Hi Kalle, >>> combo PJ/heat sink/12V fan for ~$25 (HSC# 16106). It was originally >>> intended as a cooler for non-metallic Pentium Chips. A drop of water >>> will boil on one side and freeze on the other. >> As I recall, you get about 20-25 deg C temperature difference > The temp difference can be over 150 degC. The problems are when the > hot side gets hotter than the solder that is used in the assembly and > the whole unit destructs. > Just had a look at the RS components (UK) catalog at some specs. > Max temp differential is 65 .. 70 degC for the single stage units Thanks, I stand corrected. Unstated in my comments, but I assumed that for $25 the device in question was a single stage unit. > You have to build a pyramid with more cooler required to cool > the cold one with a minimum recommended ratio of 2:1 but better > at 4:1 unless you only use a small fraction of the coldest > coolers capacity. Expensive. > > -50 degC > XXXXX > 0 degC =========== > XXXXX XXXXX > 50 degC ======================= > XXXXX XXXXX XXXXX XXXXX > 100 degC ======================= > ||||||||||||||||||||||| > BIG fan Remember it's temperature difference. So for any given final target temperature (say -50 deg C), if you start with a lower temperature you need fewer stages. An alternate scheme is to cool the "hottest" side stage with something better than ambient air. For example, pump chilled fluid (alcohol or anti-freeze) through passages in the heat sink. Modifying Kalle's example above: -50 degC XXXXX 0 degC =========== <-- heat sink OOOOOOOOOOO <-- passages for chilled fluid =========== <-- heat sink In astronomy (cooling CCD cameras), I've heard of a small 12V pump in a bucket of ice water. Portable and environmentally friendly -- at the end of the observing run, you can water the plants around your site. This scheme has much lower power consumption for an increase in "mess" (cooling fluid, ice, hoses, etc). Lee Jones