[sdiy] NCO Jitter (was Large Numbers)

[Richie Burnett]

I agree with Mr Gillet. They are just two different ways of describing the same phenomenon. (Jitter is a description in the time domain and aliasing is a description in the frequency domain.)

Decimation doesn't bring the aliasing back provided that you do a sufficiently sharp lowpass filtering operation *BEFORE* you decrease the sample rate.

FWIW, algorithms like BLIT, MinBLEP, etc. are really just highly optimised ways of arriving at the same result as what you get from massive oversampling, lowpass filtering followed by decimation. But they achieve it through some clever mathematics that avoids having to massively oversample, lowpass filter and decimate every single sample.

As you have already identified it is the sharp transitions and turning points in waveforms that cause the problems, and herein lies the power of the algorithm. It turns out that most of the raw NCO generated sawtooth samples are fine too use just as they are (the ones on the straight line part of the ramp), but a few of the samples either side of the sharp "reset" instants need a bit of tweaking to reduce aliasing...

Sent from my Xperia SP on O2

---- O Gillet wrote ----

>> The BLITs and BLEPs deal with the aliasing you get from sharp edges. I am more concerned with the jitter you get from the mathematical residue of the ratio of F/Fs. In my experience it is an issue mostly with sawtooth waveforms. Triangles and sines seem to have much less trouble.
>
>BLITs and BLEPs deal with both, because they are two facets of the
>same problem (sharp transitions happening in-between samples).