Help with the control of a TEC (thermoelectric cooler ) on FPA(focal plane array )

Discussion in 'The Projects Forum' started by qisaiman, Feb 18, 2009.

  1. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    hi ,everyone, I am dealing with a temperature control job. And the object is 200K with a 0.05C stability.
    Demand on TEC control is like following :
    The maximum current supply of the TEC shall be less than 4.2 A for 7.4 V such that the maximum electrical power is less than 31 W.
    At first I prefer a PWM control based on a H-bridge, but there are some app tips from Marlow and RMT which stated a DC current is more efficient than PWM. I wander if this is true.
    I’d like to use a H-bridge shown below, and the LC filter to smooth the current , but I don’t know how to limit both the voltage and current within the requirment mentioned before in the same time , thanks alot for any suggestion.
    I had known the high-side driver issue, and I will employ an IC to simply my design , but during one period, how the voltage changed on the TEC1 and TEC2 . I want to figure out how to operate the MOSFETs using PWM. For example, turn off Q1 and Q3, turn on Q2 , and a PWM on Q4 to result a current from TEC2 to TEC1, is that right?
    An last, any programmable DC current supply scheme is appreicated.
     
  2. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    is there anything I did not make clear?
     
  3. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    If this thing is resistive in nature, as suggested by your graph, then Ohm's Law will allow you to limit current by limiting voltage.

    Why do you want to use an H-bridge? Is this for heating as well as cooling?
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Do you mean 200° Kelvin aka -73.15°Celsius aka -99.67°Fahrenheit?

    And do you really mean to control that temperature within 0.05°Celsius? :confused:
    If the TEC is a Peltier device, it will be subjected to much less stress if DC is used vs PWM. Note that Peltier devices will have their life shortened (perhaps drastically) if their temperature is rapidly cycled due to metal fatigue.
    Inductors will keep the current relatively constant.
    However, do you need both heating AND cooling? If not, you will probably be better off to use a synchronous rectifier configuration rather than an H-bridge.
     
  5. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    Is this for a thermal imaging camera?
     
  6. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    yes I think it is an IR application.
     
  7. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    the FPA device demands a constant tempature to work well. so I guess both heating and cooling will be better.
     
  8. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    I would make L1, L2, C1, and C21 as large as practical.

    You will require excellent resolution on your thermocouple ADC, as well as good ice-point compensation in your thermocouple amp.

    I do not envy you the task of compensating for any thermal hysteresis.
     
  9. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    yes ,the datasheet requires the device to work under -73.15°Celsius , and temperature within 0.1°Celsius will also be OK.
    actually ,it works like this: after power on ,the circuit will keep cooling the device till about -73 Celsius, and mantain the tempature around it ;
    if it's needed to be heated slowly after power off,I do not make sure now.
    by the way , what is a synchronous rectifier configuration ?
    thank alot.
     
  10. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    In the device ,there is a NTC and other compents , a measure of the voltage on the pin of FPA can give a tempature inside , I will use a DSP with ADC module,and a PID control method , if PWM is used, the L1, L2, C1, and C21 is like a filter, and a transform function is upon to L1, L2, C1, and C21 , thus maybe it's not as large as good .
    I did not consider the ice-point and thermal hysteresis at first ,thanks for reminding .
     
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
  12. qisaiman

    Thread Starter New Member

    Feb 11, 2009
    8
    0
    the TEC inside is a four stage one ,and the device area is quite small ,so the thermal load can be small too, and when working ,the ambient tempeature is about zero Celsius. thus I believe a good heat sink will be enough.
     
  13. DedeHai

    Active Member

    Jan 22, 2009
    39
    0
    How do you plan on controlling the temperature? You mentioned a DSP, so i suppose you use a digital control algorithm? Heating/Cooling applications can be hard to control because of the dead-time between output signal and sensor feedback. I hope you are familiar with control loops (i.e. anti windup) or this will never ever work within the tolerance of 0.05°C.
    What you could use is a TL494 (see http://forum.allaboutcircuits.com/showthread.php?p=124641#post124641). I used it in a buck-converter like configuration, but instead of having an output capacitor i use a series resistor as a current sense and feed that voltage back into the opamp of the TL494.
    I used a similar circuit like the one shown here (page 27):
    http://focus.tij.co.jp/jp/lit/an/slva001d/slva001d.pdf
    Just much simpler. I only used about 6-8 external components, but i only need 100mA output. Sorry for not posting my circuit, but I only have a hand-drawn sketch and no scanner :)
     
    Last edited: Mar 8, 2009
  14. ifixit

    Distinguished Member

    Nov 20, 2008
    639
    108
  15. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    The very small sensor is what will be a nightmare to keep within even 1°, the mass difference between the coolers and object cooled is huge, as well as the temperature gradient across the coolers.

    Have you thought of using multiple controlled stages, with a controlled small final temp change at sensor, rather than a single large cooling unit? Or is this already broken up into calibrated areas?

    e.g. Outside box/insulation, inside is kept x degrees cooler, another box held regulated x cooler than outer shell, then a final small TEC on the array itself.

    With a 100 degree gradient, simply not cooling should give a quick enough 'warm up' so that heating wouldn't be required, except maybe on the final, small TEC.
     
Loading...