Randomize Shape, Phase or Amplitude of Low Frequency Sine Wave

[pnielsen]

I have generated a 32Hz sine wave using a crystal-timed CD4060 followed by a two stage, passive RC filter.

I would like to make each cycle of the sine wave vary "randomly" (non-repetitively), either by distorting its shape, phase or amplitude ... or multiples of same. A 10-30% distortion of these parameters is about what I am aiming for.

Can anyone offer a suggestion that does not involve DDS, external equipment or something strange like a photo cell aimed at a tree? This is a simple circuit so I am looking for the simplest solution. Perhaps something involving an op amp mixed and non-linear feedback loop, but I am unsure of the design approach given the low frequency.

I tried mixing in unprocessed noise. It did not work since, to achieve the desired effect, the distorting influence needs to occur below 32Hz.

 

[unclejed613]

you can use a 4 quadrant multiplier, and use a noise source rolled off at 16Hz as the "y" input, and your 32hz sine wave on the "x" input. either an AD633 or MC1495 would work as the multiplier. a zener noise source and a 16hz low pass filter could provide the noise source. if you want lots of noise, use a zener with a voltage above 6V (a 12V zener would be good). zeners above 6V operate in avalanche mode which is more noisy.

 

[pnielsen]

A circuit like this, but without the LF411? That is about what I was looking for. Does the random modulation created by the filtered noise only affect the amplitude of the 32Hz sine wave, or does it affect other characteristics as well?

 

[unclejed613]

with a 4 quadrant multiplier, you will get amplitude changes, phase inversions (which you wouldn't get with just adding noise), and intermodulation products (which also would be absent with just adding), which means you will get random things in the frequency domain as well.
you don't need the output op amp, which looks like it was put there to have two gain ranges (unity gain for AM modulation, and a gain of about 3 for full 4 quadrant multiplication).

 

[pnielsen]

I was going to simulate this concept in LTSpice and post it here, but no multiplier model is included with the software.

I assume only one multiplier is needed. A discrete type, like the Gilbert Cell below, could be used instead.

Or, here is how to import an AD633 SPICE model into LTSpice. https://www.electronicspoint.com/threads/ltspice-and-ad633.25141/

However, since I have an AD633, I will just try it on the bench.