DPRG List  

[DPRG] [Fwd: h-bridge discussion]

Subject: [DPRG] [Fwd: h-bridge discussion]
From: James Andrew Smith -- Onnimikki jasmith at ee.ualberta.ca
Date: Fri Jun 2 09:40:46 CDT 2000

There are a couple of good plans for H-bridges at

One of the problems with most H-bridges is trying to figure out how to
drive the high-side FETs (if you're using MOSFETs).  I've found five
possible solutions:

1. Use four N-channel FETs, two as high-side switches and two as low-side
switches.  Use a charge pump to drive the gates of the high-side switches
so that they can be driven at a voltage level above the main power rail.
(see http://www.cadvision.com/blanchas/hexfet/nch-brdg.htm) I tried
implementing one of these and blew up the charge pump.

2. Use four N-channel FETs, same as in 1.  Don't use a charge pump, just
tie in a 9-volt battery so that the gate can be driven at 9 volts above
the main power rail.  This is a simple hack and the 9volt battery will
last a couple of weeks.  (It doesn't really discharge any current, so it
will last a while)  I've used this type of design for three years and it's
pretty easy and reliable.  You just need to remember to change the 9 volt
batteries every once-in-a-while.

3. Use four N-channel FETs and drive them using a special full-bridge
driver chip like the LT1162 or the HIP4081A.  My attempt at a LT1162
circuit can be found at
ftp://larcs.ee.ualberta.ca/pub/arvp/h-bridge/.  The LT1162 circuit works
fine if all you're controlling is one motor.  I've had some serious
coupling issues when I've tried to drive more than one motor using two or
more of these circuits.  The pulse-width modulated signal should never be
DC with the 1162 as they have a built-in charge pump that can't recharge
properly if the PWM duty cycle is not less than 90%-95%.  

I've bread-boarded a HIP4081A circuit and it seems to work better than the
1162, but I haven't built more than one so I haven't seen coupling
problems with more than one motor.

4. Use one N-channel FET as a low-side switch and replace the high-side
FET with a double-throw relay.  Drive the relay with a BJT (2n2222) and it
will allow you to switch direction of the motor while the FET allows you
to use a PWM signal.  This is fine for a speed control of a motor where
direction changes happen infrequently (i.e. at least a second between
direction switches).  I've got four of these soldered together and they
seem to work just fine.  I'll be driving them under load today and I'll
see what happens.

5. Use P-channel FETs for the high side switches and N-channel FETs for
the low-side switches.  See
http://www.cadvision.com/blanchas/hexfet/npch-sw.htm.  A friend of mine
and I etched a version of this design last night and we'll be putting it
together in the next few days.  Initial bread-board tests worked out okay,
so the etched version should be fine.
This doesn't help with the IGBT design, but I thought that I'd put my
two-cents worth in. (-:


*	    James Andrew Smith		*  jasmith at ee.ualberta.ca	*
*  Autonomous Robotic Vehicle Project	*  www.ee.ualberta.ca/~jasmith	*
*          University of Alberta	*  jasmith at internaut.org	*
*					*  james.smith at ieee.org		*

More information about the DPRG mailing list