passive peak detector

I have a square wave (output from a mosfet bridge) that swings from -100V to 0V referred to AGND (see picture) at ~60kHz; in some situation the signal has very narrow positive peaks (width of few ns). I need to detect those peaks, since if they go too positive they can destroy mosfets. A threshold of 15V would be OK.
I tried the circuit in figure: when there are positive peaks the capacitor is charged, and when it goes over Vzener + V[D1] + V[opto] then the capacitor is discharged over the opto (with darlington output) and the output generates an edge. However, as soon as the capacitor begins discharge, voltage drops and current quickly goes to zero, so the output of the opto goes down for about 1ns which is very difficult to detect.
Now, the biggest problem is that on the left of the opto I do not have any voltage supply available: is there a way to improve this circuit without using any supply?
Any ideas?
Thanks
**broken link removed**

I think I would have turned another way i order to solve the problem.

I would have put a timer circuit combined to some logic circutis behind the opto isolator.

A 1N4007 is a slow recovery power diode. It is not suitable for detecting short transients.
You need to use a Schottky diode instead.

Try increasing the value of the capacitance to lengthen the output pulse.

Do you see the positive pulses with your circuit connected? If the width of the over shoots are only a few nano seconds, you should try to clamp them because the speed of any circuit will be too slow to do much else with them.
Frank

chuckey said:

Do you see the positive pulses with your circuit connected? If the width of the over shoots are only a few nano seconds, you should try to clamp them because the speed of any circuit will be too slow to do much else with them.
Frank

Click to expand...

Hi, unfortunately I cannot clamp the peaks with SMCJ100CA. Not sure why... the peak reach >200V undisturbed.