[time-nuts] Some thoughts about Crystal Oscillators

[Gerhard Hoffmann]

Am 15.04.2018 um 20:27 schrieb Ulrich Rohde via time-nuts:
> Before I forget it, Bernd Neubig seems to build the worlds best Colpitts circuit 100 MHz oscillator and uses my published thought to use the crystal filter as the noise filter to the next stage. I was never able to get the Driscoll oscillator spurious free. Bernd  Neubig can offer more details .
>

We had a space-proof USO of his, nice. I don't know what's inside.

When it goes below ~ 140dBc / 100MHz / 100 Hz we found that only 10% of 
the crystals were really
good, no matter of Q and extra-extra lapping. I got the idea to build a 
notch array from the
other 90 %. That could be done even at elevated carrier levels that a 
passband crystal filter would
never survive. One just would have to be sure not to tune the oscillator 
over the notches by accident.
I mean, the crystals are there, paid for already and gathering dust.

Their phase noise properties are completely secondary, what matters is 
only series resistance and that
they stay on their frequency.

And 6 dB better is a lot there,  even over a limited range around the 
carrier.
For example, 6 dB less averaging time in an analyzer for the same result.

73, Gerhard