I'd love to know more about the voice board cpu through to the CVs for the VCOs etc. Are you using a DAC channel per CV, or demux and sample and hold? Your tuning sounds so precise :-)
Ok, from the beginning: STM32F103 per every voice, DAC8554 x 3 totaling 12 channels of 16 bit CVs, two AS3340, AS3372E, AS2164 for modulation routing. Expensive, but I don't have to multiplex CVs. I bought DACs on AliExpress cheaply back in 2020 anyway :D
@@leeselectronicwidgets You have a lot of CVs flying around :) In my approach every voice have its own CPU, so auto tune is done simultaneously. Still, algorithm could be faster, but circuits needs some time to settle to get accurate frequency measurement. Worth noting is fact, that higher frequencies needs less time to measure
@@Krzychu-bh4rl Are you counting zero crossing cycles to determine period of note? I tried that but it was too slow on low frequencies for accuracy, so I sample and use an FFT now. It would be handy to tune all voices simultaneously! I have 24 oscillators to do, one at a time!
@@leeselectronicwidgets Yes, I have LM311 circuit cloned from Prophet 600, signal from comparator is connected to digital pin to count frequency. I'm impressed that you have FFT analysis for that, i wouldn't even think about it :). I had some troubles with my method on my previous project, made on ATMEGA328, but on STM it works with accuracy to fit tuning within one cent tolerance
![Bastian Balders - The Unknown [BAL010]](http://i.ytimg.com/vi/p6bmSFam_UU/mqdefault.jpg)
Fajne brzmienie. Dobra robota!
I'd love to know more about the voice board cpu through to the CVs for the VCOs etc. Are you using a DAC channel per CV, or demux and sample and hold? Your tuning sounds so precise :-)
Ok, from the beginning: STM32F103 per every voice, DAC8554 x 3 totaling 12 channels of 16 bit CVs, two AS3340, AS3372E, AS2164 for modulation routing. Expensive, but I don't have to multiplex CVs. I bought DACs on AliExpress cheaply back in 2020 anyway :D
@@Krzychu-bh4rl nice! I have 152 channels of CV, so had no choice but to demux / S&H... I like your tuning algorithm too, it's fast!
@@leeselectronicwidgets You have a lot of CVs flying around :) In my approach every voice have its own CPU, so auto tune is done simultaneously. Still, algorithm could be faster, but circuits needs some time to settle to get accurate frequency measurement. Worth noting is fact, that higher frequencies needs less time to measure
@@Krzychu-bh4rl Are you counting zero crossing cycles to determine period of note? I tried that but it was too slow on low frequencies for accuracy, so I sample and use an FFT now. It would be handy to tune all voices simultaneously! I have 24 oscillators to do, one at a time!
@@leeselectronicwidgets Yes, I have LM311 circuit cloned from Prophet 600, signal from comparator is connected to digital pin to count frequency. I'm impressed that you have FFT analysis for that, i wouldn't even think about it :). I had some troubles with my method on my previous project, made on ATMEGA328, but on STM it works with accuracy to fit tuning within one cent tolerance