hello,

i have a question. if i use a bandlimited sawtooth wave oscillator to make another waveforms, these latter will be anti aliased too ?

thanks

I guess that heavily depends on what you do with the wave. If you chop it up violently it will probably generate new higher frequencies which will fold back at the nyquist-frequency - at least as far as I know.

i was thinking about making a triangle wave or square wave from a bandlimited sawtooth wave.

It really depends on what you do to it. If you just sum two waveforms together (which you can do to get a square wave from a sawtooth), they will still be band-limited. I'm not sure how you would get a triangle wave from a sawtooth. If you're thinking of just playing it forward-reverse, I don't think you'll end up with what you're expecting.

if i use a bandlimited sawtooth to make a square wave, can i choose duty cycle percentage without making aliasing ?
same question with triangle wave : if i change its slopes (reverse saw/tri/saw) aliasing will occur ?

@pouk said:

if i use a bandlimited sawtooth to make a square wave, can i choose duty cycle percentage without making aliasing ?

Yes. To make a band-limited square wave, you take a band-limited saw wave and subtract another band-limited saw wave that is offset by the length of the duty cycle. Subtraction is just an inverted addition. Just think about it this way: whenever you are mixing tracks, you're adding them together. There is no aliasing as a result of that.

same question with triangle wave : if i change its slopes (reverse saw/tri/saw) aliasing will occur ?

This, again, depends on how you are generating the triangle wave. I don't know how to make a band-limited triangle wave with sawtooth waves. The ones I made sum band-limited parabolic waves, but it doesn't really result in "true" triangle waves, just reasonable approximations.

You can also make band-limited triangle waves by running a band-limited square wave through a leaky integrator, such as [rpole~ .99], and doing some amplitude scaling (because the integrator itself makes for a really loud triangle wave). But that has it's own problems, one of them being that it takes some time to settle when the frequency is changed, and there can also be some DC drift that needs to be taken care of. I don't really like it.

https://ccrma.stanford.edu/~stilti/papers/blit.pdf

i try to make a bandlimited triangle wave with slope control in order to choose between sawtooth and triangle. i' ve got some errors, a part of the signal stay without
being changed by the slope control. so the result is a triangle wave preceded and followed by a vertical line...
here the patch

http://www.pdpatchrepo.info/hurleur/triangleAliasSlpe.pd

I just took a short look, but did you meet the [tabread4~]s table-conditions??

What you're doing here is switching between two inverted saw waves of different amplitudes. When the switch occurs, there is a sharp corner in the wave. These corners result in an infinite spectrum at that time and will cause aliasing.

In addition, the up and down sections of the ramp may contain frequencies of the same waveform, but they are at different amplitudes and they are inverted. If you were to split these into individual components, you would see sine waves that suddenly change amplitude and invert at the same time. At that point, each wave will have a jump, like you see at the end of [phasor~]. The jumps will cause aliasing, and the amplitude modulations will create additional distortions.

ok it' s out of my skill...maybe one day
so i give up for the moment...
Flipp i don' t understand, what is table condition ?

thanks

I just mean the [tabread4~]s requirements concerning the table it reads from.. Have a look at [tabread4~]s and maybe [tabosc4~]s helpfile..