[time-nuts] Close-in phase noise question...more info...

[John Miles]

> Maybe I'm missing the decimal point.  What's the bandwidth of a crystal
> filter relative to the spectrum out of the same crystal used as an osc?
>

The oscillator's noise floor itself is superb, near -180 dBc/Hz beyond 1
kHz.  That is only true until you do something with it.  The idea behind the
crystal filters is to get rid of the broadband noise introduced by the
buffers and multipliers.

You don't put the crystal directly in series with the oscillator, but rather
after various multiplier stages.  See the HP note "Generation of low phase
noise microwave signals" at
http://www.ke5fx.com/Scherer_Low_PN_Signal_Generation.pdf , especially page
23.

The additional filtering won't necessarily solve Brian's problem of
minimizing phase noise at circa 1 KHz from a 630 GHz carrier, though.  That
is beyond anything I have tinkered with.

-- john, KE5FX