Understanding Streams schem: DAC vs PWM out

I’m studying the Streams schematic, after stumbling upon the VCF’s with VCA’s document: http://mutable-instruments.net/static/documentation/vcfs_with_vcas.pdf

Each channel has a VCA into a 2 pole VCF, which is 2 simple cascaded stages. I’d like to know why the (somewhat expensive) DAC feeds the VCA section, but the PWM out is adequate for the filter cutoff. Is it a characteristic of how the filter circuit behaves that it doesnt need such precision?

I don’t actually have a Streams to play on yet, just a student tinkering with the idea of DIYing a digitally controlled V2164 based mixer

Also is there a repository somewhere where these docs by Oliver can be accessed? I’m finding them valuable. I’ve stumbled upon the SMR4 filter analysis and the one above through google

When using PWM, one needs a low-pass filter (R4/C6 for example in Streams’ schematics) to eliminate the PWM carrier.

If the cutoff of this LPF is too high, the PWM carrier is still present, and can “leak” through the circuit that makes use of the filtered PWM signal.

If the cutoff of this LPF is too low, there will be a perceived “slowness” or “inertia” when the CV changes.

That’s why, for Streams, I decided that PWM was not suitable for the VCA: not only I wanted the VCA to be as fast as possible, but also, a residual PWM signal fed to a VCA control port creates an annoying high frequency tone in the resulting signal. As for the VCF, it could deal with more slowness (in the order of 1ms), and the ~100kHz modulation of the VCF cutoff caused by the residual PWM carrier can be barely detected.

Thanks Olivier for the detailed explanation