[music-dsp] Paper on bandlimited analog waveform synthesis usingFM
antti at smartelectronix.com
antti at smartelectronix.com
Wed Jan 17 18:25:18 EST 2007
On Wed, 17 Jan 2007, James Chandler Jr wrote:
> Well, one thing to consider poking a probe on a circuit board-- if the probe
> isn't high-impedance and properly compensated, the picture you see on the
> scope can be 'deceiving'. I think one would have the same trouble recording
> to audio off some point inside the circuit board. Not rocket science, just
> something that would need to be done right. Maybe use a 10 meg 10X scope
> probe somehow interfaced to an audio preamp?
You should tap the output buffer of the oscillator core. If you tap the
actual timing capacitor, you Will influence the oscillator (at least
frequency and possibly waveshape).
> Anyway, am pretty sure my scope probe was usually well-compensated, and
> various artifacts were not uncommon on real-world oscillators-- Sometimes
> spikey switching artifacts on wave transitions. Sometimes more slew than you
> might expect on a wave-transition which 'ought to be' instant. Sometimes
I've simulated finite slew rate of oscillator and I can't say I heard any
difference at all. Nonetheless, my Moog Taurus VST does simulate that, so
you're free to try it.
> slight curves where there are supposed to be straight lines. Square waves
> that don't have 50% duty cycle, and perhaps have different artifacts on the
Non-50% square waves are quite common and there is significant difference
to sound (just try any digital synth with 49% and 50% pulse wave). This is
due to the comparator threshold not being at exactly 50% of sawtooth
amplitude.
> Since in digital, aliasing wants to be the travelling companion with sudden
> transitions, am not sure how you could add some of this 'imperfect analog
> character' to digital oscs without inviting aliasing. Maybe you could get the
It's not actually that difficult to simulate the various artifacts. I'll
leave the implementation to the reader ;). You do have to be careful to
simulate the right artifacts of course.
Antti
More information about the music-dsp
mailing list