TEC Peltier calculation software

[doktor pyta]

I'm designing a cooling system that uses two stage TEC to cool object to -35'C (ambient =30'C, Heat sink 0,6K/W, power applied on cool side 1W). The power supply can give 260W of continuous power. I found out that this is not trivial task and the TECs have to work close to optimum currents. Does anyone knows good and free software to design such system. I personally use Kryotherm and I divide my system in two parts and calculate them separately but I assume there must be some better way.

[rs20]

It's just a simple system of equations? Heat transported per amp is a property of the TEC; dissipation is I^2 R (hence the need to not have too much current or else the TEC will be overwhelmed by its own heat dissipation.

Can be solved as closed form, or you can fall back on Solver in Excel. Really no need for any specific software, although I'm sure easier suggestions will come.

[ajb]

It's just a simple system of equations? Heat transported per amp is a property of the TEC; dissipation is I^2 R (hence the need to not have too much current or else the TEC will be overwhelmed by its own heat dissipation.

You could do that if you had an accurate model of the TEC you're using, but often you just have a handful of graphs from the manufacturer.  Coefficient of performance decreases with dT, and assuming you're controlling the cold side temperature then the hot side (and hence dT) goes up as you increase heat load, so there's a substantial feedback effect that prevents you from just picking your operating point from the graph and being done with it.

There are thermoelectrical models for TECs you could use, but when I've had to design single-stage TEC solutions in the past I've just created a basic thermal model in a spreadsheet, ballparked my initial figures, then taken the results back to the graphs to refine the TEC operating point and adjust the model inputs.  With a few iterations you can pretty quickly get your operating point to within the range of a reasonable control system.  A two stage system will take a bit more juggling to figure out what your intermediate temp and heat load are if it's anything more complex than literally just two identical TECs stuck together, but the same method will work.

Laird has a program called AZTEC that will select parts and spit out graphs based on input parameters, but of course it's limited to their TECs, and trying to match their part numbers to other manufacturer's equivalents is a pain. 

[doktor pyta]

I think there is a typo in the first article by Mr Woodward.
instead:      T1 = (-P*Itec + Itec2 *Rp/2 + Q1)/(C1 + Cp) + (Q1 + Itec2 *Rp)/(Ch + T3)
should be    T1 = (-P*Itec + Itec2 *Rp/2 + Q1)/(C1 + Cp) + (Q1 + Itec2 *Rp)/Ch + T3
otherwise the units doesn't match.
Am I right?

[CatalinaWOW]

You will find much information of interest at   http://tetech.com/

The need for a two stage cooler may imply that one of your stages is intended to be outside the range covered by the data sheets.  There are papers on the tetech site which can help in estimating what will happen. 

[johansen]

I found out that this is not trivial task and the TECs have to work close to optimum currents.

they also have to be the optimal size. since you need about 70K difference in temperature, you should split that temperature in half, the first stage needs to pull 1 watt of heat out of -35C and dump it into 0 C.

so you design it to do that, measure or calculate how much heat it will generate, then design the second stage to pull that heat out of 0C and dump it at 30C

i made a two stage tec cooler to cool a chip.. dunno how much heat it generates, but i cut a standard 40mm square tec, i cut 10mm off one side of it, glued it onto another 40mm square tec with an aluminum heat spreader between them.

i did not know at the time that you have to seal them so moisture doesn't get in.. anyhow it worked but it may not have been an optimal size. but it worked.

[doktor pyta]

You will find much information of interest at   http://tetech.com/

The need for a two stage cooler may imply that one of your stages is intended to be outside the range covered by the data sheets.  There are papers on the tetech site which can help in estimating what will happen.

Wow, there are a lot of great papers in their Download section.
Thanks !

[TiN]

Also since you looking for 65K temperature delta, your insulation quality of both cold-side to ambient and cold-side to hot-side is key importance to avoid thermal transfer as much as possible, where you don't want it to happen. This kind of setup is not something you can test without getting everything insulated and assembled they way it supposed to be used.

[John Heath]

I do not mean to be pessimist however there is a difference between theory and applied technology. Rule of thumb take any spec and divide by two to be on the safe side. For a 30C gradiant to a - 30C at 1 watt you could be looking at a 3 stage peltier system. Either that or brute force  freon pump then peltier to fine tune the final stage.

[rs20]

I do not mean to be pessimist however there is a difference between theory and applied technology.

Datasheets are measurements, not theories.

Rule of thumb take any spec and divide by two to be on the safe side.

That's nuts!

[TiN]

If going down with phase-change system using freon, it's much easier than just to build small one without any TECs to get -50..-40°C range and then just use resistive heaters at cold plate to bring temperature to desired level with simple SSR/PID controller. Compressor, condensor, fan and piping can be easily reused from automobile AC unit, or similar. Larger units from house AC systems aren't optimal as they not designed for idle/no load operation, which cause compressor overheating and possibility of hydraulic damage. I'd think in the end power consumption of whole unit would be in range or smaller than TEC rig.

Building phase-change loop seems complex at first glance, but it's not that difficult given one investment into toolkit (copper tube brazing gear, vacuum pump, manifold with manometers/valves, bottle of R404/R507 freon).
There are some freeware software tools on web to calculate such systems as well.. Sorry for derailing thread.

[johansen]

If going down with phase-change system using freon, it's much easier than just to build small one without any TECs to get -50..-40°C range and then just use resistive heaters at cold plate to bring temperature to desired level with simple SSR/PID controller. Compressor, condensor, fan and piping can be easily reused from automobile AC unit, or similar. Larger units from house AC systems aren't optimal as they not designed for idle/no load operation, which cause compressor overheating and possibility of hydraulic damage. I'd think in the end power consumption of whole unit would be in range or smaller than TEC rig.

Building phase-change loop seems complex at first glance, but it's not that difficult given one investment into toolkit (copper tube brazing gear, vacuum pump, manifold with manometers/valves, bottle of R404/R507 freon).
There are some freeware software tools on web to calculate such systems as well.. Sorry for derailing thread.

since he only needs 1 watt of cooling it can be done with the guts from an office water cooler. may need to replace the R-134A with propane, if you can get real propane that boils at -40C. the stuff i can get here boils at -30C because its mixed with butane. it is probably impossible to get to -50 C with an unmodified office fridge cooler. you can try crimping the capillary tube to effectively make it smaller, but its very easy to crimp it shut while trying to do this.

however, 65K of cooling can almost be achieved with 1 TEC. two of them puts the COP of each one in the most efficient range.

this should be rather trival to do, provided the TEC's are sized correctly. as i mentioned.. i cut one in thirds.. i believe i used an aluminum oxide grinding wheel to do that.. but i'd rather use a diamond wheel next time.