[time-nuts] Re: Best frequency to start for GHz synth ?

[Lux, Jim]

On 4/1/21 9:41 AM, Richard (Rick) Karlquist wrote:
> 1. Whatever the advantage of cryogenic operation, Keysight
> could not consider that due to marketing reasons.  They
> were happy to leave the lunatic fringe to our friends in
> Western Australia.
>
> 2.  They were already limited by microphonics even at
> room temperature, so cryo would be wasted.
>
> 3.  My understanding of the oven was that it was simply
> a substitute for a Q reducing varactor diode tuning mechanism.
> There was never an expectation that temperature compensation would
> allow standalone operation of the sapphire.  Again, for marketing
> reasons, it was always going to be locked to an OCXO at exactly
> 10 MHz.  That OCXO's EFC could be used to lock the OCXO to an
> external reference with a very slow loop that would prevent
> phase noise contamination.
>
> Also of interest to time nuts:
>
> Back when Keysight was part of Agilent, and before the sapphire
> resonator, they had architectures using DRO's at about 8 GHz.
> So any Sapphire replacement had to be in some sense a drop in
> replacement.  Before that, there was a boondoggle project using
> a 1 GHZ DRO with a resonator about the size of a hockey puck.
> It kind of reminds me of the giant 1 MHz quartz resonators.
> There was also a 1 GHZ Surface Transverse Wave Resonator
> (STW) oscillator for a while, until the fab shut down.
> It evolved from the 640 MHz SAW resonator used in early 5071's.
> Again, the fab shut down and I designed it out.

SAW devices were all the rage in the 80s for signal processing. 
Dispersive delay lines (SACs and RACs) were a standard thing in pulse 
compression systems and in "real time spectrum analyzers" - Enormous 
temperature coefficients, so there were all sorts of schemes to have 
multiple devices that would cancel changes out. When ADCs and DACs got 
high enough performance, doing signal processing in the analog domain 
became passe.

Same with electro-optical systems (like real time spectrum analyzers 
using a Bragg cell and a laser) - When DSP got to ~40 dB SNR, all of a 
sudden, optical processing wasn't as interesting.




>
> They also wasted (IMHO) a lot of resources on a "opto-electronic"
> (I think that was the term) oscillator around 15 or 20 years ago.
> It depended on some spool of fiber optic cable.  I never thought
> that was going to work, both because of basic principles and
> because of the cable spool microphonics.

Oohh, I worked on systems using this kind of thing for delay lines and 
coherent analog processing. You can build an oven, sure, but...


And for what it's worth DROs have their microphonic problems too. We had 
a breadboard deep space transponder and you could demodulate your voice 
(poorly) using the spectrum analyzer's FM demod feature. Very cool. 
Unimpressive to the folks who wanted to push the technology.  We were 
trying to build a DRO oscillator that would tune 100 MHz BW (8400-8500 
MHz) while also being able to be locked to a very narrow band source.  
We just couldn't get there because of fundamental physics - couple the 
varactor tightly enough to get tuning range, and Q goes down.
Never did think of changing the temperature to move the "rest 
frequency".. Interesting idea.




>
> Rick N6RK
>
> On 4/1/2021 9:06 AM, Chris Caudle wrote:
>> On 2021-03-31 14:27, Richard (Rick) Karlquist wrote:
>>> When I left Keysight in 2014, they were still trying to solve the
>>> microphonic problem in their sapphire resonator oscillator. Also,
>>> it is still necessary to lock the oscillator to a 5 or 10 MHz
>>> OCXO.  The oscillator is tuned by varying its oven temperature
>>> set ppoint.
>>
>> What kind of oven temperature range?  I thought sapphire oscillator 
>> was pretty much synonymous with "cryogenic sapphire oscillator."Â  I 
>> found a paper which described sapphire as a "low loss material with 
>> loss tangent of 5×10^−6 at room temperature, 2×10^−8 at 77 K and 
>> 7×10^−10 at 4 K giving Q-values of more than >10^7 at low temperatures."
>>
>> That paper seemed to be describing some kind of temperature 
>> compensation they  had developed to reduce the temperature 
>> sensitivity of 10ppm/K and also move the turnover point from 77K to 
>> 92K.  92K isn't exactly what I think of when I hear "oven" so 
>> presumably there is some mode that works at around 300K that I didn't 
>> find discussed yet.
>>
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe 
> send an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.
>