this gave a better cooling compared to the 120mm (the 80mm fans covers a larger area of the heatsink). This design was also faster to reach temperatures below 10°C. The cold side heatsink was equipped with mounting brackets for a single 92mm fan.

To do

* Test out the final construction, with fans at both side. Mostly, the amount of moisture at the cold side have to be reduced before it can be put inside a PC.

* Testing with Papst 80mm fans for lower noise levels. The current fans are too noisy to be useful in a desktop PC. I have to see how much in temperature I have to sacrifice to get a more silent unit.

* Build a power supply for the TEC - it must be regulated to prevent over-voltage (and too much current) at the TEC, but still run it at it's optimum performance. The design of the power supply will be presented here at this page when I got something that works.

* Put it in the PC - this is mostly a question of space - does it fit?

	First, a suitable heatsink have to be found - one with low profile but enough surface to cool the TEC. Two of these was brought from a local electronics store - one of the was cut down and used for the cold side. For the hot side, it was kept in it's original size.
	The TEC with Arctic Silver II thermal compound on both sides. The heatsink shown is for the cold side.
	Put together - the two heatsinks bolted together with the TEC in between. A mount for a 92mm fan have been prepared at the cool side.
	The hot side - two 80mm fans serves as cooling together with the bigger heatsink for the TEC.
	Here is the TEC squeezed in between the two heatsinks.
	Time to give it a run for the money - first, a reading of the room temperature - a rather stable 21°C (small differences was unavoidable).
	After about 30 minutes running at 14.5V (taking a current of about 6A - it went lower when the TEC heated up), I got an amazing -0.5°C at the cold side!
	Another proof of the cold - a lot of moisture formed at the cold heatsink, dripping down at the table...