It took me a while to figure out the huge difference between our Benjolins - your resonance is restricted to between 85-99% (actually maxes out at 98%?) Whereas mine goes from 0-97%. Also your drive is significantly lower, and your default cutoff significantly higher. It took me a long time to figure out how to get comparable sounds But I finally got there.
I find I am able to discover interesting sounds more readily with mine, but I don’t know if that is because I am more familiar with it, or if the filter settings really make that much difference.
I see you have a non-linear oscillator frequency response curve via shape master. I simply scale my oscillator frequency CV by 92%, and set the frequency knob to max out at ~8060 with 10V CV (before scaling). My CV range is +/- 10V. That seems to give an oscillator range that adheres closely to the After Later Audio spec, and is definitely not 1V/Oct.
I have reworked my Benjolin to use the linear feedback mechanism - I love it!
I have also created a version where I can switch between which oscillators are used.
- VCV LFO
- Befaco Pony in LFO mode (all x8 oversample)
- Befaco Pony in VCO mode (triangle x2, pulse x8 oversample)
I use the Count Modula Startup Delay to automatically send a sync pulse to the VCOs one second after the patch loads. I also added a manual sync button, but as you say, the Ponies do a great job remaining in sync, so I don’t think the manual button is really needed.
I think the Pony VCO mode includes a BLEP treatment (or whatever, I’m not fluent in DSP), whereas the LFO does not. But both are oversampled,
I was able to configure them so their frequencies nearly match perfectly, so comparisons can be made, and I am shocked how little difference there is in the final Benjolin output.
If I listen to the raw pulse output at high frequencies, then of course I hear tremendous aliasing out of VCV, much better (less) out of Pony LFO, and hardly any aliasing out of Pony VCO unless I go extremely high. (I’m running Rack at 48 kHz)
But typically with the Benjolin you are listening to some amount of the PWM and/or XOR signals, and they are not anti-aliased. So the differences between the oscillators largely disappear, even at high frequencies. Both Pony configurations generally are smoother in the high register, and the VCV is a bit grittier (which I happen to like). If/when I create my own Benjolin module, I will look into having an option to oversample pretty much everything, including the shift register, comparators, and logic gates. It will be interesting to see what effect that has.
Here is the patch to compare the oscillators. It is configured the same as my prior resonance CV swell video, except I raised the frequency to highlight the oscillator differences better. I’ve included the delay and reverb (not too much), but it is easy to hit the bypass button on the delay to kill both if you want to hear the raw Benjolin.
Benjolin V4 Resonance CV Demo Osc Compare.vcv (9.2 KB)
And here is the same patch with some different settings that produce some absolutely gorgeous drones.
Benjolin V4 Beautiful Drones.vcv (9.2 KB)
I was thinking my next “official” Benjolin patch release would use the Pony, but after these results, I think I may just stick with the VCV LFO - no worries about triangle/pulse sync, and less CPU.