Just got the miniwoog working and examining the wave generators.
I see that the [pd sawtooth] creates a much cleaner spectrum than a normal [phasor~]->[~-0.5]. The latter has a bunch of "junk" across the whole spectrum when looked at with fft analysis and it sounds hazier. Looking at the inner workings of it, but is there an explanation anywhere of how these clean waveforms are made and why they're not the default? Looking at fft analysis of complex FM in PD (done with only [osc~] objects) and those spectra are naturally clean. Are there any vanilla objects for generating "clean" versions of sawtooth, triangles and pulse/square waves? I guess I could abstractify this patch's generators, but there are a lot of arguments being sent in so it might help to understand what's going on...

EDIT: I see that the [phasor~]'s frequency is being jiggled around its center, the result is being multiplied by 1024, but not sure what goes on after that with the "number of harmonics" value and the tables (whose location or contents I don't know...)

phasor~ is "dirty" because is not anti-aliased (or band limited), so harmonics will "fold down" back onto the spectrum - there are many different methods of antialiasing oscillators.. there are none in pd-vanilla though.. I have some in my library using stilson & smith's "blit" (which doesn't work so well for frequency modulation though) https://ccrma.stanford.edu/~stilti/papers/blit.pdf

the miniwoog uses mmb's band limited saw https://github.com/dotmmb/mmb, which uses pre-computed band limited waveforms and crossfades through a large table of them and selects the appropriate number of harmonics based on the fundamental frequency.. this is a very safe method but takes more memory afaik

there are polyblep-based band limited oscillators in heavy lib https://github.com/enzienaudio/heavylib
and also mini blep ones in creb