MOSFET fail short - searching for cause

Discussion in 'Automotive Electronics' started by Marcel DaJanchar, Jan 1, 2019.

  1. shortbus

    AAC Fanatic!

    Sep 30, 2009
    6,068
    3,478
    And I still say the zener isn't a good enough gate protection. Be willing to bet it's the gates blowing out.
     
  2. MisterBill2

    Well-Known Member

    Jan 23, 2018
    2,365
    440
    How about a shunt capacitor, gate to source, and a 1K series resistor, to reduce the height of any spike, IN ADDITION to the zener protection diode. The RC protection will work for spikes of either polarity. And still, I suggest a higher rated FET, perhaps 200V and 20A.
     
  3. Marcel DaJanchar

    Thread Starter New Member

    Sep 9, 2015
    24
    0
    @shortbus It seems like the gate of the FET fails (FET stays shorted). Could I replace zener with TVS, or is there any other way to fully protect Vgs from exceeding its maximum ratings?

    @MisterBill2 Wouldn't shunt capacitor make the rise and fall time of the gate voltage greater? If not, what value would you suggest? I have trouble finding appropriate FET rated with 200V. Even with Vds of 200V, they still have Vgs rated at 25V max. And they are expensive (5€ per piece)
     
  4. MisterBill2

    Well-Known Member

    Jan 23, 2018
    2,365
    440
    Yes, certainly it would slow the turn on time, and that is a consideration indeed. Putting the filter on the input to the gate trigger supply will solve that, since then the capacitor will serve as a voltage source for the gate drive. And now I am wondering if it might be better to supply the gate driver circuit from the same power connection as the drain feed, instead of the acc switched connection. If the FET is biased off while the starting is going on that may be a better condition. When do the lights being controlled need to be switched on?
     
  5. Marcel DaJanchar

    Thread Starter New Member

    Sep 9, 2015
    24
    0
    @MisterBill2 The circuit adds a function to front blinker bulbs to act like DRLs (with adjustable brightness - provided with 555, PWM and FET outputs). When you use turn signal, the FETs are switched off and the selected blinker blinks normally. When the blinking stops, the PWM enables again after 1 second (so the both front blinkers are dimmed again).
     
  6. Sensacell

    Moderator

    Jun 19, 2012
    2,140
    681
    I would try adding fast clamping diodes to the "Output" terminals, anode to ground, cathode to the drain.
    The intrinsic body diode in the FETs already clamps the Output to Vcc, so you just need one to clamp negative transients.

    Since the lights are probably off at the other end of the vehicle, away from the power input, a big positive transient might make the "Output" terminal look like a big negative voltage, the diode will help mitigate that.
     
  7. MisterBill2

    Well-Known Member

    Jan 23, 2018
    2,365
    440
    Now that I understand the application, here is a suggestion, which is to change the sockets and the bulb type to use a dual filament bulb, which has been standard in automotive stop/tail lights for many years. Then you can add a voltage control for the low power sections and have the higher powered sections for the turn signals. AND, as an option, add a diode so that when the turn signal is on the DRL filaments are more brightly illuminated with the full 12 volts. AND, this will provide a redundant turn signal indication.
    And a separate question is if there is any real proof that DRLs actually make anything safer?? My guess is that it was a rule created totally based on the emotions of the unknowing folks, based on some skilled orator's bleatings.
     
  8. wayneh

    Expert

    Sep 9, 2010
    15,552
    5,760
    Studies showed that having your lights on made you safer even in daylight. But I'm not sure they ever proved it's safer if everyone has their lights on.
     
  9. MisterBill2

    Well-Known Member

    Jan 23, 2018
    2,365
    440
    Were those studies paid for by those folks selling the DRL technology? Just wondering. AND, has any study verified that, now that they are so very common? And has it been shown that it was the DRL that improved the safety, or could it have been the widespread ABS systems helping folks stop straighter? LOTS of safety things have appeared in the recent years.
     
  10. dl324

    AAC Fanatic!

    Mar 30, 2015
    7,963
    1,901
    I know I can see cars further away during daylight hours if they have their headlights on. It helps with being able to see when it's clear to pass.
     
  11. wayneh

    Expert

    Sep 9, 2010
    15,552
    5,760
  12. MisterBill2

    Well-Known Member

    Jan 23, 2018
    2,365
    440
    What is absolutely certain is that DRLs take energy to light them, and that energy comes from fuel burned. And fuel burned reduces miles per gallon, thus increasing the pollution for every mile driven. Nothing is free, especially power, in cars.
     
  13. MaxHeadRoom

    Expert

    Jul 18, 2013
    16,873
    5,024
    DRL's are mandatory in Canada, my particular GM model just drop the low beams in series in the DRL mode.
    I find them very handy in conjunction with the beam flash switch it indicate to other drivers to go ahead etc.
    Bus drivers appreciate it and usually acknowledge with a short burst of the 4way flashers.
    Facing sun while driving helps see oncoming traffic also.
    Max.
     
    shortbus likes this.
  14. Marcel DaJanchar

    Thread Starter New Member

    Sep 9, 2015
    24
    0
    I know about dual filament bulb, but I want to make this circuit to work, because I will need this kind of application a lot (not just for blinkers). I am aware about safety and visibility, but nevertheless, I want this circuit to work like it should. So please stay on topic :).

    The last circuit which has failed had two 12V TVS in series (one has 13.3V breaking voltage), because I didn't have other TVS available for testing. Tomorrow, I'll get SMAJ15CA TVS diode (with breaking voltage of 16.7V). Maybe then it will be OK.
     
  15. Marcel DaJanchar

    Thread Starter New Member

    Sep 9, 2015
    24
    0
    Hello everyone!

    I have an update regarding this project. I placed TVS diode (Vbr of 13.3V) across VCC and GND and the circuit still failed after some days. I have my last theory of what could go wrong. Is it possible that bulbs and wires produce induction spike at drain of the FET when its closed? Drain voltage potential would rise above source voltage and exceed its Vdss rating (-30V) pretty quick - at that point inner body diode would fail short (due to avalanche), source is pulled at the potential of drain (which is still high) and Vgs is then also exceeded (because gate is tied to GND that moment). Could my theory be correct?

    A lot of failed FETs had source and drain shorted. Few of them had gate shorted too.

    I have wired TVS across drain and source to clamp the voltage Vds before it reaches -30V. I am currently testing it in my car and it is still operational.
     
  16. MisterBill2

    Well-Known Member

    Jan 23, 2018
    2,365
    440
    An inductive spike due to wiring is very unlikely. And I am thinking that the TVS right across the mosfet because that is what you want to protect. One possibility is that somehow the PWM has changed and has the bulbs in a higher current state. Bulb resistance is a function of filament temperature and that might be what changed. It may be that something in the PWM circuit has changed. That is a possible cause that would be a challenge to spot. It may also be that you have a charging system fault that has boosted the voltage a bit. It is probably NOT a short circuit because that would lead to an open fault, not a shorted fault.
    It is also possible that the gate drive has become weak , leading to more heat because of not fully switching on.
     
Loading...