Re: [CAD_CAM_EDM_DRO] Re: L297/L298 And a "correction"

Mariss Freimanis wrote:

>Thanks for the comment. That circuit works much better than any half-
>bridge driver from IR or anyone else. Why?
>
>It does not permit the MOSFET intrinsic diode to conduct. When it
>does, the reverse-recovery time of said diode results in truly
>prodigious "shoot-thru" currents. These currents are simultaneous
>with full supply voltage imposed so the power content of the pulse is
>enourmous; 50A @ 80VDC is 4 kiloWatts. Thank goodness they last only
>100nS (400 uJ per transition). Still, at 20kHz they amount to 8W of
>switching losses.
>
>
>

Maybe I shouldn't give away my trade secrets, but they are pretty obvious,
anyway. That's why I use the IR2113S, with separate drive inputs for the
2 transistors. That allows me to use simple circuits at the input of
the driver
to control the dead time between the two transistors. There are some
other tricks
to handling the tremendous di/dt transients that can fool the driver
into turning
on both outputs at the same time, mostly in the form of PCB trace
layout. Because
I put output filters on my drives, I get some pretty large circulating
currents that
can cause the body diodes to start conducting during this dead time.
So, I have
to put ultra-fast diodes across the low-side transistor to keep the
voltage across
the FET from going negative. The FET body diode is a HORRIBLE diode, and I
have seen over -12 V across the transistor. This doesn't hurt the
transistor, but
it pops the IR driver chip. The fast diode clamps this negative
transient, protecting
the driver. This is in a 20 A DC brush servo drive, so the currents are
roughly 3 X
higher than in a Gecko stepper drive.

I wanted to be sure my drives could survive peak current into a dead
short with
a 108 V supply voltage.

Jon