N-MOSFET half bridge driver - is this a good design?

Discussion in 'The Projects Forum' started by RichardThiessen, Nov 19, 2012.

  1. RichardThiessen

    Thread Starter New Member

    Nov 18, 2012
    I'm just getting started in electronics. I'm designing a bipolar stepper motor driver. There are two H-bridges in this driver one for each of the motor coils. These H bridges are controlled by a microcontroller based on sensed current to bring the current in each coil to the desired level.

    I want the H bridges to handle >3 amps at 24 volt. I considered using an integrated h-bridge driver but the low cost drivers seem to be made for <20V supply voltages and 24 volts brings the NCP5359's bootstrap pin to ~34volts, 1 volt under the absolute maximum. I don't like that.

    So I decided to design a driver from discrete components. Here it is:

    Mosfets- NTD4963N (Vds max>30V /Qgate <16nc @ Vgs=10V /Vgs max >20V)
    PNP bjts- BC807 (Vec max > 45V)
    NPN bjts- MMBT4401 (Vce max > 40V)
    diode-SBAV99WT1G (Vr max >70V / Trecovery <6ns)

    These were the cheapest parts at mouser that look like they'll do the job
    The fets require 16nC of gate charge which will take roughly 100ns for the buffer to supply.
    Their threshold voltage can be pretty low (min 1.45v) but that still gives 0.7v above 0.7 diode drop limit of the buffer to discharge, worst case. That could be a problem.

    I've already found one problem in the design:
    -if the driver is told to turn both fets off (input=high impedance) and the output is pulled high by whatever it is attached to it, current from the output is drained through the upper push pull buffer and into the attached transistor. Which will dissipate enough power to fry itself (>350mW).
    -the boot capacitor drains under the same circumstances but I care less about that since it can be recharged.

    I won't have to worry about those because of how I'm using it (one fet on at all times) but I'm new at this so there could be other problems I'm not aware of.
    Last edited: Nov 19, 2012
  2. JDT

    Well-Known Member

    Feb 12, 2009
    Interesting, but I would not try to build the half bridge driver from discrete components. You have already found one problem. You may have problems with cross-conduction and switching speed.

    This is all taken care of with an integrated circuit like this:-

  3. RichardThiessen

    Thread Starter New Member

    Nov 18, 2012
    I suppose that I should clarify. I was hoping to get the total cost down to under 10 dollars which includes a microcontroller, current sensing and the power output stages. Similar products based on stepper motor driver chips like the A4983 cost roughly that. Given my budget, I was hoping to spend under 50 cents per half bridge driver. The mosfets themselves come out to $1.60 for all eight. Four dollars in integrated gate drivers is a little expensive.

    The cheap drivers (price>=$.25) all seem to be made for <20 volt power supplies. Most can handle high side driver voltages of 25-35 volts relative to ground but I need 34V at the bootstrap pin before transients. The more expensive ones (price>=$1) handle hundreds of volts, they're meant for handling mains voltages. There seems to be very little in between. and I've found nothing in my price range capable of doing what I need comfortably. If there is I'd love to hear about it.

    I vaguely remember reading about level shifting in half bridge mosfet drivers and the "problem" I talked about is something people have already run into. Current based level shifting can create a lot of heat when done over a large voltage gradient. There are ways of mitigating this (pulse signaling+latch) but I don't need to worry about it. It is inelegant but not of practical significance for the voltages and currents used in my design.

    The switching speed is what I'd like to ask about. How fast will it switch? I can put in as much dead time as needed to eliminate cross conduction but an initial estimate is a good thing to have and I have very little experience with bjt turn on/off characteristics.
  4. liquidMikado

    New Member

    Jan 6, 2013
    2 Suggestions:

    Go for a NCP5359 Solution and drive it with 22V
    Use a Mikrocontroller that offers Syncronized PWMs with Hardware Deadtime (ie locked AntiPhase) for your discrete solution. Cortex M0 from STM or NXP can be had for sub <2$ even in single qttys

    I hope this helps.
  5. shortbus

    AAC Fanatic!

    Sep 30, 2009
    Maybe I'm missing some thing here (constantly do), but where are you finding mosfets that take that much gate voltage?

    The gate drivers for only need a ~10V supply. The gate voltage is referenced from the 'floating' bootstrap capacitor, not from ground,and only for the high side mosfet.