100V Switching using power mosfet

Discussion in 'The Projects Forum' started by arvinik, Apr 21, 2008.

  1. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    Hello dear all,

    In the attached file, you see a 100 Volts pulse train with 25% duty cycle and period of 800 ns. I need to generate such pulse train using IRF640 power mosfet and a 100V DC power supply. This pulse-train is fed into a rather long coaxial cable with 50 ohms surge impedance.

    The problem is the correct configuration of the mosfet circuit!! :confused: Is it possible to switch 100 Volts from drain to source, with 5 or 10 volts gate signal?? Do I need additional elements or IC's...?! note that I've made the control signal of the gate through a high frequency counter (74HC4017) with output amplitude of maximum 10 volts.

    I'll be happy if you help me on this...

    Many thanks in advance,
    Yours,
    Arvin
     
  2. scubasteve_911

    Senior Member

    Dec 27, 2007
    1,202
    1
    Hi,

    The maximum gate threshold is 4V for the IRF640, which means you can use 5V to turn it on. You should be using 10V though. The problem will be your very low output current of +/-25mA, which will be extremely difficult to raise your gate voltage up fast enough to turn the transistor on. I recommend a push-pull pair or a dedicated IC driver to charge the gate.

    Some calcs:

    72nC max for 10Vgs
    transition ~20nS (10% of your pulse width, should go a bit shorter)
    Igate=(72E-9)/(20E-9)
    = 3.6A !!

    Much higher than your 25mA can possibly source..
    72E-9/25E-3=0.00000288 = 2.88uS

    Steve
     
  3. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    Yes steve you are quite right. I had not paid enough attention to that point. about the push-pull pair, would you tell me in more detail how I should use it with the counter??
    The reason I've used the counter to generate the contro signal, is that I need 4 identical pulse-trains, with a 200 ns phase shift between each of them, in order to switch 4 power mosfet switches (see the attachment named 'pulses.jpg'). (I've used 4 pins of the IC outputs and the 5th is attached to the reset pin). is there any IC driver which can provie such pulses???!



    By the way, is the attached schematic (mosfet.jpg) a correct switching or chopper configuration?? (if we assume that the gate current is high enough).

    Yours,
    Arvin
     
  4. Caveman

    Active Member

    Apr 15, 2008
    471
    0
    You shold low-side switch it instead. You should
    1. Disconnect the 100V positive side from the FET
    2. Connect one side of the load, RL to the 100V positive side.
    3. Connect the other side of the load to the drain of the FET.
    4. Connect the source of the FET, the negative of the function generator, and the negative of the 100V supply all to ground.

    This is the circuit you now need (barring gate current issues).
     
  5. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    thank you caveman, but in low-side switching with the same control signal from the function generator, the outputs are the inverted of the desired outputs (attachment of the 1st post). we'll have pulse train with 200ns off and 600ns ON! is it possible to use an inverter IC after the counter, in order to achieve the desired outputs from the MOSFET??!
     
  6. Caveman

    Active Member

    Apr 15, 2008
    471
    0
    You could use an inverter IC, but you will still have the same drive problem. Or you could just change your function generator signal.
     
  7. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    What about using the same function generator (74HC4017 and the inverter) along with a mosfet gate driver IC? does the output provide the needed gate current??
     
  8. scubasteve_911

    Senior Member

    Dec 27, 2007
    1,202
    1
    Sorry, I am writing finals right now and haven't been checking.. I recommend just using a driver IC, unless it is meant for production. You'll save yourself a headache and they work very well.

    You might also want to look into mosfets that match your specs, which may save you some gate charge. That way you can get away with a <2A driver.

    Steve
     
  9. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    Thank you steve, but there's something I still don't know! Does the driver IC generate the pulse-tarin itself, or it must be used along with my previous logic circuit just to provide the gate current??

    By the way, Do you think an IRF520 Mosfet, along with an IRF2130 driver, would work fine in both time/current range??
     
  10. scubasteve_911

    Senior Member

    Dec 27, 2007
    1,202
    1
    You still have to apply logic to it, it simply acts as a high current buffer. Sometimes they incorporate a charge pump for driving high-side fets. Anyways, if you meant the IR2130 driver, it doesn't supply enough current. The IRF520 is significantly lower gate charge, but not low enough. You didn't mention your slew rate requirements for the signal.

    If you need a really square waveform, you'll need a higher current driver. You're probably going to have a bit of ringing going on though..

    The same math applies..

    16nC/0.2A=80nS risetime which means if it were fed a 160nS squarewave, you'd have a triangular waveform and probably a ton of heat if you're driving any sort of load since you're in the linear region, rather than saturation or cutoff.

    Steve
     
  11. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    I tried low-side switching configuration, it works well. BUT I can't use it since I have no lumped element load! My load is a transmission-line (a rather long RG58A/U coax). So I have to use the high-side switching method in order to turn the input voltage/current of the coax OFF, while the control signal is low.
     
  12. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    Well the shape of the signal does not need to be "completely" square, and it shouldn't be triangular! :)

    by the way, I've used NCP4425 driver. it provides 3A at the output. The results were not bad (high-side switching). The problem is now rather long rise/fall times of the signal at the end of the coaxial line.

    The schematics of the circuit is shown in the attached file.

    I'm waiting for your comments,
    Yours,
    Arvin
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
  14. arvinik

    Thread Starter Member

    Apr 21, 2008
    11
    0
    By the way, as I mentioned before, I need 4 stages of switching using power mosfets, with 200ns time-delay between each stage.
    Now what I need is 4 optocouplers for each stage, to separate the control signal of the stages.
    Can anybody introduce an optocoupler (or any other device for such purpose!) which can work in this rather tight timing (200 ns pulses, with period of 800 ns) ?? :confused: I searched for it but I couldn't find a proper one. (how about H11N2-M?)

    Thanks in advance,
    Arvin
     
  15. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    The source impedance from the Mosfet must match the 50 ohms impedance of the cable and of the load to avoid reflections which cause cancellation of some frequencies.
    Then the Mosfet needs a 200V p-p output signal.
     
Loading...