303 clone (free and open source) - feedback welcome

[Uncle E]

Personally I've had the MAM MB33

The MB33 was a great synth. Not accurate at all but excellent sounding in its own right. I would love to see an emulation of that.

For Squelchbox, the sawtooth and filter sound quite good now and I could definitely see myself using it in a track. The square and distortion are the things that need the most work, IMO. If you use NAM for the distortion, you'll beat the distortions in Phoscyon 2.

A nice way to deal with the sequencer would be to reduce the range to a single octave, then add octave up/down buttons. This would make it more accurate to the real thing, too. If you want to keep the current range, the sequencer section should probably be about twice as high because it's pretty hard to set as-is.

Sync should probably default to Host and there's probably a way to do it that takes less room and integrates better with the rest of the UI. For example, make it a single button that changes between the three modes, switch the position with the waveform buttons, and change the waveform buttons to be a switch like the real 303 (except positioned on the face).

Lastly, it would be nice if you add a step mode to the tuning knob. It's pretty hard to dial in the way it is.

[Mike Janney]

Instead of using an Attack-Decay envelope for the accent you might want to try feeding a simple Decay envelope through a one-pole filter to create a smooth curve, which should be a quick way to get much more 303-ish accents

[JackedPotato95]

Wow, I love this, definitely feeling like the dumbest person in the room at the moment. This is quite useful, specific and actionable feedback.

As for the UI scaling I totally agree ofc, accessibility is a prime concern for these plugs - this should be solved in the latest release (works on my machine, although needs re-opening the window). Yes indeed the CV etc row at the top is simply eye candy, might just clean that up.

Back to tweaking!

[mystran]

Instead of using an Attack-Decay envelope for the accent you might want to try feeding a simple Decay envelope through a one-pole filter to create a smooth curve, which should be a quick way to get much more 303-ish accents

That's not quite right either, because when the envelope coming through accent switch (when it's active) is higher than the accent cap voltage, the cap will charge pretty fast through the diode, so there is actually a sort of "attack" here. Then when the envelope falls below the cap, the diode stops conducting and the accent cap needs to discharge through the filter CV circuitry, which is sort of release... although like your typical analog envelope follower, the release is sort of "always active" really and the diode might not really be a perfect switch... so the actual shape is somewhere between a smoothed bump and an AD-envelope... and then ofcourse depends on the resonance pot.

ps. If you wanted to really get it accurate, then running an actual circuit sim for just the VCF-CV circuitry (ie. the stuff between envelope and the expo current source) could probably be done, then take the expo current and put it through tan()-prewarp for a trapezoidal filter model? Idk.. not terribly interested personally on this right now, but .. that part of the circuitry is definitely a bit tricky.

[Mike Janney]

Mystran, I have spent countless hours analyzing this, what I was suggesting was a quick improvement not necessarily right though.

[mystran]

Mystran, I have spent countless hours analyzing this, what I was suggesting was a quick improvement not necessarily right though.

Oh, don't worry, I did recognize your name and I'm not under any illusion that you would suggest such simplification due to how you wouldn't know any better. I'm just in technical disagreement about whether this sort of approximation is actually convincing (well, you can do worse, for sure, but idk).

[Uncle E]

Does Phoscyon get the accent right?