I'm not looking for an description of how you are trying to solve your problem, I'm asking for a description of the problem you are trying to solve.

Kudos to @WBahn for finally eliciting a response that cleared up the question.

 

If we can assume that the pulse amplitude is not changing drastically from one pulse to the next, then a simple low pass filter with the appropriate time constant should be sufficient in lieu of the peak detect circuit.

 

If we can assume that the pulse amplitude is not changing drastically from one pulse to the next, then a simple low pass filter with the appropriate time constant should be sufficient in lieu of the peak detect circuit.

Definitely worth consideration. A lot of devils are in the details on that, though. Are the pulses tightly related, or largely independent? What is the purpose of detecting pulses that fall within that window and how accurate does the detection have to be and on what scale? But those are now questions that can be put to the customer that might not have ever been thought about previously.

 

First, the LM311 is not a regular comparator, it has an output transistor with both the collector and emitter open, and it also has an input bias adjustment connection. So it is highly flexible. A scheme providing the different output signals can be provided by multiple comparators and some digital logic. An LM339 quad comparator would be simpler to apply and require less board space.

 

First, the LM311 is not a regular comparator, it has an output transistor with both the collector and emitter open, and it also has an input bias adjustment connection. So it is highly flexible. A scheme providing the different output signals can be provided by multiple comparators and some digital logic. An LM339 quad comparator would be simpler to apply and require less board space.

Given that the response time of the LM339 can be over a microsecond, I'm not sure how well it can be relied on to work with pulses that last only half a microsecond.

 

Wish we knew the actual application of this window detection circuit.
If you can average the peak voltage over say five pulse duration, then yes, a slower comparator (5 μs response) would be appropriate.

 

Given that the response time of the LM339 can be over a microsecond, I'm not sure how well it can be relied on to work with pulses that last only half a microsecond.

The slow response is for small signals. The large signal response time is much faster.

 

The slow response is for small signals. The large signal response time is much faster.

For TTL level signals it's only 300 ns. That's not much margin. Plus, there will be signals that only rise above the threshold by a small amount, which interacts with the question of how accurate the threshold values need to be and how long the output pulses need to be.