Cooling the Camera ... the trials and tribulations

This is my second attempt at a cooler box for my Canon XS. I have used it several times and I am pleased with the results of 33C or about 60F temperature drop below ambient while imaging and mounted on the telescope. I am still trying to drop the temperature further with another design.

The design was inspired by Gary Honis,, who also modified the filter on my camera. Many thanks Gary. His website has many tips on building these boxes.

The Goal was 40C or 72F temperature drop, light weight, reproducible and relatively inexpensive. The results differed and the weight at 1930 grams, or about 4.25 Lbs is heavy. I experimented with many series of thermoelectric modules and several experimental boxes and collected many CPU coolers, thermal sensors and other items before settling on final design(s) as I still have an ongoing project on the bench.

The box is 0.025" aluminum 7W X 4D X 4H", but it only requires 6.5W X 3.5D X 4"H. Smaller is better, and I suspect that 1/8" Acrylic might be better than aluminum, a plastic box with a small aluminum cold plate similar to my modern food freezer box, as the energy losses through the walls become noticeable beyond temperature differences of 20C . With only 1/2" Styrofoam, the outer walls can be 6C or 10.8F cooler than ambient when the inside is recording 30C below ambient.

The TEC(s) I settled on for this camera cooler box were 127 thermo-couple dual stage with a total power draw of 8 amperes, thus TEC2-25408. Since I generally ordered several units in each series I trialed, I noticed that with most series of TECs, there can be temperature variances between them under equal conditions. This dual stage series ranged from 44 to 53C on the desktop, that's the cold side of TEC on thermosensor which was notched into a 1/2" styrofoam with a heatsink on the hot side, basically monitoring only a small point source. At 10 volts, their most efficient voltage, they produced the greatest temperature drop, and at 12 volts the temperature change is as much as 15% warmer but the cooling power increases considerably.The increased voltage will increase the thermal transport across the thermocoupler junctions but net temperature will drop because of the increased heat produced.The units typically operate most efficiently at 70 to 75% Vmax.

At the desktop thermosensor, single stage TECs from 12702 to 12707 produced temperature drops from 12C to 38C using the same heat sink with a typical range of 28 to 35C and a few

units only dropping 12 to 16C which I labeled faulty. By the time I was testing 12708 and above, the temperature drops where typically 22 +/- 2C, and with a 12709, I was detecting temperatures of 58C on the heat sink with an ambient temperature of 21C, that is a lot of heat to remove. Beyond that I tested 12 and 16 ampere units relying on water cooling and they only produced modest drops of about 20C. But multistaging a 12706 and 12716, both 40 mm chips, wired in parallel and running at 10 volts, I have tested a smaller single cooling unit, in a poorly insulated water cooled box, without a camera, and reached 38C drop, but it is still in development stage. Also, from multistaging tests, I could not appreciate any performance gain until the second stage was as least twice the power of the first such as 12703 and 12706 and they seem to optimize at a ratio of 1:2 2/3, such as a 12703 and 12708 that repeatedly yielded 43C temperature drop at the probe and a 31C drop in the smaller water cooled box previously mentioned. Single stage Tec unit 40mm in size from 12702 to 12709 can easily be purchased on Ebay for $6 to $7 USD a piece, I could not find any 12704 TECs on Ebay. The higher power units cost $20+ USD, and I kept with 40 mm TECs to better match my methods of cooling.
To cool a camera box of about 80 +/- 5 cubic inch with 1/2" insulation 25C, a 40 +/-5 watt single stage TEC should be sufficient. Considerably more power is required to go past that because conductive and radiant losses become excessive and over compensation of power is required for greater temperature deltas. Because the losses are exponenti
al, the effective losses at lower temper
ature deltas are relatively small, but even dropping the temperature an additional 5 degrees inside the styrofoam insulated box (38C) and the outside of the box was 10C, 18F, cooler than ambient versus 6C cooler when the inside of the box is 30C below ambient.

The adjustable, 10v - 15v, 12 volt regulated power supply(s) I purchased were rated at 30 Amp and can be found on Ebay for about $50 - $60. With 20 Ampere draw, they can get quite warm, finding spots with the laser thermometer, $20 Ebay, of 60+C and noting the regulator's internal cooling fan was operating about 50% of the time, I made a simple well ventilated stand with 2, 80 mm fans attached to a 1" foam spacer with large slots for ventilation that keeps the power supply's temperature below 45C.

The USB thermosensors, also purchased on Ebay, were about $7 each. I have 4 of them and each can have some variance, so I calibrated them with a mercury thermometer.

The TECsĀ  I used to get 33C drop at the image sensor were orderd through with a minimum purchase of 5 units the cost was $125 USD delivered and they sent TEC2-25408 by mistake, the added power was useful. These were manufactured by seems to sell a similar product and with only a $79 US minimum order.

The Cpu Coolers, Scythe Shuriken Rev. B Low Profile CPU Cooler, were purchased at a local computer store for about $40 each and the LGA775 mounting brackets worked very well and not too heavy at about 400 grams with the fan (320 gram cooler and 80 gram fan). The heat pipes help considerably but their effect can vary depending on their orientation. As for cooling capability, up 12707 series of TEC units, these coolers will typically produce desktop probe reading from 28 to 34C below ambient, and at least at the desktop, I could stack a 12703 and 12709 in parallel and reach 42C below ambient. Beyond that, a much larger mass cooler is required or switch to liquid cooling. I tried units up to 600 grams and the drop of 1 or 2C was considered to little to justify the added weight and a 280 gram unit with heat pipes lost performance beyond 12703 TECs.