Post #2542771
2026-04-12 12:16 UTC
The problem is that my "genius" FM demodulation algorithm isn't so genius after all, at least with the sample rate involved, where I get 7-ish samples per period of the carrier.
The algorithm measures the distances between zero crossings using inverse linear interpolation, to (try to) directly get the inverse value of the frequency. And it turns out that linear interpolation doesn't approximate a sine wave well enough with this few samples. This manifests as some rather nasty noise:
Replies (1)
-
@grishka@mastodon.social 2026-04-12 12:17
The only reason my video samples from this thing didn't look as nasty is that I was putting this through a low-pass filter that mostly smoothed it all out. It's also much less noticeable on signals coming from a video tape in the first place.