High-side Pushpull driver for IGBT?

[cupoftea]

Hi,
We’ve got a IGBT driver. It’s a fast opto chip with common emitter pull down in the high side. This pull down simply pulls-down-then-lets-go of the centre point of the pushpull stage. Top of pushpull is Darlington, with 4k7 pullup. Bottom of pushpull is PNP…as said, the base of the PNP is pulled down by the common emitter opto output txtor.

There is a signal diode connecting bases of top and bottom txtors of the pushpull (cathode to PNP base). There is also a 1MEG feeding into the PNP base, from the positive high side rail…..we believe this 1MEG is meant to keep current flowing in this signal diode when the top Darlington is ON…….would you agree?

The purpose of keeping current flow in this diode is so it acts as a path to sweep out minority carriers from the PNP, would you agree? Thus this PNP will switch quicker.

 

[dick_freebird]

I'd need to see a sketch but believe 1Mohm isn't going to "sweep" very much or fast. Might just be a passive turnoff?

 

[cupoftea]

Thanks, i am also considering that, but then wondered if the "safe passive state", would be with the PNP ON...so that IGBT gate was grounded.
But anyway, it doesnt seem to have a good "raison d'etre".
Maybe when its value is lower it can sweep out more minority carriers.

 

[cupoftea]

Also, i think we have found a bad circuit because a double NPN Darlington is slower than a single NPN...so you would surely never use a Darlington in the top transistor of a pushpull gate driver?

It also has an RCRC filter going to the gate of the IGBT...where the final C is the IGBTs gate_emitter capacitance.....this is surely a mistake............it coudl be done, but you woudl simply use just an RC?

Also, the gate drive protection zener is not directly at the gate of the IGBT...but is a bit back from that...its on the collector of the BJT that drives thee pushpull (AKA totem pole) pair.

I believe we have found ourselves a duff circuit here.

What on earth the signal diode is doing between the bases of the Darlington_NPN and the PNP is anyones guess.....doesnt do any good thats for sure...probably doesnt do much harm though either.....but id rather be without it and be sure the upper Darlington was OFF when the lower PNP was ON.

 

[FvM]

Is it asking too much to expect a circuit sketch?

 

[cupoftea]

Sorry i cant. Not for a few months at least.....i am pretty sure its duff though. I woudlnt give it if it was good.....but this one's got duff written all over it.

 

[dick_freebird]

A Darlington made of individual discretes can let you sink charge from the lower's base which is the nut of the speed problem. One made of common collector slab without that tap will be a pig.

If your two bases are driven by different paths or there are back-clamps from output emitters to bases, the signal diode may be another bit of that BVebo protection network. Keep the next layer in, right side up (or only a little bit upside down).