[sci.electronics] HELP! me cross an H-bridge

grege@gold.gvg.tek.com (Gregory Ebert) (02/27/91)

 2hneirony@kuhub.cc.ukans.edu writes:
>
>	I am attempting to rig up a MOSFET H-bridge to run a bidirectional DC
>motor from a 24 volt source;
> [...]

International Rectifier (213-772-2000) has devices which allow you to sense
current without a resistor in series with the load; the devices are
called 'HEXSense' mosfets. The data sheet I have on the IRC150 says it's
100V @ 30A, but is a preliminary data sheet as of 1987.

The diode you mention is to provide a return path for inductive loads. You
dont need them in HEXFETs because they are an inherent device. In fact, almost
all MOSFET and bipolar power devices have these 'anti-parallel' diodes in
place because they are imperative. The zener across the gate-source
is to limit Vgs, but is totally unnecessary with good design techniques.
A low-impedance gate driver protects you from high dV/dt at the drain.

I assume your H bridge is something like this:

V+ o-------------------*---------*
		       |         |
                    MOSFET 1   MOSFET 3
		       |         |
		       *--motor--*
		       |         |
                     MOSFET 4  MOSFET 2
		       |         |
GND 0------------------*---------*


MOSFETs 1 & 2 are both on, or 3 & 4 are both on, or all are off. Oy Veh ! DONT
turn on 1 & 4 or 2 & 3 at the same time !

The gate drive for MOSFETs 2 & 4 is trivial: just give them about 8-10 volts
and they're turned-on rock solid. For MOSFETs 1 & 3, use separate pulse
transformers. I've even used the 1:1 audio transformers from Radio Snack
with perfect results. The only trick is that you need to drive the
transformers with pulsating DC (10Khz is great). I put a full-wave bridge
rectifier at the gate-drive end, with a parallel RC network whose RC
time constant is about 5 times greater than the period of your pulsed-DC.
Adjust the duty cycle so that you just start to get little 'dips' in the
gate-source voltage; this assures there is no DC component in the 1:1
transformer. DC and transformers don't mix.

===> make sure your 'dead time' is greater than 5*R*C <===

(dead time is the time interval between MOSFETs 1&2 truning off, and
MOSFETs 3&4 turning on, etc)

Finally, make SURE you minimize inductance in the source lead. You can
generate enough Ldi/dt to zap the device with excessive Vgs. Always use
a separate wire from the HEXFET device to the ground of the gate-driver.

                                                                 #####
                                                               ##  |  ##
                                                              #    |    #
							      #   /|\   #
                                                               #/  |  \#
                                                                #######