[time-nuts] Oscillators and Ovens

[Richard (Rick) Karlquist]

On 11/1/2017 3:44 PM, Dana Whitlow wrote:
> Bob,
> 
> This discussion is getting really interesting.  In thinking about the
> crystal Q versus
> tuning range conundrum, two (presumably-overlapping) concerns come to mind:
> 
> 1. The motional parameters of a high-Q crystal are such that the external
> network
>      needed to pull it very far would be wholly impractical.
> 
> 2. Varactors themselves probably have pretty limited Q over much of their
> range.
> 
> Is my thinking on the right track at all?
> 
> Dana  K8YUM
> 


Sorry your thinking is NOT on the right track.

What determines the pullability of a crystal is the ratio of the
motional capacitance to the static capacitance, commonly denoted
as C1/C0.  The Q of the crystal has nothing to do with it.  The
only thing significant about the Q is that it limits the
how QUICKLY you can change the crystal frequency.

What determines the noise of a crystal is the intrinsic
flicker of frequency noise.  The Q has nothing to do with it.
If the Q is degraded somewhat by adding varactors to pull the
frequency, it doesn't affect the noise.  It is true that if
varactors are used, it is possible that the noise will be
degraded if the tuning voltage is not clean enough.  The
HP smart clocks were always limited by this problem because
no realizable voltage source was good enough, at least 20
years ago.

In the 5071, I modified the 10811 to increase its tuning range
by an order of magnitude.  This did not affect its noise
at all, AKAIK.  The zener diode reference in the 10811 is
actually quite good.  This modification was done to eliminate
the need to tweak the coarse tuning of the 10811 as it aged.

Having said this, with currently available technology, I recommend
using frequency synthesizers to do a "virtual pull"
on crystal oscillators, rather than trying to pull them
with varactors.

Rick N6RK