What is the best way to process this signal?

[Deese]

The signal is a decaying positive dc voltage,there is a voltage limiter diode clamp in circuit so the max voltage of the signal is 600mv. So when the diode clamp turns off the dc voltage decays to zero in a curve. I want to measure/process that voltage close to zero,at a point in time,at the same point each cycle. The absolute value of the voltage is not really relevant,but the change in value from cycle to cycle is. If the voltage level increases we want to show the increase on an indicator/meter. So whatever the quiescent level happens to be we are going to null it/zero our meter etc.. Then read the change on the meter. The present way that is being used is a preamp amplifies the signal,then it is switched by a analog swith to an integrator. Supposedly there is alot of demand on the preamp in this method. It is said that it must have low noise,high speed,high BW,and must come out of saturation quickly. I was thinking that a sample and hold circuit,then a precision amplifier, would work better. Isn't the sample/hold ic optimized for high impedance/fast operation? Then a precision amplifier can amplify the voltage from the storage capacitor. The signal repeats at 1kHz frequency.

[danadak]

Seems like if that voltage decays according to a known functional curve,
like exponential decay, all you have to do is timeout fromn some start V,
like your 600 mV as a starting point, and then some time, any time,
measure V again and use the relationship to interprolate to the value
you want at the low V point.

If decay is not periodic then simply start a timer, and when it gets close to 1 mS,
take a reading. Certainly a S/H where you go to hold at timeout is fine, of an
A/D fast enough to meet your goals, like a SAR, A Cypress PSOC could handle
this quite easlily.

You will have to do an end to end error budget to see if it meets your accuracy/
precision requirements, taking into account T, Vref T & V, clock timing, offsets,
etc....


Regards, Dana.