Connect with us

H-Bridge MOSFET blowing up

Discussion in 'General Electronics Discussion' started by positron, Jan 23, 2012.

Scroll to continue with content
  1. positron

    positron

    9
    0
    Jan 23, 2012
    Hi,

    I am using a MOSFET H-Bridge IC (ZXMHC3F381N) driven by a MAX4427 MOSFET driver IC to control a motor.

    Actually I've got two pairs of those ICs to control 2 motors using the same signals from a microcontroller and, for some reason, the ZXMHC3F381N blows up after a while. VDD is 12V and the current consumption of each motor at stall is 1.2A.

    The first output of the MAX4427 is connected to the gates of the two mosfets (P&N) of one side of the H-Bridge and the second output is connected to the gates of the other side. I fix one of the MAX4427 inputs to 0V while making PWM (10KHz) at different duty cycles to the other. If motors run freely (no load), the circuit works just fine but after changing speeds and applying some resistance to the wheels, it will blow up.

    Does anyone see any design flaws? I didn't add extra flyback diodes but I expected the ones of the Hbridge to work for me.

    Many thanks and any help is very much appreciated.
     
  2. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,411
    2,779
    Jan 21, 2010
    Do you have diodes across the mosfets to protect them from spikes? (I think the answer is no) The body diodes are far too slow to perform this function. Without them your mosfets will die pretty quickly (and as you note, as soon as you apply a load to the motor)

    Back to back zener diodes across the motor may also be effective (less spikes on your power supply rail, but you need to consider the power rating.
     
  3. positron

    positron

    9
    0
    Jan 23, 2012
    Thanks steve for your speedy response!

    This circuit is working just great with no body diodes if I drive two motors with 2 different signals for each motor. This was the reason why I didn't use diodes across the mosfets.

    However if I drive two motors with the same 2 signals for both motors the MOSFETs die after a while.

    I was wondering whether the reason of the failure was a shoot-through due to possible short-circuits between the two mosfets of one side of the bridge or, on the other hand, to spikes due to reverse currents flowing from the inductor back through the mosfets.

    The body diodes of the ZXMHC3F381N are rated to 12ns reverse recovery time at 2.1A. Aren't those fast enough for my case? I have no clue about how fast my diodes should be.

    Many thanks once again
     
  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,411
    2,779
    Jan 21, 2010
    Some mosfets have built in diodes (in addition to the parasitic body diode). I'm not sure if these do, but the specs for the diode seem better than I would expect.

    Can you show me the schematic of the 2 versions?

    If the working one has some dead time and the non-working one does not, or if the second one has a lower switching speed then shoot through could be an issue.

    Is there any difference in operating temperature between the two options?

    How do the mosfets die? Are they shorted D-S, or something else?
     
  5. positron

    positron

    9
    0
    Jan 23, 2012
    I don't have the schematic handy right now but I'll try to upload it later.

    Anyways, the board is the same for two version but the logic signals for the wrong case are connected together for both motors (ie, motors run at the same speed). In the good version, I drive the two motors from different control signals (ie, motors can run at different speed)

    This is the datasheet of the mosfet IC and I can't really see whether the diodes would work as flyback: http://www.diodes.com/datasheets/ZXMHC3F381N8.pdf

    The operating temperature and motors are exactly the same for both cases.
    I have blown up some mosfet ICs, in some of them, D-S are shorted and in some others, every pin is shorted with the others (D-S-G of the 4 mosfets are shorted).

    Many thanks!
     
    Last edited: Jan 23, 2012
  6. positron

    positron

    9
    0
    Jan 23, 2012
    Here is the schematic:

    [​IMG]

    Regards
     
  7. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,411
    2,779
    Jan 21, 2010
    You would connect them across the drain and source of the mosfets, essentially the same as the body diode.

    Here are a couple of things I would try:

    1) place small value resistors in series with the gates rather than tying them together directly. This will help ensure that any voltage spike which appears on the gate as a result of the device turning off quickly is not transferred to the other device.

    2) consider external flyback diodes.

    3) consider a snubber network across the motor

    4) investigate adding some dead time to your design

    edit: 5) retain the pair of MAX4427's and link the inputs of these rather than the gates for the pair of drivers.
     
    Last edited: Jan 23, 2012
  8. positron

    positron

    9
    0
    Jan 23, 2012
    Hi

    Thanks steve once again for your help.

    Won't these resistors introduce a delay in the transition?

    I'm not sure if I understood this well. The two inputs of the 4427s are used to drive 4 gates so if i'm getting you right, I would need to use two 4427s to drive the 4 gates tying their inputs together.


    Also, I'm not using the mosfets IC in the schematic but the ZXMHC3F381N and their turn off/on and rise/fall times are the following:

    N channel: Ton=2.5ns Trise=3.3ns Toff=11.5ns Tfall=6.3ns
    P channel: Ton=1.9ns Trise=3.0ns Toff=30.0ns Tfall=21ns

    Could this lead to a real shoot-through problem? I mean, are those few nanoseconds enough to blow the mosfets when driven from the mosfet drivers?


    And last thing, the forward voltage and recovery time of the internal diodes, aren't enough fo protect them from spikes due to inductor current flowing back to the battery? I'm still not sure whether the problem is shoot-through or returning current.

    Anyways I will try some of your suggestions and let you know!

    Thanks!
     
  9. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,411
    2,779
    Jan 21, 2010
    yes, but they will also isolate the gates so each mosfet can turn on and off as fast as possible. When you parallel the gates you are placing their gate capacitance values in parallel. This means that the gate voltages all move as one together. It can result in one gate holding another one up (or down) a little longer. Placing a series resistance (maybe only a couple of ohms) will allow each gate to discharge at its own rate.

    Here is a quote from an IR application note which discusses the problems with paralleling mosfets (different problems, but similar causes):

    Well, you say that using a single 4427 to drive 2 gates was fine, but when you use it to drive 4 of them it's a problem. I'm just suggesting that you retain the pair of them so that you're not connecting more gates to the one driver. Having one driver per mosfet will make your switching faster, just as having 4 per driver will slow it down.

    It's pretty clear that the P channel mosfets turn off far slower than the N channel ones turn on. this will lead to a short time when both are somewhere between fully and partially turned on. Yep, shoot-through.

    This *may* be survivable when the switching is faster, but less so when it is slower. And that may be the reason they're dying when they are (i.e. when you combine them but not before). The inductance in your wiring may be sufficient to limit the rise in current.

    You'll note my suggestion for some dead time. A driver designed for H Bridges will often have some dead time circuitry built in.

    If you check out this device, the timing diagrams on the bottom of page 3 stow dead time (T2 and T4) when operating in PWM mode.

    Normally I'd say no. However there are some mosfets produces with high speed diodes connected drain to source. The datasheet for this device doesn't say either way, so it's hard to tell. They don't show ant sample circuits so you also can't see if they recommend them that way. Try to find an application note perhaps? I might play it safe and add them anyway.[/QUOTE]
     
  10. positron

    positron

    9
    0
    Jan 23, 2012
    Oh okay, I should have explained it better. Sorry.
    The sketch I posted, was showing a whole PCB for driving two motors.

    When I used that board to drive two motors using 4 different signals: L_PWM, L_DIR, R_PWM and R_DIR, it worked just fine.

    However, when I drove the two motors with just 2 signals (ie, connecting L_PWM and R_PWM together and connecting L_DIR and R_DIR together), the thing started to blow after I started varying the duty cycle on the PWM lines.

    Sorry again for the wrong explanation
     
  11. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,411
    2,779
    Jan 21, 2010
    The power of explanation! :)

    With separate driving, did you impose a load on the motors?

    Are your drivers well decoupled from the power supply? It could be that you are seeing dips in your power rail, ground bounce, etc. that could affect either the logic levels or the amount of drive available. (are PS and Vcc different? Describe your grounding)

    This may have been less of an issue when the motors were not in sync.
     
  12. positron

    positron

    9
    0
    Jan 23, 2012
    Yes, actually they are on a robot and the mass of the robot is the same for both cases so they are supposed to be equally stressed.

    PS is the output of a DCDC converter and it's 12V. VCC is the output of a 3.3V LDO .
    I have separate ground planes and they are connected using just one track.
    I use 10uF aluminium capacitors and 100nF ceramic capacitors as close as possible to the MAX4427 and HBridge power supply.

    Actually I can't really see the difference in the two cases but, I thought that reverse currents were likely to be the reason of dying mosfets because they are higher when in sync (both motors are stopping and varying speeds at the same time). Does it make sense to you?

    Thanks!
     
  13. positron

    positron

    9
    0
    Jan 23, 2012
    I can confirm that flyback diodes didn't help :(
    I used 4 1N5819 for each HBridge but MOSFETs are still dying after some seconds
     
  14. positron

    positron

    9
    0
    Jan 23, 2012
    I can also confirm that flyback diodes + 22ohms resistors in the gates of the MOSFETs didn't work either.
    Power supply was decoupled using 2x 4.7uF/50V and 100nF ceramic capacitors.

    :(
     
  15. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,411
    2,779
    Jan 21, 2010
    Well, I guess that shoot-through is the most likely problem.

    Have you considered how you will introduce dead time?

    Essentially you need to delay the turn on of each mosfet but turn them off as quickly as possible.

    It means that you need to use a different driver for each gate. If the drivers have schmitt trigger inputs, you could use an RC delay with a diode across the resistor to delay only one transition.

    ...Or get an H Bridge controller that has inbuilt dead time.

    Some circuits I've seen use gate resistors, but have a diode across them so that the resistor is effectively out of circuit when the device is being turned off. Whilst this doesn't equate to dead time, it probably can't hurt.

    Since you're driving an inductive load, the energy lost during turn on of the mosfets will be lower than during turn off, so slowing the turn on will not be too bad in terms of increasing dissipation and will contribute to reducing shoot through.

    If you can place a low value resistor in series with one of the mosfets and monitor the voltage across it with a scope, you may be able to see the shoot through as a voltage spike. If you can see it (and thus confirm it's an issue) you has killed 1.5 birds -- the first being the confirmation of the problem, and the second being the ability to observe when you've reduced/eliminated it.
     
  16. GonzoEngineer

    GonzoEngineer

    321
    2
    Dec 2, 2011
    Wait a minute here.......if your driver is blowing up, then you are driving it wrong., plain and simple.

    Can you revise the schematic to show EXACTLY which gate drive is attached to which drive signal?

    You show four gate signals, labeled 1-4, but in actuality you have 2 P-channel devices and 2 N channel devices in thebridge.

    (All of your gate drive signals are referenced to ground.):confused:
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-