2 CEM3379s contain 2 VCF (with cutoff and resonance CVs), 2 VCA (with gain CVs), and 2 panners (with pan CVs) = 8 CV in total ; so it would make sense to drive them from a single DAC and a 2300 if we lived in the 80s. I don’t know where you get those 4 VCA / 1 VCF (maybe from the chip diagram) - but what matters are the CV “exposed” to the outside of the chip (for example the panner works internally with 2 VCA, but it only has one CV).
You mention CV1 and CV2 and I am under the impression that you want to do a Shruthi-1 filter board with a SSM2300 and some CEM3379s. This makes very little sense. To be useful, the SSM2300 needs 8 time-multiplexed CVs as its input, and the matching data on its address bus. What the Shruthi outputs is one single, plain CV and no address data. I don’t know how to explain it - but in short you are trying to drive 8 different parameters from a signal containing only information for 1 parameter.
Let’s take an example… You have two CEM 3379 corresponding to two voices, and your patch and envelopes are such that at a given point in time you need to output the following CVs to the voices:
vca gain: 5V
vca gain: 4V
Instead of having 8 DAC outputting those voltages to the 2x4 CV inputs on the CEM3379 (the modern solution), you can have one DAC and 3 digital lines connected to a SSM2300 (the 80s solution), in this case your microcontroller needs to cyclically emit the following sequence
digital lines = 000 ; DAC = 2.5V
digital lines = 001 ; DAC = 0.0V
digital lines = 010 ; DAC = 5.0V
digital lines = 011 ; DAC = 2.0V
digital lines = 100 ; DAC = 1.5V
digital lines = 101 ; DAC = 0.5V
digital lines = 110 ; DAC = 4.0V
digital lines = 111 ; DAC = 3.0V
If you look at the CV signal on a scope, it looks like staircases which will be “demultiplexed” into the 8 CVs by the 2300: what you will get on the output 1 of the SSM2300 is a continuous 2.5V signal, on the output 2 a continuous 0V signal, etc… And this is why you need the sample & hold aspect : the voltage on the input of the SSM2300 will carry the voltage to output on its output 1 only 12.5% of the time, so the sample and hold “fills the gaps” while the input voltage successfully sets the other outputs.
If you connect the CV output of a Shruthi-1 to the input of a SSM2300, one of the following thing will happen: either you’ll get the CV signal on output 1 and crap on the other output ; or you’ll get the CV on 8 outputs. Both are equally useless. You need logic in the firmware to write the sequence of values, and 3 digital lines (which you don’t have) to write the address.
Enough said… the SSM2300 is a boring plumbing chip which made sense in synth architectures the Shruthi clearly doesn’t belong to. This chip is from an age when DACs were parallel and horribly expensive, and MCUs slow. Now for the price of its modern days equivalent (the AD SMP08) you can get 2 or 3 awesome octal SPI DACs from TI. Don’t be fooled by the rare status of the SSM2300: in a modern context it’s worthless.