> And for a purely hypothetical practical example, let's say that you
wanted
> to design a FM Broadcast transmitter kit which covered from 88 to 108
> MHz.......... :-)


> The two signals are then compared in a phase/frequency comparator,
which
> generates an error voltage which steers the VCO onto exactly the right
> frequency (and phase).
>
> The only tricky bit is getting the time constants of the control loop
> correct. The maths of this can get nasty. It is basically a PID, with
just
> the right amount of leading and lagging phase to overcome the
mechanical and
> thermal instabilities in the VCO.
> Most people cheat and tweak the values empirically.

There are a few other considerations if you want to use this for
broadcast audio. Essentially you have a conflict between loop settling
time and audio quality. Too short a time and the thing risks unlocking
on bass audio signals.  Many professional transmitters (and even a few
of the "unofficial ones") use a dual time constant which runs fast when
the VCO is a long way off frequency but switches to a very slow rate
when it is close.
Think 4066...




Cheers,

Andy.