Michael Rigby-Jones wrote:

>Note that some TEC's are not very consistent in terms of their voltage drop, i.e. different units will have differening voltage drops for a given current and temperature difference.  Check the datasheet for you devices to see what the min/max voltage is.  If your devices have a significant spread then paralelling them will likely cause one TEC to work harder than the other.  Series connection would be prefered in this case.
>  
>
So two parallel channels of PWM would actually be more effective, then.  
That was actually my original design.  I think the only thing I 
necessarily need to add then is synchronization in order to gain a 
180deg phase shift between the PSUs, which can help the input caps much 
like a multi-phase system would.

>Response time will be limited by:
>
>a) Temperature difference between hot and cold sides (delta T).  The higher the temperature difference, the less the heat pumping ability of the TEC.  A 226Watt TEC can pump only pump 226 Watts when the dT = 0.  A some value of dT (dTmax), the TEC will be unable to pump any heat at all.
>
>b) Heat capacity of the load.  Obviously the bigger the chunk of metal etc. that is being heated/cooled, the slower the response.
>
>c) Thermal resistance of the junctions between TEC and source/load.  Big TECs need to be clamped down quite tightly (and very "squarely" to avoid cracking).
>
>d) Current capacity of power source
>
>1 second may or may not be a realistic expectation of response time.  The control system will of course affect this as well, what system are you using?
>  
>
When I made the comment about a 1s response time, I was refering to the 
PSU response time.  I'm aware of all of the limiting factors on my 
thermal system.

>Ripple simply degrades the TECs performance.  If you want to extract the maximum possible performance then ripple should be kept as low as is practical.  However, a 10% ripple factor typicaly reduces TEC performance by only 1%, so it's certainly a law of diminishing returns. I'm fairly sure a well designed single phase switch mode regulator should be able to meet your requirements.
>
>What really kills TEC's is using them in a thermostatic/bang-bang control scheme, i.e. switching full current through them on and off, rather than a proportional control system.
>
>Regards
>
>Mike
>
Thanks, that clarifies what I knew more as hearsay.  With all of this in 
mind, I think I can go back to my 2x single phase design.


Thanks to all for your input,
Brendan
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist