Pete KE9OA wrote:
I want to come up
with a design that tracks the signal all the way down into the noise. I
think that a big part of it will be having a long enough time constant on
the loop filter, so that if a signal fades out, there is enough hysteresis
to "hang on". The tradoff is acquistion time of the system, but I think that
I will be able to find a happy medium.

Ah, you addressed my concern already. I think slow
acquisition time may be entirely acceptable. Question:
would the receiver be mute prior to achieving phase lock, or
would there be some audio recovery as you wait for it to get
an initial phase lock?