Need an analog ckt to integrate under a curve

[biff44]

Have a waveform I have to detect. Most of the time the voltage is hovering around zero. But every once in a while an event happens where there are some voltage humps over a short burst of time--maybe 3 voltage "humps" over a 1 ms period. Is there some way to integrate the area under the V vs. Time curve to make the 3 humps add up to a bigger detectable signal?

I am aware of analog integrators. But I need one that integrates up to a big voltage output, but then slowly droops off back to zero output so that I can see the event easily. In other words, I need an integrator with different attack and decay times.

Any circuits?

Thanks.

 

[keith1200rs]

Just add a series resistor to the capacitor of an integrator and a parallel bleed resistor across it?

Keith

 

[FvM]

A real integrator has to be reset at discrete times by a switch or continouosly by a finite time constant (which turn the integrator into
a first order low-pass). The latter method is more common and used in most pulse processing circuits, e.g. for charge amplifiers in nuclear
instrumentation and other pulse measuring applications.

By adjusting the time constant, you can try to filter low frequency signals and zero drift appropriately. If this doesn't give sufficient selectivity,
non-linear "zero-restore" circuits are an option.

P.S.: An integrator has - by it's principle operation - zero attack time. If actually helpful (I doubt about), you have to add a low pass
to the signal chain to introduce an attack time.

 

[biff44]

Thanks. Just dawned on me that I could reset the integrator pretty easily. The system frequency sweeps, and the waveform I am integrating is the detected power. I can easily reset the capacitor after each sweep.