High Speed Design ----> Knee in the falling edge?

[EdV]

I worked on this problem a couple of years ago and never did
understand what the deep down root of the problem was beyond the
vendor shrunk the die on an optocoupler we used to drive an IGBT and
we needed to slow down the propagation times.

The weird thing was the manifestaion of the problem:

A kink or knee in the middle of the falling edge would turn into a
"ring". Big time! The IGBT would then be driven in the linear region
for a a half second or so. After this happens umpty ump times the
IGBT would become an open.

So what is with the knee thing in the middle of the falling edge? I
have seen and have an intuitive feel for ringing at the bottom or top
of a square wave near the level transitions. But in the middle?

I was reminded of this by the falling edge (of all things) a PID
temperature controller. Going from 100C to -40C there it was a knee
right in the middle.

It has got to have something to do with some sort of level transition
for the temp controller case I would imagine SW. The opto coupler
though. . .

Sorry if I got rambling here.

Thanks,
Ed V.

PS - yeah, I know I should say the part numbers and edge rates but hey
it was two years ago. : )

 

[James Arthur]

I worked on this problem a couple of years ago and never did
understand what the deep down root of the problem was beyond the
vendor shrunk the die on an optocoupler we used to drive an IGBT and
we needed to slow down the propagation times.

The weird thing was the manifestaion of the problem:

A kink or knee in the middle of the falling edge would turn into a
"ring". Big time! The IGBT would then be driven in the linear region
for a a half second or so. After this happens umpty ump times the
IGBT would become an open.

So what is with the knee thing in the middle of the falling edge? I
have seen and have an intuitive feel for ringing at the bottom or top
of a square wave near the level transitions. But in the middle?

I was reminded of this by the falling edge (of all things) a PID
temperature controller. Going from 100C to -40C there it was a knee
right in the middle.

It has got to have something to do with some sort of level transition
for the temp controller case I would imagine SW. The opto coupler
though. . .

Sorry if I got rambling here.

Thanks,
Ed V.

PS - yeah, I know I should say the part numbers and edge rates but hey
it was two years ago. : )

That's Miller effect--when the collector starts to move, a current
opposing the base drive flows through the collector-base (in this case
collector-gate) capacitance. That causes the "knee" in the drive
waveform--the transistor's fighting back.

Since the transistor's collector has started to move, the thing is
linear, and prone to oscillate.

An opto-coupler driving an IGBT is kinda wimpy--that's not what we'd
usually call high-speed design!

A heftier driver (i.e., spend less time in the danger zone), a
damping resistor / ferrite bead at the gate, or some combination
thereof might be the cure.

Best,
James Arthur