AW: Fast VCOs/V->F converters

[Haible Juergen]

	>>I share your doubts about PLLs, for the reason Martin and others
	>>mentioned earlier: No clean transition between notes. But I don't
	>>understand "never quite getting there" - once it's locked, it's
locked
	>>in *phase*, which implies being locked in frequency, too. (In
theory,
	>>you might have an infinite response on pitch changes in the PLL,
but
	>>in reality it will be really locked after some time.)
	>>
	>
	>Not true.  There is always a small amount of phase noise in the
loop.  The
	>tighter the loop bandwidth the smaller the noise, but it still is
there and
	>it shows up as phase jitter.  If you clamp down too much on the
loop
	>bandwidth to get acceptable jitter performance then the transient
response
	>of the loop is way too slow for making note to note transitions
(portamento
	>anyone?).  Whether the jitter is acceptable or not is application
dependent,
	>but it is there.

Hmm, are we speaking of noise from your input signal, or from parts
of the PLL ?

Parts of the PLL can be more or less noisy. You always have a certain
amount of noise in any system - the questions here are will it be noticable,
and will it be much higher than on a "straight" VCO without the loop.

The PLL also acts as a filter for noise from the input signal (i.e. the
frequency you want to track), and is in fact often used to clean noisy
carriers. Apart from the frequency dividers in the loop, it's just a sort
of (auto tracking) BP filter, after all.

It's clear that one can build a PLL (think of a DPLL with high quantisation
error and low bandwidth) where the internal noise is so high that it will
never lock - but is this valid for the musical applications we were
talking about ?

JH.