[time-nuts] Re: E1938A phase noise improvement

[Richard (Rick) Karlquist]

On 12/26/2021 10:42 AM, Tom Van Baak wrote:
> The correct link is:
> 
> https://febo.com/pipermail/time-nuts_lists.febo.com/2012-September/053009.html 
> 
> 
> /tvb
> 

I think what I said here is self explanatory.
The loop amplifier or its resistors have noise that corrupts the AFC 
loop.  Or else the loop bandwidth or damping factor is not optimum.
I would say many readers of time nuts would be capable of optimizing
the AFC.  The fix should be able to be kludged on the existing PC board.

Regarding Attila's question about the bridge stabilization:

I started with the 10811 by spraying coolant from a can on the
2N5179 transistor.  The frequency is extremely sensitive to the
temperature of this transistor.  It is also extremely sensitive
to humidity.  The bridge stabilized circuit improves this by
many orders of magnitude, as can be seen if you spray the
oscillator transistor in the E1938A.  The bridge stabilized
circuit doesn't add any significant cost to the oscillator.
The zero gradient oven achieved thermal gains as high as
2 million in hero experiments, but that was only for the
resonator.  Its thermal gain would be considerably less
for the entire oscillator circuit.  I never tried disabling
the AFC in a working E1983A and then testing its frequency
stability over temperature.  Maybe I would have been pleasantly
surprised.  Again, many readers of this list could do this
experiment if curious.

It is worth considering that a typical E1938A has better temperature
stability than a 5061B.  That's a fairly non-trivial feat.
The double integrator in the oven control loop has been optimized
to provide extremely good transient temperature control to the crystal.
I'm not sure that the whole oscillator circuit could be made to have
such transient performance.

Thanks to everyone for their comments.

Rick N6RK