Help! My MOSFET H Bridge is burning!

Discussion in 'General Electronics Chat' started by Barnaby Walters, Jul 19, 2011.

  1. Barnaby Walters

    Thread Starter Member

    Mar 2, 2011
    103
    4
    Hi there,

    I am having a problem with my SMD MOSFET H Bridge. One of the N Ch MOSFETs burns up when no signal is being applied to the driver (a 4427, also surface mount), i.e. the inputs of the 4427 are pulled down.

    The board can be seen below:
    [​IMG]

    The FET with the arrow pointing to it is the faulty device. I have labelled the others, the power rails and drawn in the traces under the 4427 for clarity.

    Does anyone know why this might be happening? The MOSFETs are rated at 4A each, and this also happened with no motor connected.

    I'll try to dig out the datasheets for the MOSFETs… Can't tell the part number from the device, but I'll have written it down somewhere…

    EDIT: Forgot to mention, the circuit is running off 7.4v unregulated from a LiPo battery. The logic signals coming in are 5v regulated.

    Thanks,
    Barnaby
     
    Last edited: Jul 19, 2011
  2. Barnaby Walters

    Thread Starter Member

    Mar 2, 2011
    103
    4
    Found datasheets:
    P Channel: SI2303DS
    N Channel: AO3402

    Looking through them, everything looks to be within the right ratings… Hmmm…

    Cheers,
    Barnaby
     
  3. debjit625

    Well-Known Member

    Apr 17, 2010
    790
    186
    Post the schematic,and are you sure that the connection is all well as their is a lots of solder...
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Where is your schematic?
    Who made your 4427 driver? IRF? Microchip? Someone else?
    Microchip's datasheet specifically cautions against leaving unused inputs floating, which also means inputs that are not connected to some form of logic input.
    If you don't have anything else connected, then connect them to ground.
     
  5. Barnaby Walters

    Thread Starter Member

    Mar 2, 2011
    103
    4
    Here's the schematic:

    [​IMG]

    (Sorry about rough drawing, I always have to think too hard about how to draw MOSFETs. They don't exactly have the most simple symbol.)

    All of the connections are good (I think), I've tested them all with a meter, made sure signals are getting where they need to be and not going anywhere else.
     
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    If you're planning on using this circuit for PWM, you really should consider adding diodes to clamp the motor terminals to +V and gnd. it'll take four diodes, and they'll need to be fast recovery type or Schottky.
     
  7. Barnaby Walters

    Thread Starter Member

    Mar 2, 2011
    103
    4
    It's a Maxim chip, and I've pulled the inputs down to GND with resistors (one of them is on the other side and can't be seen in the photo).

    I checked with a meter, all of the signals are going to the right places. I've spent about three hours debugging this thing (and the exact same circuit again on the other side, also not working) and can't figure out why it isn't working.

    EDIT:
    Thanks for the advice — this circuit won't be used for PWM, but I'll be making another soon that will be. When I do I'll post a plan schematic to make sure I've got them in the right place :)

    Cheers,
    Barnaby
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Well, you may have overheated ether the 4427 or the MOSFETs when you were soldering them in. Wouldn't be hard to do that.
     
  9. Sparky49

    Active Member

    Jul 16, 2011
    834
    417
    Are you sure that there are no shorts?
    Are you using Surface Mount for a particular reason?

    Other than the things Sgt. Wookie has suggested, I recommend you spend a moment seriously checking the circuit over with a multimeter. Wiggle the probes around a bit, make sure that what you're touching isn't covered in flux or dirt and test every bit which looks suspect - and more.:)

    I'd advise that you consider soldering conventional pieces (thier name evades me at the moment lol), rather than SM, as it clearly hasn't helped you produce a working circuit. The advantage of the SM, in this case, has been minimalised. Try making some practice boards (pad, after pad, after pad, all lined up) and just solder. Also, try to use as little solder as possible - but make sure that you use enough. In this case you seem to have covered the whole board is solder.:)

    If you have to still use SM due to factors such as lack of suitable components, try using a tiny blob of BluTac to hold each component in place. I mean really tiny. Just the bear minimum to hold a component in place. This'll mean that you don't have to worry about the component sliding around all over the place, meaning the iron is touching (and heating) the component for less time.

    I'd really recommend not using SM though, unless you have to.

    Keep up the good work, and I'm sure you'll get it working.:)
     
  10. Barnaby Walters

    Thread Starter Member

    Mar 2, 2011
    103
    4
    Hi there,
    In this case I was trying to squash the circuit into a small space – as it turned out, not a good idea ;)

    Well, I've just spent three hours doing just that, but I'll have another crack at it tomorrow and see if I've missed anything.

    Through Hole? :) Yes, I'm not brilliant at soldering SMD components! I made up two smaller circuits today that worked perfectly first time, and there's far less errant solder, so I'm getting better.

    I actually enjoy using SMD components — in every other circuit I've made with them they've saved drilling, complicated layout and space. But ultimately I have found both types have their uses — for example, TH resistors are huge and unwieldy compared to SMD ones, but can be used as a jumper and also as a via, so I find myself using both types on the same board. Clearly this kind of power application isn't suited to SMD at my level yet, so I'll keep using larger TH components for motor drivers, I think.

    Cheers :) I usually get there in the end. With this circuit I may end up cutting off the motor driver portion of the board, salvaging the components that work and making a separate driver board with TH components.

    Barnaby
     
  11. Sparky49

    Active Member

    Jul 16, 2011
    834
    417
    Good stuff, remember that you can never check enough!:p

    I think if I spent three hours with a multimeter I'd just snap the darned thing in half, then stamp on it, then shoot it, then burn it, then put the remnents in a lead box, then melt it all down and throw the resultant mess down a volcano - at least you've got the patience hahaha.

    Have you considered making it out of through the hole components? It should be easier to test and manufacture due to its size - that way you should be able to see if the circuit is at fault much easier.
     
  12. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    From what I understand, it happened more than once ? Is that right?

    Your Mosfets already have diodes, so even if used for PWM, you should be good.
    Maybe meaningless, but what are the pull-down resistor values at the input?
    As stated before, either you have a short circuit somewhere, you damaged the component when soldering or there was cross conduction due to some non matching time delays.

    You said it also happened without the motor, that most likely means, there was cross conduction on the damaged channel.

    1. You applied a high signal to one input, the n-channel FET was conducting / p-FET was open / everything was fine?
    2. then you applied a low signal to the input, the n-FET burned immediately?
    3. If that's the case I'd think that while the p-FET had already turned on the n-FET was STILL conducting

    But, I had a look at the datasheets, and the overlap could in worst case only be a few ns, and I'm not sure if this is enough to burn your component (I doubt it somehow)

    I'd suggest, you replace the burned part and do some measurements (oscilloscope) without motor and with low motor voltage. put a resistor in series with the motor power supply and you should be able to see if there is cross conduction measuring the voltage over the resistor.
     
  13. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488

    LOLOLOL.. I liked the volcano part... seems so much easier than trying to repair that thing :D
     
  14. Sparky49

    Active Member

    Jul 16, 2011
    834
    417
    Heck, I've nearly done it to a few of my circuits in the past. Sometimes it's just easier, quicker and cheaper to just start again lol.

    Which is exactly what I suggest doing here.
     
  15. Barnaby Walters

    Thread Starter Member

    Mar 2, 2011
    103
    4
    Unfortunately there are no volcanoes near, or I would try it out… Ah well.

    Yep, I'm going to saw off the motor driver board, finish the rest of the circuit and attach a header so I can connect another driver board later.

    Thanks,
    Barnaby
     
  16. debe

    Well-Known Member

    Sep 21, 2010
    947
    184
    Just for a start the H bridge drives are wrong in the circuit, they will short the power rail. The drives should be feed ing diagonaly oposite. This is why its burning up a FET.
     
  17. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    How exactly will it short the power rail? Could you explain in detail?
     
  18. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    debe,
    This bridge has P-ch on the high side and N-ch on the low side.
    The 4427 driver has Schmitt-trigger inputs to ensure that the outputs OUTA and OUTB will rise/fall quickly.
    When one of the OUTx's is low, the connected N-ch is off, and the P-ch is on.
    When it goes high, the P-ch turns off and the N-ch turns on.
    There will likely be some momentary shoot-through during the change of states.
    It wouldn't hurt to put a resistor in series with the supply during testing; say around 5 Ohms.
    [eta]
    After more thought, the lack of bypass caps across the 4427 may have caused this failure. The P-ch MOSFETS have around 10-12nC total gate charge, and the N-ch have 5.8nC-10nC total. So, anywhere between ~15.8nC and 22nC to charge/discharge from a driver that has no bypass caps.
     
    Last edited: Jul 20, 2011
Loading...