@phil123456

How about this test with the [vcf~] ? It sounds close to me but maybe you are hearing something else.

vcf.zip

**edit ** updated with [vcf~] <--> [moog~] low freq cutoff comparison. Try vcf~ with only the right outlet as well.

Here is another kind of array object (no structs this time) with free point plotting and curve transitions. There are two layers in the GOP box: the layer with the canvas points and one below with an array-table and a resized slider in front to block mouse interaction.

array2.zip

(see title)

arrayresize.pd

@weightless

Ok here is a final one that simulates the cz waveforms in series by writing half of the array in the first chosen waveform and the other half by the second. If both waveforms are the same, the second phase slider won't do anything.
txcz-wavs3.pd

As far as the array size I am not sure, but I believe at the really low resolutions the actual shape of the waveform is simply altered.

@weightless

I tried your idea with the alternating tables only, but even the "set" messages will cause some clicks, so here is one with tabread4~ & volume alternating that works. The extra cost is one tabread4~ & two sets of line~s and [*~]s but with effective [switch~]ing you could take away the extra tabread4~.

I also experimented with different array sizes (see the switcher at the top-right), and certain waveforms are affected quite a bit while others remain the same, so if someone wants to go this route, the whole setup could be fine-tuned per waveform. In this patch the [*~] before the tabread4~s is changed by a 20ms line~, but if you wanted to change the waveform per voice in a polyphonic setup, before the volume envelope kicked in, and each waveform had a specific table size, this would have to be done immediately with a way to avoid the click.
txcz-wavs2.pd

@weightless

I thought it might be nice to group all the waveforms to be read from a single tabread4~ for each operator, and here is the result. The trade-off is that the casio waveforms cannot be modulated at the 2nd input since they are now calculated statically (you can change them live but since the tables are being redrawn, its unusable that way). The feedback is in here too and a comparison w/ the original cos~ style.
txcz-wavs.pd

@weightless

I cheated to get some pre-made waveforms with this one (waveforms 2,4,6,8 may not be accurate), and I am wondering if there is an advantage to using the audio math instead, because this one is certaintly lighter. I added a tri-wave at 9 too.

tx-wavs.pd

@weightless

Ah that was right there in your first example. Nevertheless, amazing - thanks.

@weightless

I don't have any answers for you to get those arcsin-looking waveforms... but I have also been curious about how to use tables instead of the phasor-cos~ pairing... since the latter generally has nicer sounding modulation
cos_vs_sintable.pd

Two more versions. The first one only adds the lateral restriction to graph-adsr-v2 while keeping the minimalism. The second is a kind of hybrid between the last two postings.

adsr-g2.zip