Hi everyone!
For my String Machine project, I need to convert the Sawtooth wave coming out of a osc/phasor into a Square Wave. Below is my subpatch in charge of the conversion. The Phase is converted to Sine Wave and then clipped with the object dist/inf.
It works like a charm but the dist/inf uses a lot of %DSP and it dramatically limits the number of instances I can use...
Do you guys have a lighter idea to get a Square Wave from a Phasor?
Thanks in advance!
and you might have to move the "zero" point of the phasor to get a square instead of a pulse..
i hope it is more processor friendly...
OMG YES IT IS!!
It allowed me to double my overall polypony!!!
The aliasing seems to be a bit more hearable (do you confirm?) but that's OK, it's perfect for my need!
Thank you so much my friend!
yes, there is no filtering whatsoever, there will be aliasing. (i don't know what dist inf does in that regard)
note that you also don't need the math/sin object, maybe that could give you some more voices?
The dist/inf does have a bit of aliasing but in a more reasonable range. Anyway, it doesn't bother me more than that. Just for my knowledge, would a simple low pass filtering reduce the aliasing of your object? I guess it's a bit more complicated...
Actually, I didn't manage to get any sound out of your object with only phasor-->squarify. Like you said I need to move the zero point of the phasor to make it works, but I didn't try to modify its code yet. For now, math/sin gives me a perfect duty cycle and doesn't seem to eat that much CPU...
Do you think I'm wrong?
no you are not wrong. but i think you can simply offset the phasor with a -c object or an unipolar to bipolar object to get the same result. but as you said the math/sin is not that expensive either.
try the lowpass after the squarify, it might help a little. but it will not remove aliasing.
)
The low passed ate some %DSP and didn't reduce a single bit of aliasing 
Tried it, both "-c" and "unipolar2bipolar" objects worked and saved some %DSP !
BUT
Something really weird occurs when I play 2 octave notes...
For example :
- if I play a C1 --> I hear a C1 (wow I'm such a good programmer 
)
- if I play a C2 --> I hear a C2
- but if I play C1 + C2 --> I hear C1 + C3 !!!!! 
If I use a math/sin object before your squarify, I don't get this strange phenomenon...
I didn't mention it yet but I use a phase lock architecture, this means that C1, C2, C3, etc are coming from the same phasor that fed the polyphonic object I've showed few posts above.
Any ideas ?
I really would like to replace math/sin with any of these object to gain some %DSP...
It may be a problem with phases.
Maybe you can try changing the -c value.
As you stated, phases relations are crucial when it involves combining octave related notes.
Here's some news!
The problem does not seem to be induced by the phase the square method but by the math/satp included in the filtering block right after. Indeed, if I disconnect it, the problem disappear.
Here's a test bench I made to figure how to solve the problem :
phasor2square.axh (52.8 KB)
You'll be able to select between the 3 methods (math/sin, -c and unipolar2bipolar). Try to play a note and its octave (C2 + C3 for example), everything will be OK with math/sin but you will hear a higher octave with the 2 other methods.
Let me know what you think!
Thanks a lot for you help!