Problem with DC motor control

Discussion in 'General Electronics Chat' started by Henkie001, Jun 27, 2011.

  1. Henkie001

    Thread Starter New Member

    Jun 26, 2011
    Dear ppl,

    I'm trying to build a PWM DC motor control. It is more or less a copy of a Chinese motor control unit for just one motor. In our application we need 1 control for two motors.

    Found different schematics on inet and copied the power equipment from the print of the chinees model(that was broken).

    PWM looks ok (in my opinion) but the problem that occurs is that every time I'm testing, one of the fets brakes, flames. The motor wil run for a while and the fets don't get hot or even warm.

    When 1 brakes the others are not damaged.

    I'm not an expert and the application notes that I found will take me weeks to understand. I have measured the frequency on an other Chinese print, without opening it, on the motor side and that was about 12,5 Khz. In first I was afraid that my 13Khz was to high, switching that often with this power....But in the end this seems to be normal.

    On other parts of this forum I read that the frequency is put this high to prevent noise. Is this really the reason? What is a minimum frequency to drive a DC motor of 1000Watt?

    I have looked on the edge of the PWM and that looks pretty sharp. I suppose this should be to prevent a lot of power being dissipated.

    On the Chinese print I saw an small elco near the transistor? I Suppose this is meant to drive the FET directly in operation???

    On my first design I mounted the fets on the print against a heatsink. But this was difficult to build and after my fist burn I planned to make the PWM board different from the power unit.

    The connection is made by unshielded cables and as I read problems that can occur due to cap difference in the Fets I think this can be my problem.

    Others say that there is difference in operation of FET’s of the same type. Read the treat: what in beginning looks the same problem but what was the solution in the end??

    Someone who can give me some suggestions?


  2. praondevou

    AAC Fanatic!

    Jul 9, 2011
    While the motor is running, did you measure the voltage spikes from drain to source of your FETs?

    Are these motors running freely or are they attached to something? Can there be a counter emf (motor acting as a generator)?

    If a FET burns they often enter into short circuit, so no wonder that the others are not being damaged, almost all current goes through the one just burnt.

    MOSFETs normally don't need to be matched as the have a positive temperature coefficient.

    Is it always the same FET that burns?

  3. Henkie001

    Thread Starter New Member

    Jun 26, 2011
    Dear Praondevou,

    Last week read more theory of this issue. Read the published articles and
    getting to understand more and more. Because we have built a nice housing with external
    heat sink etc. I still want to try to get it running this way. But is this possible....

    Lessons learned for me where:
    - Make a compact PCB design to provide L and C influences
    - In the examples I found a Capacitor over the + and - of the totem pole transistor output
    - It is not only the turn on moment of the FET that can be of sync with the others but also the turn off moment can cause problems.

    I'm not equipped with stuff to make my own PCB's. Had in the past, so I have a bit of experience, just amateurish. So I make my controls on standard hole PCB's. Question I have here is, does everything working with FET's need to have an accuracy of
    0,01% or is it possible to make low budget controls the way I'm working.
    In the picture you can see the heat sink with the print. Because we already have the nice
    housing I now import the PWM signal on the small board on the heatsink. This way, I
    presume, external influences on the signals going from transistors to FET are not an issue. On the left 4x MOSFET and at the right 2 x schottky diodes.

    On our vehicle there are two motors that drive two wheels. 1 I have connected with an original control and 1 I have connected on the new control. The original control is power by three big battery's of 12V. My control is powered by two smaller batteries of 18V.
    The result of the new control, measured on the drain site of the FETS, is shown in the
    picture. And as you already suggest there is a very high spike lifting the voltage up to
    apprx. 90 Volts!! In the beginning I made a wrong connection with the schottky diodes, they
    were not working, and I saw this happening for a very long period. Now I have made the right
    connection and still a small peak is present.

    Where is it coming from?


  4. praondevou

    AAC Fanatic!

    Jul 9, 2011
    Well, if it's 90V then that's too much. Vdss of this transistor is 75V.
    Is there no capacitor from cathode of your diode to source of the FETs? I don't see any in your schematic. Try to put for example 220uF/ (>50V should be ok) as near as possible from cathode of your diode to source of your FETs.
    Last edited: Jul 16, 2011
  5. Henkie001

    Thread Starter New Member

    Jun 26, 2011
    Hi Praondevou,

    I placed a capacitor, just the first one I found, 2200 50V between the cathode and the source. There is a difference, the peak is reduced a lot, almost gone.

    One of the wheels has a brake and therefor more mechanical resistance. If I connect the control to this wheel it results in a better block wave form. Three pictures attached,

    1. Free running motor. The circle presents the generating effect of the motor
    2. Brake on the wheel but wheel still turning. Nice block waves
    3. Full brake standstill of the wheel. Peaks are forming. Not as high as they have been.

    But in the end I still see a peek and this is not what is happening with the original control. Can there be a other reason for this behavior?

    Furthermore is the influence of the generating effect of the motor less with the original control. Can be the same problem is the peak in voltage?

    Cause of the different power source these pictures are made with the battery pack power source so for both controls it is the same.


  6. SgtWookie


    Jul 17, 2007
    You might place a SMALL ceramic or poly metal film capacitor from your MOSFET drain to source terminals; somewhere in the vicinity of 200pF to 500pF, with VERY short leads.
    The diodes don't turn on and off instantly; they might be fast, but they're not instantaneous.
    The small capacitor gives the reverse-EMF diodes some time to turn on; it slows the rise time of the spike.
    You don't want the cap(s) to be very large at all, because when the MOSFET turns back on, it will discharge the capactor(s). If the cap(s) is/are too large, the MOSFET will have a very high current spike when it turns on.

    Keep in mind that inductance and fast turn-off times is what causes the spikes. Even your MOSFET's terminal leads have inductance; in the range of 4nH to 8nH. Other component leads will have inductance at about 15nH per 10mm. The shorter your component leads and circuit board traces are, the less inductance they will have.
  7. praondevou

    AAC Fanatic!

    Jul 9, 2011
    Ok, so the big cap helped. Should be good enough to avoid the FET blowing up. These caps normally have to have the lowest internal resistance possible, I don't think you paid attention to that?
    As SgtWookie suggested you can additionally diminuish the turn-on / turn-off slope of the FET by placing a SMALL cap between drain and source.

    If you don't have the exact same layout / components, there will of course be a difference in the output waveform.

    Additionally I suggest you read the two following documents I already posted in another thread.

    They give some useful info about layout considerations. Always keep in mind: high-frequency, high-current carrying wires / traces should be as short as possible.;)
  8. Henkie001

    Thread Starter New Member

    Jun 26, 2011
    Hi guys,

    I replaced the capacitor for one of 470u 100v. The one of 63V gave smoke signs when I try to run with a good load of resistance on the wheel. It gets warm now after a while but I don’t believe it will get to hot.

    Never had anything to do with internal resistors of capacitors so was indeed not aware of the fact that it could be a problem. Have googled a bit but did not found the classification of esr on elco’s and the ones I have on the shelf. If I want to measure it I need an ESR meter ……

    For now I think it will do ….

    Is the temperature a result of a possible big ESR? Or are there other causes?

    I tried the ceramic capacitor of 330pf on drain source but this had no influence on the total. When I block the wheel with the break to a complete standstill I saw a peak of over 100 amps. I’am not trying to break the stuff so I just test for some peak moments.

    On these peak moments I see Vds rise, just a really small spike, up to 70V. There is not much left with a rating of 75V Vds.

    So I don’t think it will break but I’m not happy thinking of what happens when the FET is burned up. As I have written earlier it the acts as a short circuit. That is not something you want to have sitting on the ECart.

    Are there more possibilities to get rid of the voltage peaks? Should I order a special Capacitor to solve this problem? Can it be caused by a different turn off of the FETS?
    I used standard components in the gate driver, like resistors that have 10% tolerance (measured).

    It is hard to measure in the circuit what turn on and off is of the fets in parallel. Is there a method to measure this?

    For now I’ll study the app notes you have send.


  9. praondevou

    AAC Fanatic!

    Jul 9, 2011
    You can:

    - use a FET with bigger Vds, IR has a lot of options in TO-247, it would even be possible to decrease the number of FETs, have a look at
    - use snubbers - voltage clamps or simple RC snubbers, I'd prefer a simple RCD clamp
    - try to increase the gate resistors, but they already have a quite big resistance.
    - buy Caps with low internal resistance, specially designed for SMPS, have a look at
    - in parallel with the caps you added you could put a MKP capacitor 470nF with high pulse strength, for example the B32686/B32656/B32654 series from epcos

    attached you find various snubber circuits
    Last edited: Jul 18, 2011