MOSFET working temperature

Discussion in 'General Electronics Chat' started by samuel.whiskers, Aug 26, 2014.

  1. samuel.whiskers

    Thread Starter Member

    Mar 17, 2014
    I've built a small dummy load - it uses two IRF540N MOSFETs that are driven by op-amps. Works well....
    Datasheet for MOSFET

    It runs quite hot... both FETs are secured to a reasonable sized heatsink - approx 4x3" or so in size. They have silicon pads and grease under them. At 2A total load at 12V (1A per FET - approx 11W each), the heatsink runs around 60°C and the FET body around 100°C.

    Should I declare this a maximum working load? Any suggestions to the highest FET temperature you'd be happy to use in this application??

  2. wayneh


    Sep 9, 2010
    So the MOSFETs themselves are the load? It's more customary to let a component meant for heat dissipation do that chore, for instance a light bulb, heater element, or a power resistor. Of course the MOSFET can dissipate quite a bit also, if that's enough for your application.
  3. kubeek

    AAC Fanatic!

    Sep 20, 2005
    Your numbers dont look right if the mosfet is dissipating 11W. Datsheet says 1.15K/W junction to case resistance, and another 0.5 K/W case to heatsink, so if your silicon pad adds another 0.5K/W then the body of the transistor should not be hotter than 11*2.15 above the heatsink, which is about 83°C.
    (yes the measured temperature on the case will be a bit lower than the actual juction temp, but my guess is that not by more than about 5°C)

    So either you are dissipating more than you think, your temperature measurements are off, or the transfer between the transistor and the heatsink is really bad.

    Can you show your circuit? If you could connect the drains together and bolt them straight to the heatsink without the silicon pad you will get a lot better performance.

    But anyway, the maximum junction temperature is 175°C. I would try to keep it below say 120°C for permanent use, to get longer life from the transistors, so you should be still ok with even a bit more power dissipated.
  4. MikeML

    AAC Fanatic!

    Oct 2, 2009
    Here is a way to move most of the dissipation from the NFET to the source resistor. As shown below, with only 12W of dissipation in the NFET (red), the total load current can go to 3.3A (green), and the total power can go to 40W (dk. blue).

    Assuming you have a 50W resistor...
    • D193.jpg
      File size:
      290.5 KB
  5. samuel.whiskers

    Thread Starter Member

    Mar 17, 2014
    Will post circuit soon - although it is the same as posted below, with a 1Ω load resistor....
    Temp measurements seem accurate, I have a pair of previously tested mercury thermometers, my meter reads to within 0.25°C of these at room temp, should test it at 100°C too.....
  6. samuel.whiskers

    Thread Starter Member

    Mar 17, 2014
    I'm using a 1Ω resistor (12x12Ω 5W paralleled), which dissipates 1W @ 1A, I had thought of using a higher value to dissipate more of the heat, seems worth a try, thanks.
  7. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    Swap the silicon pad for a proper mica washer, a thin one.

    You can split them thinner too for high heat low voltage applications.
  8. samuel.whiskers

    Thread Starter Member

    Mar 17, 2014
    The answer was on my bench all along.... a little 60mm PC fan - pointed at the heatsink fins, the FETs are only reading 85°C with the box drawing 3A (1.5A each FET)....
  9. samuel.whiskers

    Thread Starter Member

    Mar 17, 2014
    Grabbed some on ebay just now.... will come in handy I'm sure....
  10. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    Every single electronic load in our company uses an active element (junction or field effect transistor) to dissipate energy.

    That keeps them snappy when applying transient loads.

    If you can isolate the heat sink you can loose the isolating pad altogether. Don't loose the si grease with it, a conductive medium is always required.