Analog Filter advice required

[Nick Mulder]

Hello,

Question simplified:

Say I had a 48Hz square wave whose high V value was modulated over time (see the black line in the first attached graphic) and I wanted to extract and output the curve of the modulation as a voltage (the red line)...

How would I go about this ?

Some thoughts that may or may not be helpful/relevant:

The signal I am after is a low frequency signal compared to the one 'carrying it' (not to be confused with PWM!) - I thought that by simply adding a steep low pass filter I might be able to get way with it. But there is the worst case scenario to think about which is represented in the second graphic, imagine the 'high signal' were to fluctuate between max and min values as it does in the green box area, without a very (?) steep LPF how can I save that output from the filter ? (a 24Hz signal)

I was thinking of a combination of a LPF to get rid of all the higher frequency harmonics of the square wave but combined with a high-Q band stop filter set at 48Hz ?

I don't know much about the electronics here - just talking in terms I've learned from synthesizers ///

Again, not getting confused with PWM but thinking about it in terms of duty cycle it would be an odd situation where a %50 cycle would give a constant voltage if the high value if the original square wave yet a %25 duty cycle (the worst case scenario) would give a 1/2 freq (sine?) wave whose range was equal to the high voltage of the original square wave. Yep, its what I want, but is this odd or what ?

(addendum: I am aware that the red line appears to be 'aware' of the future signal path of the square wave in the graphics, I mean this to be real time so its output would of course lag behind the square wave signal, its just clearer in the graphic to display it this way)

[Hero999]

WHat are you doing with the signal?

If it's going to an ADC you just need to make sure the clock is in phase with the pulses so you only sample when the analogue signal is above 0V.

A sample an hold circuit clocked to the pulse frequency is another way of doing it.

[Nick Mulder]

The square wave Its the output of an LDR, which is conditioned into the red line signal which is then scaled appropriately to be sent to various analog synthesizers ... There will be at least 10 of these sensor set ups.

The idea was to keep it all analog, but maybe a bunch of sample and hold circuits could be interesting, once the sample rate is clocked with the square wave frequency and then phased correctly with the highs then I'd be getting pretty much exactly what I'm asking for huh ! - is there a way of achieving that with cheaper and/or analog components ?