I also wondered that I almost never used the PWM on the NM sofar because on some other synth I own, the PWM is the first thing I touch. Seems to be depending on the psycho acoustic quality of the PWM . I wonder where the differences in the sound of different PWM comes from and if it would be possible to "teach" the NM to sound like the PWM of a MS 10.

I realised that PWM on a NM goes in conjunction with a changing DC offset, maybe this weakens the PWM sound somehow. A patch that compensates the changing DC offset during PWM of the OSC output ? how ? Didn't had the time for DC offset research sofar.

An asymmetrical pulse has a DC offset when the signal level stays between, say -2V and +2V. The NM simply compensates by subtracting the PW control signal from the pulse output, so a 25% pulse might be more like between -1V and +3V. I suppose the reason is that the FMA inputs are very sensitive for DC offset.In the example patch you can hear what happens if you remove the DC offset when modulating a FMA input.

Theoretically the DC offset shouldn't be audible as AC coupling on the outputs of the NM and in the mixer/amplifier chain would remove any DC component.

I suppose that the difference in sound is caused by the fact that on a prepatched analog synth the oscillator signal always goes through the filter and the VCA. The filter circuitry, VCA and output amplifier do change the sound in the very high regions of the spectrum. There is a small degree of distortion in a digital pulsewave oscillator caused by the fact that the flanks of the pulse are forced in time within the e.g. 96kHz samplerate. Clavia uses a clever algoritm to reduce the 'time-jittering' of the pulsewave making it virtually inaudible, well, to my ears the pulse sounds very nice and clean.

...Until you start hardsyncing two oscillators and combine the outputs of the two oscillators. The hardsync input is quantized to the 96kHz samplerate and the slight 'phase difference' between the syncing waveform and the synced waveform can now produce an annoying spike. Which in turn can be turned into a nice noisy high by a Chorus module.

The interesting thing about a modulated pulsewave is that when the frequency is beyond the audible range, any audio signal can be coded by giving every pulse a width according to the average level of the audio signal during that particular cycle of the wave. (As is used in PCM and lots of digital gear, as many of you will know.) The PW Stepper patch, with the sequencers on the PW input, I recently sent in was based on that idea, but simplifying the idea. When taking a frequency, say 16 higher than the pitch you want to play, and separately control the pulsewidth for every one of the sixteen pulses in that period in a repeating manner, a rough sound is produced, but with a wide sound range. So its basically downsampling a PCM pattern until the coded signal itself becomes a sonically interesting wave. Or filtered heavily to get an interesting sound control over bass sounds.