Guru
Joined: 21/06/2018
Location: AustraliaPosts: 1267
| Posted: 07:27am 01 May 2021 |
Peter, the negative bias and the ringing are totally unrelated topics. In my career if I ever took a short cut and put something in production prematurely it inevitably would come back and bite me in the ass. This has caused me to be very cautious, pessimistic and strive to get it right from the get go - almost to the point of obsession.
When I made my inverter board Rev0 - that has exceeded my expectations to date - I had read hundreds of posts here and had a few odd arguments with others along the way and built my board the way I believed it would perform best and that meant not following a Mad/Oz style build (or thanks to you the dreaded EGS002) but going with my experience and intuition and drawing from all the good things I read that resonated with my thinking.
I did say from the outset that I did not want to build an inverter power board that would interface to an oz/mad controller as I did not believe the approach was right or best and considered that I would have an existing Achilles heel if I did.
The ringing that I found with each prototype I played with always responded and went away with a drain ferrite bead on the lower side FET's. But I had a totally different power stage. I used low impedance opto drivers with fast high current transistor buffers all on the same PCB avoiding the dreaded ribbon cable fiasco. All gates drives had a 2R2 and around 80mm of total track length.
I use a ribbon cable that carries opto drives in inverse parallel and I have isolated optos & power supplies the only common connection is ground which was only needed for measuring the input voltage.
So when I see an issue like the gate bounce I observed, I think maybe I should pull my head in and just be an observer. We are fighting two different animals in the end. You might build a hundred the way they are and they will all work properly on the other hand......I dont profess to know that answer only time will tell.
There is no argument that it works but perhaps there is not a large enough sample to draw all the right conclusions from yet.
I did not mean to say it was shoot through only that it was headed that way and how much space was there till it became shoot through. If it (gate bounce) could be lessened surely it would only be a positive result ?
Edited 2021-05-01 17:31 by wiseguy