Peltier circuit (common emitter) transistor overheating

Discussion in 'The Projects Forum' started by ponjavic, May 4, 2011.

  1. ponjavic

    Thread Starter Member

    May 4, 2011
    37
    0
    I'm designing a circuit applying a variable voltage to a peltier element.
    This voltage is controlled by a Labjack U6 analogue output.

    My powersource is 200W 13.8V regulated DC

    Peltier is maximum 12V maximum 9A

    I'm using two resistors of 10kohm +-10%

    The darlington pair is:
    http://uk.rs-online.com/web/search/...od=searchProducts&searchTerm=485-9042&x=0&y=0

    All I want is to be able to accurately vary the voltage applied to the peltier so any alternate circuit design would be welcome.

    I've read that PWM is not optimal for peltier elements.

    Here is the circuit, this circuit might be incorrect as well, this is one of the first circuits I ever built :p

    Naturally this is horrible. When operating at normal voltages about 3-4V the transistor will have to dissipate the reminder. At a current of a few amps the dissipation is 10s of watts which it cannot handle causing it to overheat. So what is a proper setup for my needs?
     
  2. wayneh

    Expert

    Sep 9, 2010
    12,145
    3,055
    You might enjoy reading this project, regarding controlling a TEC. It even involves a LabJack.

    Bottom line, I think you'll want to abandon your "linear" control approach, for exactly the reason you've discovered. Instead, think about on/off control, such as the thermostat I used in that project, or PWM.

    PWM is OK for a TEC (= Peltier). What's hard on them is rapid temperature changes, for instance if the TEC is switched from full current at one polarity to full current at the opposite polarity. Expansion and contraction will tear them apart. But PWM is too fast for that, and the TEC is holding at roughly a constant temperature, so it's happy.

    Your LabJack can even supply the PWM signal, but you'll need to drive a logic-level MOSFET to actually switch that big current. And check to be sure the lowest PWM frequency of the LabJack is not still too high for the TEC. Maybe there is no such thing as "too high", only too low.
     
  3. ponjavic

    Thread Starter Member

    May 4, 2011
    37
    0
    Thanks wayneh, very much appreciated.

    I wouldn't mind giving it a go so I'll order a mosfet for tomorrow.
    Anything particular other than current capacity I need to consider?

    Would this do for example?
    http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=7104736

    I have never used mosfets and only quickly read up on them but I assume it's just a high frequency digital switch.

    Also would I need anything else than two resistors in the circuit?

    That project was very enjoyable reading and it might be useful in the future, but it seems a bit over my head at the moment :)

    Would this be the correct circuit to use:
    "An example of using the MOSFET as a switch"
    http://www.electronics-tutorials.ws/transistor/tran_7.html
     
  4. ponjavic

    Thread Starter Member

    May 4, 2011
    37
    0
    Ok so apparently I'll need a buck converter like here in order to smooth the PWM into a dc signal:
    http://schmidt-walter.eit.h-da.de/smps_e/abw_smps_e.html

    I wouldn't mind buying one but there don't seem to be any.
    Not completely sure about choosing the components. Would this work for the inductor?
    http://uk.farnell.com/wuerth-elektronik/744824101/choke-common-mode-1mh-10a/dp/1636292

    It doesn't look like the capacitance is as critical, am I right in this? Any particular criteria on the diode?
     
  5. wayneh

    Expert

    Sep 9, 2010
    12,145
    3,055
    1) The current capacity should be ~4X your expected current level, to allow room for general safety. 2) The output voltage of your LabJack is...what? I think you'll need a logic-level FET that is full on at 5v. A regular FET isn't fully on until its gate sees 12v or so. You definitely want whatever FET you use to be fully on, otherwise it'll get too hot. 3) I usually use thru-hole, not surface mount devices. Just be sure which you want. 4) What frequency will your LabJack use for PWM? Things get dicey at high frequency and others here may need to help you design the driver.

    IMHO, no, because the current spec is is too low. Could it work? Maybe, but I'd go for more current. And consider the form factor. Do you really want surface mount?

    Sort of. It's a transistor, and it's capable of "high" frequency, and MOSFETs make excellent switches because they have very little resistance when fully on. But "high frequency digital switch" defines an application, nothing inherent about the transistor itself. A MOSFET makes an excellent slow, analog switch too.

    That's going to depend in part on what the LabJack is putting out, voltage and frequency. It's possible you won't even need the two resistors, just the FET to turn the cooler on or off.

    Yup, that's pretty much it except for the details of your project.
     
    Last edited: May 4, 2011
  6. ponjavic

    Thread Starter Member

    May 4, 2011
    37
    0
    Very helpful, I guess I'll have to find another mosfet as you're completely right, it won't be fully open at 5V.

    http://uk.farnell.com/stmicroelectronics/sts12nf30l/mosfet-n-logic-so-8/dp/9946381

    It's not 3x the current rating but I can't really find any logic mosfet at RS or FARNELL that does it. Also I wont be operating at the maximum capacity of the peltier (9A) anyways. I'll probably hover around 6 or less.

    The labjack will be running at 5V ~200 or ~800Hz which I have seen in use before for peltiers, so I hope this is suitable. As per my previous post I'll be building a buck converter for smoothing in order to not strain the peltier and cause large inefficiencies.
     
  7. wayneh

    Expert

    Sep 9, 2010
    12,145
    3,055
    I don't know if this makes any sense. I don't believe the TEC manufacturers recommend this approach. Straight PWM is the usual approach. Are you saying you want to provide a lower voltage to the TEC, so that it's drawing only 6A instead of 9 or so at full voltage? In that case I understand.
     
  8. ponjavic

    Thread Starter Member

    May 4, 2011
    37
    0
    This is where I got the idea from.

    http://fieldlines.com/board/index.p...d525c57cc666&topic=144978.msg982948#msg982948

    In order to avoid stresses induced in the peltier...

    I'm saying that the buck converter will turn the PWM into a dc signal, so 50% PWM --> 6V or so DC.

    I do have the option of running straight PWM as the mosfets have arrived. I guess I could try that and see if anything burns out or not :) they're easily replaced anyways.
     
  9. wayneh

    Expert

    Sep 9, 2010
    12,145
    3,055
    Ah, got it. A low-pass filter. That's a nice finishing touch but it's my understanding that TEC modules are widely used (by hobbyists, anyway) with straight PWM. It's not ideal for efficiency and long life, but I think you can get away with it just fine for non-production stuff.
     
  10. ponjavic

    Thread Starter Member

    May 4, 2011
    37
    0
    That's good to know if my converter absolutely fails which it's likely to do :)

    I'll try it as it's a very nice learning experience.

    Any idea why the zener diode is absolutely required in a buck converter (or a low pass filter in a MOSFET setup)? I'm assuming it has something to do with letting the inductor actually be in a circuit while the MOSFET is off. It's still slightly confusing to me and according to spice the circuit works without a diode.
     
Loading...