@Jens.Peter.Nielsen, are you still wanting to do get CV output as CCs in 14-bit mode? I had the same need and couldn’t find another solution, so I just coded it up. Rack is now working (excellently) with the high-res mode on my Intellijel Midi 1U/CVx pair.
Yes, I would like to have that - I have pcb’s to build a few MI Yarns, (I don’t know if they have 14-bit control - if not, the FW is opensource.) they use different 16 bit DAC’s(DAC8564) compared to the Intellijel Midi 1U/CVx (LTC2609 in the Midi1U) - but should be good enough to output 14-bit resolution CV.
As to the jitter, the LSB is sent always, the MSB is only sent on change. The preferred order is MSB/LSB. The MSB message can buffer the control value such that the terminal LSB triggers a MSBCache*128+LSB assignment. The MSB changing without the LSB changing is statistically less likely than any other combination. I’m not sure if I would go as far as implementation of dither for delta-sigma dynamic range extension based on a controller timed byte emission.
EDIT: if the emission nyquist is not high enough, the “noise” would perhaps muddy the sound. It maybe advantageous to say emit a percent of the unchanging MSB bytes, as the initialization state of what is in the MSBCache needs to be sent within a reasonable performance start interval, and in some sense be immune to frame errors on reception. All Notes Off/Program Change would be ideal MSB sync points if perfect transmission was assumed.
It really was just a data rate thing. Come to think of it, having the individual bits of program change would be “good” boolean CV signals.
EDIT2: perhaps 7 inputs and 7 outputs for x->a, y->b, z->c, etc. splitting the program change message VCA style.
Thanks for this clear explanation! Alain
I like this idea!