A reasonable theory!
I’m also a subscriber and might send a similar letter.
Nice one - CF100 sounds like a beast! Great to see you modelling more filters in Rack Andy.
2 Likes
Hey Funkybot, welcome along! I think there is still room for a Cytomic plugin modular/semimodular, as it will have more streamed modulation - and also will definitely try and add this model to The Drop - but for that I’ll also need to do a dedicated full High Pass cascade model as well, this is only a low pass cascade and the high pass is generated with a weighted sum of low pass stages. Doing it this way means the response “breaks” more easily with drive, which has a completely different sound to a dedicated high pass filter.
edit: I’ll check with Andrew Belt if there are any legal issues with providing this as a fork of VCV Rack on github, and I’ve also done a mod for better noise generation for the Noise module in Fundamental - I need consistent sounding noise at different sample rates for testing in Rack.
4 Likes
Yes, this was recorded with E440 v2.0.2 and CF100 v2.0.0 at 44.1khz.
My main concern with the E440 sound isn’t even the high resonance aliasing, it’s more the fact that with low cutoff and having a loud input signal it totally dies and goes silent. In my circuit model this doesn’t happen, and I love the sound of deep driven bass, which is part of the reason I decided to code my own version of the Discrete OTA filter. Here it is shown a bit more clearly:
https://cytomic.com/files/forums/cf100-vs-e440-vid2.mp4
The Unison mod for VCV MIDI to CV is free for anyone to use, but you have to compile VCV Rack Free to use it. It was done as a tech demo for Andrew Belt to see, and hopefully make it, or something similar part of all versions of VCV Rack. It’s a very basic change to the voice allocation mechanism, and took an hour or so to code, so I’ve no issues giving this away and signing any copyright over for that code.
5 Likes
Another thing is that the E440 doesn’t have any even order harmonics when you drive the input with a pure sine wave, or put the filter into self oscillation:
Sin input drive harmonics:
Self oscillation resonance harmonics:
3 Likes
well, as you know, no even harmonics is pretty normal - like a moog filter. On my Stairway filter, from back in ancient times, the non-linearity is user choosble. But the default is a symmetric tanh like those things tend to have.
1 Like
Very interesting! Thanks for the deeper look.
Odd vs. even harmonic structure (e.g. hard vs. softerer clipping) I can chalk up to different flavours and it loosely lines up with my “lovely” vs. “gnarly” impression from your original video (curious now to test my hardware E440 to see how it clips) but the signal-swamping (and, frankly, that amount of aliasing post-2.0.2) seem like they might be in the bug report realm. Time permitting I’ll probably write it up and contact support.
Unison, noise, etc.–nice! I have my own in-house menu of compile-requiring mods to little bits of Rack (e.g. 14-bit MIDI output, some browser mods, etc.); I suspect a lot of us do. It would be interesting to spin up a thread on that sometime, but I don’t want to hijack your own announcement thread for it 
1 Like
I’d love for you to post some audio samples of self oscillation harmonics on the hardware E440 1khz if possible at maximum resonance and also perhaps use some resonance cv to push it higher if needed to get about the same levels as you can see above. Also the sound with no resonance of a sawtooth with as much volume as possible pushed into the input, and the cutoff as low as possible and then slowly raised to see what happens!
2 Likes
I’ve never seen a perfectly symmetric all odd harmonics only bit of analog hardware, there is always a little bit of second order harmonics, but you may need to push it further till you see it for some filters, or you may have to increase the depth of the spectrum analyser down to around -120 or lower. Please post in some audio / screen grabs of analog kit you have that is perfectly symmetric, I’d love to see it!
Here are a few other synths 1k self osc:-
This is the Moog Matriarch:
And this is an original MS-20 mk2:
4 Likes
Oh, for sure not “prefect”. And those illustrations are way more than I would have suspected. I see you are entirely correct. Most of my measuring of this stuff had been from like guitar fuzz boxes and such, where the even harmonics are much less. Anyway, I learned something new -thanks!
1 Like
Will do! Should have time in the next few days.
2 Likes
So stoked you came to the vcv world to bring your talent. Cant wait to use this filter in my productions
2 Likes
Sorry to point out, but even with guitar stomp boxes there is always plenty of even harmonics. For example, the Ibanez Tube Screamer which has symmetric diode clipping is shown below:
Since the drive is fed back into one side of the op-amp, that side is “stressed” more than the other, so this generates increasingly even order harmonics with increasing drive due to things like finite Beta, the Early effect, and other real world things that go on in transistors.
3 Likes
The even harmonics are down 24 db. Not super audible.
Not sure about your ears but this is very audible to me having 2nd order harmonics at -42dB below the fundamental!
My point is that second order harmonics are always there in analog circuits, so they should always be there in an accurate model of the circuit, but they are conspicuously absent in certain models.
2 Likes
Well, @andy-cytomic . Sure you are correct. I will say that back when was “the alias police” that many ppl claimed the couldn’t hear aliasing that was much louder than that. But still yes, I agree you are correct.
Hey Squinky, what’s up with the Assiniboine reference here? Seems kinda random. What do you mean?
1 Like
Could be a very weird spelling autocorrect result for “andy-cytomic”
1 Like
I think that must be it! For sure I have no better theory. I’ll fix that.
I think @main.tenant 's theory is correct. autocorrect and sloppiness on my part. fixed now.
2 Likes