This is the patch that I used for for the 'Controlled by Sample and Hold 3" example on the Buchla 291 BPF Adaptation Page.
You need the following modules:
2 VCO's with pulse output
1 Sample and Hold capable of receiving an external trigger
1 Filter (291 BPF used in this one)
1 Envelope Generator
1 Noise Source
The random sampling rate is accomplished by triggering the sample and hold with a VCO that has its frequency controlled by the sample and hold output. Since a high sample and hold output will drive both VCO's higher in frequency, the 'high notes' will burst at a faster pace than the 'low notes'. If you want it the other way around, an inverter could be placed between the sample and hold output and the triggering VCO's control voltage input. Then the 'high notes' would come slower than the 'low notes'.
The correlation comes from mixing the output of the sample and hold with the sampled input, which in this case is a noise generator. It's a bit different than correlation normally is - in most cases I've seen, the correlation control will fade between the sampled input and the sample and hold output, so as you turn the control, you have more of one and less of the other. In this setup, you adjust the two levels independantly. The effect of correlation (at least set up like this) is that you can group high voltage bursts closely together, with little 'in between' and low sample and hold voltages. The general pattern of voltages depends a lot on these two mixer controls.
Getting all of the controls just right in this patch can be very tricky. For example, if the level of the sample and hold signal going to the VCO is too high, or if the VCO initial frequency is not set 'just so', you may get a burst of high notes, and, nothing else until, oh, around noon tomorrow, because the voltage driving the VCO dropped so low it'll take the VCO that long to trigger the next sample.
So, it's good to use inputs with variable attenuators on them so you can adjust the levels throughout the patch. That way, for example, you can adjust the speed of triggering independent of how far the sample and hold voltage affects the frequency of the second VCO without changing the relationship of the two correlation controls of the mixer.
The main thing, once the patch is setup, is to start fiddling with the controls. You find a lot of cool things along the way........