Phase consistency between 10 VCOs...
Hi guys!
First post here for me. I'm an Italian ee that develops modular synthesizer modules (mainly analog... until now!)
In the last two years I lurked EZ fascinated by the ADAU1701 and the marvels that you guys (KJBOB especially came up with to exploit the power and to overcome the limitations of the machine).
I decided to start a line of dsp based products that range from oscillator/generators to effects and filters.
Main aspect: no micro involved, so everything is done inside the 1701.
Franky I am impressed to what can be done with this limited-resources part and during the last weeks I designed at least six/seven modules based on it!
Now, the problem.
I have an additive synthesis oscillator that sums, via a series of control strategies, the fundamental and the first 9 harmonics to obtain the final result.
Phase consistency is key, and it 'should' be kept during time and during the change of frequency (that comes from the outside in the form of a 1V/octave control voltage.
I've noticed that while I am modifying params of the schematic (using the MINIZ here) it happens that sometimes, quite abruptly, one or more of the VCOs gets out of phase, destroying the
waveshape of the output signal.
I want to ask if this is normal (the fact that it never happened during normal use of the 'module' makes me think this way) when you are 'stimulating' the IC with many communications or it's not
linked at all with activity on the I2C....
Obviously I am expecting that the VCO gets 'out of phase' after some time have passed (should depend on dsp internal precision etc.) So I should plan for some kind of 're-phase' every now and then...
Thoughts?
PS this is a really PRECIOUS space, thanks AD and everyone that contributes!
Davide
Audio Signal Processors
SigmaDSP Audio Processors
adau1701
Hello Davide,
Thanks for the mention. Squeezing more from a ADAU1701 is fun!
It appears from experience that when applying exact multiples to VCO inputs, they stay in phase indefinitely. This has come up in FM Stereo encoding/decoding, where the correct phase relationship between VCOs running at 19 KHz and 38 KHz must be maintained -- for an example, scroll down to the final post in this link: FM Stereo Encoder Problem
I'm not sure why an occasional disruption occurs in a DSP when connected to SigmaStudio. Possibly it's because SigmaStudio does use block writes (as opposed to safeload writes) when modifying parameters for some blocks other than filters, such as DC sources.
Best regards,
Bob
Hi Bob, thank you for the reply.
I will check this soon. Until now I never used the board with its eeprom.
The only potential difference between the FM application and mine is that the VCO frequency (the main and the overtones) is changed continuously (as you would expect in a musical application).
In the unforntunate case of a destructive out of phase I guess the only way to 'recover' the initial solution is to use a 'syncable' VCO and trigger a re-sync for all the VCOs evey now and then.... is it correct?
Cheers,
Davide
Hi Davide,
Resetting all the VCOs hopefully will restore the phase, yet it may cause a glitch in the sound. I was wondering if it would make sense to reset the harmonic VCOs to the fundamental VCO -- that is, each time the fundamental VCO rises through zero it would force all the others to also rise from zero. So I tried the circuit shown below, but it didn't work. It causes audible artifacts in the harmonic VCOs. I did not attempt to troubleshoot further.
Best regards,
Bob
Thanks a lot!
I think your suggestion is correct! I will try to 'hide' a timed re-sync using a long, and possibly random, timer made with a counter. A glitch every now and then shouldn't be noticeable, especially with modulated sounds..
Please log in to reply.