power nfet in ohmic region

Discussion in 'The Projects Forum' started by dlatch, Jun 10, 2017.

  1. dlatch

    Thread Starter Member

    May 15, 2016
    84
    0
    I'm using a power nfet to as a brake on my Hall wiper slot car controller. The gate voltage swing needed to give brake adjustment about 1/3 of a volt. I am having a hard time creating that small of a swing with a pot.

    Question: can you think of a better way to this? control the fet in ohmic region? is there an op amp circuit that could do it?

    Second question: I tried PWM at the gate and it works but is problematic: any higher current motor blows the nfet. I better flyback diode solved it on the bench but in real conditions the brake fet blew. Question: would more fets in parallel help?

    Diagram: the brake fet is show next to the motor and the gate control pot is as shown This linear set up is trouble free but all of the adjustment comes in 1/10 of a turn. I want to spread the control out over the whole pot. I want the fet to go from less than one ohm to about 10 ohms.
     
  2. Dyslexicbloke

    Distinguished Member

    Sep 4, 2010
    559
    34
    Just make sure the top and bottom of the pot is at the maximum and minimum voltage you require...
    That said its not a great plan

    PWM is the better option but you would need a break resistor above the nFET to limit the current. As is, if the FET is on, current is only limited by the RDSON value and the motor windings, result, smoke!
    You would also need a diode as you stated.

    The speed control FET/s do not have an addiquate gate drive, which will make them slow and therefore hot. if yoiu want to keep it simple, which will probably work here assuming your frequency isn't too high, consider using a 3904 & 3906 in a common emitter totempole arrangement to buffer the LM324 providing a lower impedance gate drive.

    The same strategy would work to produce a refrence / bias voltage for the FET if you decide to stay analogue.
    Build a simple amp with one of the spare 324 opamps. use negative feedback to make the control swing say 20X the output swing.
    Obviously you will need an offset. There are probably 20 ways to do this and get a stable refrence. A simple resistive adder with one variable input, buffered with a voltage follower is another simple option.
    Buffer the output as above and take the feedback from the buffered output where it feeds the gate.

    Proper dynamic breaking isn't easy. if you considder a mechanical break the energy disipated is a function of the breaking effort and the speed but the breaking effort is its self a function of the frictional coefficient and the torque, which varies with speed... its enough to make your head hurt.

    That said I guess you just want to slow down a bit and are not worried about slow speed holding in which case I think PWM is probably your best option, with or without feedback, but your analouge solution still has merit and with a few tweaks might actually perform better.

    Have you considdered using a big BJT rather than a FET. Given that your best performance will be achieved if you control current, irispective of voltage, which would produce a steady breaking torque, at least untill the voltage was insufficient to drive the current.
    Of course you could implement this using PWM, simmiler to your speed control, by sensing the voltage acros the break resistor and adjusting the duty cycle to keep it constant.

    Hope that helps,
    Al
     
  3. dlatch

    Thread Starter Member

    May 15, 2016
    84
    0
    Thank you for that well considered response. I am a novice so there is much in it to think about. The drive circuit seems to work very well. The frequency is about 500hz. So I want to concentrate on the brake circuit for now

    Help me understand what blows the Brake fet. You are saying, with very low R the motor would have enough back emf to kill a power nfet? (540) I understand "zero" ohms equals "infinite current" and anything would be stressed. However there is considerable resistance in the track wiring. And a linear circuit that puts 10 volts to the gate seems to be trouble free. So, I was thinking the issue has to do with PWM. Switching the fet against a high current brushed motor. The motors we use are radical and they propelling a 100 gram 1/24th scale car so the emf coming off them at coast is no doubt substantial.

    If switching the FET increases the spikes, would an additional fet in parallel be helpful.

    any and all comments on any aspect of the whole circuit are also welcome
     
  4. Dyslexicbloke

    Distinguished Member

    Sep 4, 2010
    559
    34
    It may well be voltage killing the FET and you can handle that with a big capacitor and even clamp it with a zenner diode.

    Think of it like this...
    Yoir motor is essentally a set of coils being turned on and off, by the comutator, whilst spinning a magnetic field.
    essentally it doesn't get much noiser. Oh wait you are adding PWM.....

    I would expect you to see significant voltage, and therefore current, spikes, skinny tracks or not, every time the motor is coasting and connected to a high impedance, your break FET when it is off.
    If you add a large capacitor to ground, round the FET, with a discharge resistor in paralell, which will waste some power when motoring, your motor will always be connected to a low impedance, provided its voltage is above the cap voltage, which is the only time you care.

    However this causes you another problem...
    The low impedance supply, that is the cap, relative to the break FET you are about to turn on, will kill it with current so you add a current limit resistor, high enough to keep the FET healthy but low enough to discharge the capacitor quickly...

    Motor to current limit resistor / cap / discharge resistor. Cap and discharge resistor grounded.
    FET to current limit resistor. FET grounded.

    When breaking the motor is trying to charge the cap, which significantly limits the voltage it can muster either as a generator or as a result of switching the inductive load.
    The FET now has a simple job, empty the capacitor at some rate defined by the width of the PWM pulses but limited by the resistor.
    The lower the capacitor voltage the more current will flow, at lower motor voltages, hence more breaking and in addition breaking will always 'soft start', to some degree, because the charged cap will take time to discharge.

    If you wanted to get really clever you could use the motor pulses as a trigger for a monostable that served up a pulse of prescribed width, controlled by your pot. If you did that you would have PWM but with variable frequency and configurable pulse width, thus applying more break at higher speeds for the same setting because the energy disipated would be directly proportional to break pulse width X motor speed as opposed to a fixed clock.

    Al
     
  5. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,768
    2,862
    Any idea on the motor current and/or winding resistance?
     
Loading...