It's similar, if the MUNT source code is correct: It takes a sinewave and plays the corners fast or slow to round or un-round them to act as the "filter". Up all the way, it's a squarewave. That is then put through a waveshaper to produce the saw. I believe the resonance is something similar to the CZ's. You can fake resonance a few different ways though (one other way is cross-fading between a couple sinewaves, going up the harmonic frequencies)

@@groovydsp1716 ok, Thnx :)

@@groovydsp1716 Roland's D50 manual says that the SAW is formed from a ring modulation of a SIN and a square .... interestingly, is it the same in the D110? Nothing is said about this in the manual ... but waveform, it also resembles a quarter of a sine. Or maybe it's just sampled?

@@groovydsp1716 and maybe You know ... how implemented PWM in D-110 or D-50 ? Algorithmically? Or just sampled the cycle table for each percentage value?

From what I gather, the synthesis approximates it's waveshapes through the mathematical formulas he presented. So they don't appear exactly like the normal shapes. It's more of an abstraction. I think that is why the starting and ending points of the squares appear as more of a sloped shape.

it's look like a square wave passed in a high pass filter with cutoff frequency lower than the note. 🙂 Several analog synths have square wave like this, with a slope. It is a question of phase rate ,the core of PD. With a nominal phase rate of 1 all along the phase, you have a sine or cosine wave , and you have only one frequency. The part that have a strong impact on the spectrum , the harmonics content are parts with higher phase rate , >= 1 , while lower phase rate have less impact or no impact on the spectrum . This is why the flat part of the square wave can be changed in a slow slope without impacting the spectrum, while it keep the symmetry.