On Fri, Jan 13, 2012 at 2:42 PM, Jim Lux <[email protected]> wrote: > On 1/13/12 2:33 PM, Chris Albertson wrote: >> >> On Fri, Jan 13, 2012 at 2:19 PM,<[email protected]> wrote: >>> >>> Are we not overdoing it. 7E-13 for $15 KISS, and if you want to go beyond >>> that use a $ 30 analog version. >>> Looking right now at 1 E -14, I do se temperature influences, with my >>> heat >>> sink stable to .1 C. Mainly due to the fact that is only one side of the >>> unit and it is clearly designed to use all sides for heat dissipation. >> >> >> Technical question, how does one measure 1E-14 changes in frequency? >> >> I'm thinking about building a controller and of course you can't >> control what you can't measure. But at that small level what do you >> use for a reference? >> >> > > Your other sources. Seriously.. > Of course, the man who has two clocks doesn't know what time it is. > > If you have 3 sources, you can do three cornered hat type measurement > comparisons, for instance. > > Typically, you might have clocks of various kinds. For instance, a nice > quiet quartz oscillator has great phase noise, but does drift with time and > temperature. So you could use that to measure the phase noise of something > with not so hot performance.
OK, in general I understand. But now we are talking about a $15 controller that needs to make a decision wetter to increment the DDS by 7E-13 or not. What is this controller looking at? People are already asking if people can write software and making PCB designs but first one needs to know basic things like: (1) what to use as a reference? (2) what kind of phase detector has the sensitivity to enable decisions at the e-13 level or e-14 level. (3) What time constants and loop gains would be reasonable (4) does the controller need input for temperature? (5) should the controller look at the saw tooth correction from the GPS (6) seems that a Kalman filter would help? Seems it could if you have a good temperature and aging model. Lots more I'm sure but basic requirements are needed before any design can begin. Hence my plan to use a desktop computer as the controller. I suspect the first few prototype solutions will not work well. Chris Albertson Redondo Beach, California _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
