High Side Mosfet Burn

Discussion in 'The Projects Forum' started by circuit1234, May 25, 2016.

  1. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Hi I am trying to drive Half bridge converter using IR2101 and MOSFET SIHG20N50C-E3.

    The voltage i need at the output is 100Volt . at my first attempt i vary my voltage until 72volt and it vary nicely with some noise. after that high side mosfet spoil .followed by IR2101 , and optocoupler which is connected to TL494.

    I replace the mosfet , IR2101 and isolator , now the condition is it can go upto 16volt after that mosfet is spoiled again

    Please help me to figure out the issue
     
  2. Sensacell

    Well-Known Member

    Jun 19, 2012
    1,131
    267
    Schematic?
     
  3. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Frequency is 100khz , current is 1.5amps
     
    Last edited: Jun 6, 2016
  4. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    With this setup you need to make sure that the high side fet never stays on for a long time, and that you have enough dead time between the top closing and bottom opening and vice versa.
    Out of curiosity, what kind of software did you use for that schematic?
     
  5. crutschow

    Expert

    Mar 14, 2008
    13,025
    3,236
    As kubeek said, the input signals need to be have a dead time so that both transistors aren't momentarily on at the same time.
    How you generating the signals?
     
  6. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    * Software is cadstar
    * But first the circuit was working upto 72volt, once break down occur it is not going above 16 volt ,I tried three times , 3 times also same high side mosfet blown with driver and isolator . and I believe if dead time is the issue both mosfet will blow
     
  7. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    * signals are generating using TL494
     
  8. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    "kubeek, post: 1002938, member: 4023"]With this setup you need to make sure that the high side fet never stays on for a long time, and that you have enough dead time between the top closing and bottom opening and vice versa.
    Out of curiosity, what kind of software did you use for that schematic?[/QUOTE]

    The wave form for the mosfet gate pulse
    Does it looks good, green is the high side pulse
     
    Last edited: Jun 6, 2016
  9. crutschow

    Expert

    Mar 14, 2008
    13,025
    3,236
    I meant what is generating the signals to the TL494?

    The signals you posted have no overlap between when the green signal goes low and the yellow signal goes low. There needs to be a non-overlap time between those two.

    Only one MOSFET will blow (the one dissipating the most power during the time they are both ON).
     
  10. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    Actually I don´t think they do, he is measuring the voltage at the gate, not Vgs, so there seems to be roughly 3us of deadtime. However the edges look really slow.
    Why do you have 20 ohm resistor between gate and source? That looks like a really bad move, that should in the order of 10K.
    And more to the point, why is that a completely different schematic than before?
     
    circuit1234 likes this.
  11. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Hi 20 ohm resistor form gate to source is incorrect its 1kohm resistor .
    After the above schematic i modify , try to add some snubber .
    regarding the dead time, is 3us is enough dead time or do i need to increase more .

    If i dont give the Bus voltage the waveform is like Print 001. and when i give the bus voltage the waveform looks like print 000a.
    i worried about the red portion in the pic , is it normal

    Even after adding the snubber the circuit cant go above 42volt., anything else i need to try again
     
    Last edited: Jun 6, 2016
  12. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Hi crutschow
    Signals are generated by TL494 itself depends upon the feedback of voltage and the frequency . I hope this is what you are expecting .
    May i know what is non overlap time . is that the dead time you mean .
    the waveform i posted above is gate pulse of the mosfet .
    May i know which waveform you want
     
  13. crutschow

    Expert

    Mar 14, 2008
    13,025
    3,236
    True when the yellow signal goes low and the green goes high, but not when yellow goes high and green goes low.
     
    circuit1234 likes this.
  14. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Hi crutschow
    Can you please explain what is the difference between these two situation.
    from my understanding if there is problem of interchanging the signal it will show when the circuit on.
    But now circuit can work until 2volt to 42volt ( by changing the voltage at the TL494 input ) above that only the driver and the mosfet burns
    # So from your point do i need to interchange the signal
     
  15. crutschow

    Expert

    Mar 14, 2008
    13,025
    3,236
    No.
    You need to provide a non-overlap time for both transition points, not just one.
    Thus the green must go low before the yellow goes high (only one signal high at a time).

    The circuit burns at higher voltage because the current spike and dissipation increases with voltage.
    At some point the dissipation becomes high enough to zap the device.
     
    circuit1234 likes this.
  16. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    I still think the green does go low before the yellow goes high, but it it riding on the output voltage so it looks as if it did not.
    If we knew what the output load was things would be much easier.
     
    circuit1234 likes this.
  17. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Hi
    Can you please explain more about , "non overlap time for both transition points"
    With my starting knowledge in this field , and with how i prob what my understanding is Green is highside mofet and Yellow is lowside mosfet
    So 'the green must go low before the yellow goes high" mean There is should be dead time ( Non overlap time from My understanding ) after high side mosfet Turn OFF and Low side MOSFET Turn ON

    Considering that scenario Print 001 is satisfying where as Print 000a is not fulfilling the condition

    Am i right , if any change please high light
     
  18. circuit1234

    Thread Starter New Member

    Jan 7, 2015
    22
    0
    Hi
    the output load now i given is resistive load 10kohms , so at output not much current is flowing .
    from the waveform i think pink high lighted portion is inductor voltage . since i dont have differential prob i cant get the waveform for high and low at the same time . but when i prob with respect to source the gate voltage is normal as in Print _001a
     
  19. crutschow

    Expert

    Mar 14, 2008
    13,025
    3,236
    I see what you are saying. The top transistor (green trace) turns off when the voltage drops, but doesn't go to zero until the bottom transistor turns on.
    If that's true then there would seem to be no overlap or shoot-through problem.
     
  20. ronsoy2

    New Member

    Sep 25, 2013
    19
    4
    This looks more like an implementation problem than design. How close is the driver IC to your power FETs? The connection between the lower FET source terminal and the ground pin on the IC is particularly critical in using these drivers. Nanosecond pulses can easily be generated with interconnects a few inches long, and these pulses can exceed the maximum rated terminal voltages of the IC. Using a solid ground plane with the IC ground pin and FET source pin directly soldered is best. I would remove the 20 ohm resistors and replace them with 12 volt 1 watt zener diodes soldered right at the FET pins.
    Another passing comment is that if you absolutely do not need 100khz drop the frequency. If you can use 50khz you make everything half as critical! If you can use 20khz you may not even need heatsinks on the FETs! The physical size will usually not be enough larger to make any difference in most applications. Just a thought from an engineer with many years of experience.
     
    circuit1234 likes this.
Loading...