Thank you very much for all this useful information, it is very helpful.
The main difference differences I can see are:
1) cascoding of the input devices
tried and didn't make any difference with regard to the undershoot.
2) Diode on the low side.
I am trying to understand the reason for it, but to me, it seems it provides a fast discharging path for the bootstrap on negative slope signals and basically clamping at -15V basically. I wonder if that is the culprit, I am curious to try to add the diode.
3) Limiting BW at the input.
I wonder if this prevents the undershoot by slowing down the input. I will try this as well.
4) You use 15V while I am using 24V for the rail. While this should only change the stability point, I don't recall seeing any difference on the undershoot but I can try.
I will try 2 and 3 and report my findings.
The main difference differences I can see are:
1) cascoding of the input devices
tried and didn't make any difference with regard to the undershoot.
2) Diode on the low side.
I am trying to understand the reason for it, but to me, it seems it provides a fast discharging path for the bootstrap on negative slope signals and basically clamping at -15V basically. I wonder if that is the culprit, I am curious to try to add the diode.
3) Limiting BW at the input.
I wonder if this prevents the undershoot by slowing down the input. I will try this as well.
4) You use 15V while I am using 24V for the rail. While this should only change the stability point, I don't recall seeing any difference on the undershoot but I can try.
I will try 2 and 3 and report my findings.
Just to update on this topic and sorry for not posting earlier. I wanted to clarify that if I take the buffer out of the feedback loop, I have the same results as anybody else. Square wave response is good.
The problem starts when the buffer is inserted inside the feedback loop. I have dug into this and the simulation also predicts this when closed in the loop. A simple pole compensation across the feedback loop proves to work and shape properly the response in the simulation.
Unfortunately, in practice, when I put a cap across the feedback resistor, the circuit oscillates.
Anyway, regardless of this, I wanted to close the topic on the buffer so that if anybody will read this in the future it can have a full picture.
Thanks to all those who kindly stepped in and helped, to the others who have seen this and my private messages and ignored, no need ta name and no need to comment.....ahahah....other than....you are not nice aahahahhahaha!!!
The problem starts when the buffer is inserted inside the feedback loop. I have dug into this and the simulation also predicts this when closed in the loop. A simple pole compensation across the feedback loop proves to work and shape properly the response in the simulation.
Unfortunately, in practice, when I put a cap across the feedback resistor, the circuit oscillates.
Anyway, regardless of this, I wanted to close the topic on the buffer so that if anybody will read this in the future it can have a full picture.
Thanks to all those who kindly stepped in and helped, to the others who have seen this and my private messages and ignored, no need ta name and no need to comment.....ahahah....other than....you are not nice aahahahhahaha!!!