HELP-Logic Level Power MOSFETs keep failing

Discussion in 'The Projects Forum' started by MartyMitchell, Sep 11, 2012.

  1. MartyMitchell

    Thread Starter New Member

    Sep 11, 2012
    23
    0
    I'm trying to switch a 12v air pump (20A) on and off with a logic level power MOSFET and a microcontroller (0-5v). See attached schematic. Works great about 5 cycles, then blows up the MOSFET, leaving the Drain to Source partially open always. Heat sink on MOSFET, gets warm but not hot. Rds ON is 17 mOhm @ 10A, 10V.

    What am I doing wrong? Any suggestions or ideas would be great! I can use another part if you know of one that would work better and not blow up.

    Part info - http://www.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=568-9506-5-ND
     
  2. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
  3. MartyMitchell

    Thread Starter New Member

    Sep 11, 2012
    23
    0
    Hi Mike - You're right. I checked out figure 3 and it looks like the safe zone is 50 amps for no more than around 100ms. So I guess I must be operating beyond the safe zone.

    The diode is to protect the MOSFET from high voltage spikes when the pump circuit turns off.

    Can you think of a way to solve this puzzle? Different part? Current limiter?

    Thanks,
    Marty
     
  4. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,387
    1,605
    If it works OK for 5 times, but then does not turn on the 6th time it sounds like the turn off spike is taking out the FET.

    If the motor makes a positive spike the diode should catch it. If it makes a negative spike that will come thru the body diode of the FET. I don't have any real experience to say if the latter is possible, but I'm going to follow this thread to see what your results are. I need to drive a similar pump soon.
     
  5. MartyMitchell

    Thread Starter New Member

    Sep 11, 2012
    23
    0
    Hi Ernie,

    Thanks for your thoughts. Might be the case, I'm not sure. Do you think I could solve the problem, if it is caused by inductance kick, by slowing down the microcontroller's high/low states? In other words, drive the signal into an R-C network to reduce the high speed switching.

    Any other ideas would be great.

    Thanks,

    Marty
     
  6. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    The inductive kick as an inductive load in the drain of an NFET turns off increases the drain voltage. The snubber diode is connected such that the drain voltage is clamped one diode drop above the battery voltage. However, if the snubber is a slow turn-on diode, then it may allow Vds to shoot up before the diode turns on. That is why I asked what part number you were using for the snubber.
     
  7. MartyMitchell

    Thread Starter New Member

    Sep 11, 2012
    23
    0
    The diode is a 1N4001. The reverse voltage rating is only 50v. Not sure how fast or slow they are.
     
  8. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    If the motor running current is 20A, then the snubber will have to carry 20A as the FET shuts off. The 1N4001 is too light for that. Also, if the motor is a segmented commutator brush-type motor, the brushes make and break because the motor is spinning. I dont know what voltages might appear at the FET drain due to the brush arcing. Have you looked at it with a 'scope?
     
    Last edited: Sep 12, 2012
  9. crutschow

    Expert

    Mar 14, 2008
    13,004
    3,232
    Have you checked the diode to see if it has blown?
     
  10. MartyMitchell

    Thread Starter New Member

    Sep 11, 2012
    23
    0
    The diode has not blown when measured with the diode checker on my DVM. Still 0.7v and open the other way.

    Mike-you're right, I will need to find a new diode that can handle 20a, the amount of current that may want to flow backwards when the MOSFET turns off. What about the voltage? Should I find one good for 400v?

    I'll put the motor on a scope and see what kind of votages are present and report back.

    Marty
     
  11. MartyMitchell

    Thread Starter New Member

    Sep 11, 2012
    23
    0
  12. JDT

    Well-Known Member

    Feb 12, 2009
    658
    85
    This pump will draw a very high current on start-up. Your MOSFET needs to be oversized in respect of it's current rating. Probably by about 6 - 10 times.

    Also you need to drive the MOSFET properly. The gate has a high capacity (it will have if you use the correct MOSFET) and needs to be driven with a proper gate driver that can deliver AMPS to the gate so that it charges and discharges FAST.

    Also vital that during power-up, the MOSFET cannot be switched on before the supply volts has ramped up to the correct supply voltage. Some gate driver IC's have undervolt lock-out included.

    A partially driven, partially turned-on MOSFET will instantly destruct with a high current load due to localized heating. I know, I've done it!
     
  13. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    A common 1N diode could be to slow to turn on therefore not properly clamping the spike to the supply voltage. I would use something faster. Unfortunately you will hardly find specs for that in their datasheets. Try a UF4001 for example.
    (also related: http://cds.linear.com/docs/Application Note/an122f.pdf)


    [​IMG]
     
    RamaD likes this.
  14. donho

    New Member

    May 16, 2010
    5
    0
    Forget the MOSFET when controlling an inductive load. Use a simple bipolar transistor to drive a solid-state relay. The SS Relay will last for years and much less expensive than the high power MOSFET.
     
Loading...