This may be the trivialest post of the year.

After a frustrating hour of debug I found that the sin~ abstraction in Pd extended can't compute a reasonable sine for very small signals. That is probably due to the small costable and the interpolation method. It helps to compute sin~ another way, as:

cos(phase+cycle/4) * -1

Decided to post this for fellow nit-pickers.

Katja

see attached patch for comparison.

http://www.pdpatchrepo.info/hurleur/twoWaysForSin~.pd

Or [expr~ sin($v1)]. That one seems to give you exactly what [sin] does.

Maelstorm, you're right of course. [expr~] and [sin] both call the FPU sin() instruction, which does 80 bits precision in a fair amount of clock cycles.

Katja

http://www.pdpatchrepo.info/hurleur/threeWaysForSin~.pd