[time-nuts] OT: xtal osc PN
magnus at rubidium.dyndns.org
Sat Sep 18 15:01:10 UTC 2010
On 09/18/2010 04:28 PM, Richard (Rick) Karlquist wrote:
> A couple of disclaimers here:
> 1. Leeson's oscillator model was mentioned. That
> doesn't apply much to crystal oscillators. The close
> in noise will be limited by the intrinsic noise of
> the crystal and the far out noise will be limited by
> the buffer amplifier. Leeson's model never comes
> into play. BTW, the intrinsic noise of the crystal
> is higher than the thermal noise corresponding to
> its ESR, at least below 1 kHz or so.
Actually, I did mention the output buffer noise explicitly.
The crystal noise adds ontop of the amplifiers noise, yes... but the
crystal noise will also be feedback to the crystal through the
> 2. Using crystal filters to clean up after the fact
> is limited by the intrinsic noise of the crystals in
> the filters. Hence, this usually only makes sense
> if you multiply a xtal oscillator then filter, as
> in the HP8662, and then it only improves the noise
Essentially only useful as "relatively narrow" filters.
> Unless you have access to very low noise crystals,
> you are not going to get very far building your
> own oscillator.
Agreed. There is a danger in expecting any spare AT-cut crystal laying
around being able to provide the ultra high performance if only
stabilized and "perfect design" of amplifiers. But for the purpose at
hand it may suffice.
More information about the time-nuts