[sdiy] Linear response VCOs?

[brianw]

Not just overkill, but also not scalable.

If you needed 40 CV outputs, you're not going to pay $250 for that. Besides, the R-Pi probably can't keep 5 audio CODEC chips fed properly at 32/384k.

I am also suspicious of the DC coupling. 24-bit converters have high precision, but low accuracy. I'm not even sure what it means to have a 32-bit DAC, because the noise floor is around the 20-bit level. Extending from 24-bit to 32-bit adds a lot of precision, but absolutely no accuracy is gained. Fortunately, for audio signals, the noise is ignored by the human hearing system. For CV values, I suspect that the noise would have to be filtered before it would be as usable as a simple DAC.

A 14-bit parallel multi-bit DAC can run at 125 MHz, with 8 ns latency. The 24-bit/32-bit delta-sigma DAC might have around 450 us of latency. Probably not an issue, but larger latency is something that has to be designed around.

Realistically, the parts for a design geared towards generation of far more than 8 CV outputs would be cheaper, but it depends upon how popular such a thing would be in the R-Pi world. It might end up costing more due to low quantity, compared to an board that can handle surround audio - which is clearly more popular. Sometimes, shoehorning a cheap product into a task that it was not designed for costs less than an efficient design.

Brian


On Mar 19, 2026, at 4:00 PM, Thomas Hudson wrote:
> Very interesting. I just came across two different dc coupled eight channel DAC's for the Raspberry 5 supporting up to 24-bit/192kHz or 32-bit/384k, for around $50,  though probably overkill.