Re: [CAD_CAM_EDM_DRO] Re: H bridge progress

Chris Stratton wrote:

> First, I want to thank everyone who has offered advice on the H bridge
> circuit. What I have at the moment is the FET's and 2104's on a
> little milled PCB, with the L297 on a breadboard. Next revision is
> going to put it all on the PCB, but I want to be more sure of the
> design first.
>
> The gate resistors in question are shown in the IR2104 example circuit
> between the 2104 and the gates.
>
> > > 1. The gate resistor is there to prevent oscillations. Keeping the
> > > driving circuit VERY close to the FET helps to reduce this, as do
> > > ferrite beads on the gate line placed very close the FET.
>
> It sounds like maybe I should go for a 10 ohm or so resistor, possibly
> with a ferrite bead on one of its leads? Another idea, from the ARRL
> handbook, would be to use a larger value resistor as a physical
> carrier over which to wind a small coil of wire (shunting the
> resistance). The main issue I need to decide now though is if I plan
> for a component like this on the PCB, which does increase the distance
> between the 2104 and FET's, but also provides a convenient way to
> cross over a power trace (at right angles, so coupling should be
> minimal).

A 10 Ohm resistor should be fine. They not only prevent oscillations
(not likely with square wave drive) but they also slow the rise and fall
times of the fets so that the dv/dt is not so fast that it produces damaging
spikes. I am using 13 Ohms with IRF640s, and get rise and fall times
in the range of 80 V / 50 nS! that's 1.6 giga volts/second! (Or, in that
wierd EE unit, 1600 Volts/microsecond!) Still, a vary fast rise time,
and you don't want it to be any faster than that.

I have R/C selected to give about 500 nS minimum offtime at the input
to the FET drivers.

>
> > > 3. The L297 turns both sides of the bridge high when turning off a
> > > winding as this is the fastest way to get the energy out of the winding.
> >
> > No, that actually allows current to recirculate (through the reverse diode
> > on one of the transistors) indefinitely, reduced mostly by the IR drop of
> > the coil itself.
>
> Fortunately, switching modes on the 297 is a matter of chaning one
> jumper, so I can try both ways.

Recirculating the current reduces the amount of copper and iron losses
in the motor, which can become significant if the (voltage) polarity of the
winding is reversed every time the current reaches the limit. A gradual
current drop is fine, at least for plain drives. Microstepping drives may need
to reduce winding current without a current reversal, and therefore may
need a different switching approach.

Jon