posted 26 January 2002 00:41            

voice_mod_fm_drone.kym

The system is currently set up with an input test signal containing a fundamental, a second harmonic at -13 dB, and 1/F noise. The ratio of noise and signal can be varied.

But the improvement in this circuit comes from a phase detector implemented as the arctangent of the sine and cosine amplitudes, running through a differentiator, and then into an integrator. This integrator is multiplied by a constant (in this case a nn range) and added to a base nn to derive the frequency used by the sine and cosine antinoise oscillators. The constant is effectively the lock range of the phase locked loop. It must be a negative value.

This system manages to track and kill the fundamental of the input signal over a 3 octave range when the signal to noise ratio is decent (1:1 or better). But interestingly it manages to track and null out the fundamental even when the signal to noise ratio is as low as -30 dB, although the frequency lock-up range is somewhat restricted. I have managed to null out the fundamental over at least 1 octave range under these poor signal to noise conditions.

This is only a first attempt at a phase locked loop in Kyma, but it performs remarkably well.