[time-nuts] State of the art of crystal oscillator measurements

[Richard (Rick) Karlquist]

On 8/11/2016 3:47 PM, John Miles wrote:
> Right, I'm speaking specifically of L(f).  The device being driven by the oscillator doesn't care about the NF of the driver stage, only what a PN analyzer would measure at the output jack.
>
> For any 50-ohm source, the practical L(f) floor is -177 dBm/Hz - the carrier power in dBm.  No oscillator with an output of 0 dBm can be quieter than -177 dBc/Hz at any offset, but an oscillator that puts out +20 dBm could approach -197 dBc/Hz.
>
> Given a proverbial black box containing a +17 dBm oscillator that measures -195 dBc/Hz at 100 kHz, the interesting question is, "What's in the box?"  There could be a passive resonator that's shaving off the broadband noise after the last active stage without contributing additive noise of its own.  Another possibility might be cross-spectral collapse due to correlated thermal noise from the splitter.
>
> -- john, KE5FX
> Miles Design LLC
>

If the oscillator output is thru the resonator, then at large offsets,
the source impedance is reactive.  It can easily have an effective
temperature less than room temperature.  If this "source" is then
used with a low noise temperature preamp, it is entirely possible
to go beyond these supposed theoretical limits that are based
on T=300K.

Rick