There is a photocoupler (LTV-817) that I am using that is seeing a power glitch (~1 microsecond) and it seems to take a long time for the photocoupler to recover (~35 microseconds).
The blue trace is the power glitch that I am seeing and the yellow trace is the "output" of the photocoupler.
I was wondering if anyone had ideas on reducing the response time of the photocoupler (the yellow trace). At the output (yellow trace) what is happening? The part is saturated?
Please show your complete input and output circuit of the photocoupler.
We need to determine where exactly your scope probes are connected
And we need to determine the photocoupler current (not the voltage) at input and output ... at normal conditions and at glitch...
This is output of the photocoupler to our comparator:
The blue trace in the previous post is the measurement across R421 (1O OHM) with a differential probe when the high voltage lines (HV_P and HV_N are shunted to ground.)
The yellow trace is what is measured from the emitter of transistor across R422 to ground. (signal Over_I).
The purple trace is the output the comparator U94 meaused on test point TP59.
Relaxation time during cutoff can be reduced by reducing Re to RL=100Ω and turn ON, VCE=2V, IC=2mA . Load capacitance affects the RC decay (pF) with emitter current shutting off, output impedance rises rapidly.
If lowering the load resistor value still does not give a fast enough response, you may have to use a faster opto such as a 6N136 (LTspice simulation below).
If you read the datasheet of the used opto coupler, you realize that the shown recovery delay is in the expectable range. The fast attack time is in contrast suprizing, but it's only achieved by overdriving the optocoupler massively.
your circuit:
I don´t like the idea to drive a LED without current limiting resistor.
I assume this makes the LED current unpredictable. In worst case you kill the LED.
When I see this correctly, then the pulse is about 6V .... this is way off specification, probably several Amperes of LED current.
--> Try to add a series resistor.