[sdiy] Generating a large number of CV outputs

[brianw]

On Dec 8, 2023, at 6:50 AM, Mattias Rickardsson <mr at analogue.org> wrote:
> Roman Sowa <modular at go2.pl> skrev:
>> OTOH I would love to see someone doing 100-outs MUX with single DAC 
>> spitting 16-bit precision at 1Msps
> 
> That's basically what we do since 2012.  
> 16-bit 1MSPS DAC running at 0.96 MHz, multiplexing its output to 40 CV channels, each at 24kHz update rate (no zipper noise in the audio band). 
> Works well, but I wouldn't want to try controlling that DAC/mux timing with a microprocessor, or build it on a single-sided board. :-)
> 
> /mr

One question I have is, "How much reconstruction filtering is necessary?"

Granted, my question is about a large number of signal outputs, NOT a large number of CV outputs.

One DAC with many MUX S/H outputs cannot be filtered at the DAC output, because you don't want to mix the channel signals together with the filter. Instead, each individual MUX S/H output needs its own reconstruction filter. I went with a single-pole low-pass, but that requires significant oversampling compared to the desired signal frequency bandwidth. Higher-order reconstruction filtering helps lower the oversampling requirements, but adds to the cost of the analog circuitry, not to mention the PCB real estate.

What do folks tend to choose in this scenario?

Brian Willoughby