How do I get an itty bitty laboratory grade freezer? Can I build one?

It would be awesome if it could fit on my shelf and freeze up to -50°C (-58°F) maybe even lower, if possible. I want do deepfreeze some vacuum packed spices. I know it's overkill and I do not really care. (Oh actually, I do...) Would some big Peltier Elements suffice? I guess they would not be very efficient, and this would be a definite plus? Maybe if heat is transported away from the element with a standard CPU water cooling system? I'm afraigt they're not up to the task. Even getting a portable fridge cool takes ages... What about modifying a 'normal' small freezer? One could maybe just control the pump with another kind of controller, ie an arduino with a PT1000 and a relais... I read though that one might need another cooling agent capable of even lower temperatures then those used in commercial grade freezers? Any links to tutorials, builds or products would be deeply appreciated.

Picture of How do I get an itty bitty laboratory grade freezer? Can I build one?
Its doable with Peltier elements, but not inexpensive I have developed a water cooled rig for -45 C that gets there in less than 3 minutes from +30 C, under a 20W load. It uses 4 BIG cells, in three layers, with a 1 kg watercooled heat sink dumping nearly 400W to ambient.
Maniacy (author)  steveastrouk7 years ago
hmm, sounds awesome. I'd love to see that too! Do you use another cooling agent to transport the heat from the area to be cooled to the cool side of the peltier element? What did you build it for? Greetz Martin
No, we are trying to cool a special sample cell about 10mm long and 6 in diameter as fast as possible. We take a lot of care to mount the cells in absolutely flat heatsinks, and we take even more care to insulate the hell out of anything on the very cold side. If you want a micro-freezer, you can reduce the thermal load on the cells to the point you don't need a 400W system like mine, but -50 isn't at all easy with Peltier cells, their efficiency drops through the floor in the negative temperature region - hence the massive investment in cells (3 x 150W +1 x 60W rated cell power) for only a 20W mechanical load and the nasty thermal losses. As an aside, I am also aiming to achieve +/- 0.035 C precision in that 3 minute cooldown. THAT really is a trick.
I would love to see that!