|Author||Topic: Making exponential waveform using FFT|
Posted: 2nd March 2004 12:38
I'm working on a exponential waveform, and i'd like to make the wavetables using FFT. Does anybody know what is the difference in freq domain between the sawtooth and the exponential sawtooth (capacitor unload) ?
I made a prototype, it sound really nice, but i'd like to make it bandlimited.
Posted: 2nd March 2004 16:16
wish I could help you, it sounds like a pretty cool thing yer doin' there. Probably get more response in the host/plugin develoment forum though.
Posted: 2nd March 2004 16:52
This link might help...
Posted: 3rd March 2004 01:12
Thanks Summa !
It's a great startpoint.
Posted: 3rd March 2004 02:43
Won't it alias like hell if you use FFT?
Posted: 3rd March 2004 04:17
Nope. It can't.
Posted: 3rd March 2004 06:19
It wont alias if you add f Nyquist/freq harmonics.
But who cares about aliasing... well, i do.
Posted: 3rd March 2004 08:12
As you probably know, the normal saw coefficients are just 1/n for the n-th harmonic. The exp saw is probably messier. You can do the math and find the fourier series coefficients analytically. The waveform would be something like:
1 - exp(-t-pi) t= -pi..0
exp(-t) - exp(-pi) t= 0..pi
Posted: 4th March 2004 01:35
Thanks Tom, I apreciate your reply. However i think that method will create aliasing. Processing a saw with exp(saw) or pow(e,saw) or pow(2,saw) will create zillions of extra harmonics.
But! I created the waveform, and it's bandlimited.
The answer was here : http://www.optimolch.de/jens.groh/K5000/Kenji/digests/phat.txt
You need to add some cosine too to your waveform, to modell that capacity charge.
Thanks for the reply's, to all of you.