[time-nuts] Frequency standards for different tau in Allen Dev measurement

[Bob kb8tq]

Hi

For most good standards, all a frequency counter is measuring at a 1 second gate is the
floor of the counter:

A typical counter might have a 100 ps rms sort of “gate error” at 1 second. That means that
the best it actually can do is 100 ppt = 1x10^-10. A good oscillator will have an RMS count
to count error of around  10 to 100X better than that.

If you set the counter to a 100 second gate time, it still has the same 100 ps sort of error.
Now that error is at 1x10^-12. The error in the counter and the 100X better “good oscillator” 
error now are roughly the same. 

In any measurement setup, you would prefer your instrument errors to be 5 to 10X *less*
than what you are trying to measure. For a 1x10^-12 measurement, you would like the counter
to contribute < 2x10^-13 and better yet, to contribute < 1x10^-13. 

To hit a 1x10^-13 sort of “contribution” your 100 ps counter would need to be set to a 1,000
second gate time. 

Yes, this assumes that the reference to the counter good enough to stay out of the way at
all these tau’s. If the reference contributes, then that also needs to be figured in to the mix.

If your counter is in the 400 ps range, that makes things more difficult. If it is in the 20 ps 
range that makes it a bit easier. 

There is a bit of hand waving in the above analysis simply to keep it simple. Despite that
the basic conclusions are pretty darn close.

=====

So if you *do* want to accurately measure <1x10^-12 at a 1 second tau, what counter do 
you use? Well, you can’t do it with any counter that I am aware of. You need to use a 
different device or a different technique. The DMTD is one of several approaches that
could be used. 

http://www.wriley.com/A%20Small%20DMTD%20System.pdf <http://www.wriley.com/A%20Small%20DMTD%20System.pdf>

https://white-rabbit.web.cern.ch/documents/DDMTD_for_Sub-ns_Synchronization.pdf <https://white-rabbit.web.cern.ch/documents/DDMTD_for_Sub-ns_Synchronization.pdf>

https://www.osti.gov/servlets/purl/1494164 <https://www.osti.gov/servlets/purl/1494164>

A Phase Station is another way to get the job done:

http://jackson-labs.com/index.php/products/phasestation_signal_source_analyzer <http://jackson-labs.com/index.php/products/phasestation_signal_source_analyzer>

A TimePod / 3120 also would do the trick:

https://www.microsemi.com/product-directory/phase-noise-and-allen-deviation-testers/4131-3120a <https://www.microsemi.com/product-directory/phase-noise-and-allen-deviation-testers/4131-3120a>

Even with one of those fancy devices, you still need a *reference* that is “good enough” 
to do your measurement. All any of the devices above do is compare sources. With multiple
sources, you can do some fancy math. The pitfalls of doing that have been mentioned in 
a number of posts here on the list.

The basics:

http://www.wriley.com/3-CornHat.htm <http://www.wriley.com/3-CornHat.htm>

http://www.miles.io/PhaseStation_53100A_AN_53100A_2.pdf <http://www.miles.io/PhaseStation_53100A_AN_53100A_2.pdf>

And off the deep end :) :

https://www.researchgate.net/publication/314137002_Three-cornered_hat_versus_allan_covariance <https://www.researchgate.net/publication/314137002_Three-cornered_hat_versus_allan_covariance>

(and on and on …..)

Bob



> On Feb 20, 2020, at 1:40 PM, Taka Kamiya via time-nuts <time-nuts at lists.febo.com> wrote:
> 
> I have a question concerning frequency standard and their Allen deviation.  (to measure Allen Dev in frequency mode using TimeLab)
> 
> It is commonly said that for shorter tau measurement, I'd need OCXO because it's short tau jitter is superior to just about anything else.  Also, it is said that for longer tau measurement, I'd need something like Rb or Cs which has superior stability over longer term.
> Here's the question part.  A frequency counter that measures DUT basically puts out a reading every second during the measurement.  When TimeLab is well into 1000s or so, it is still reading every second; it does not change the gate time to say, 1000s.
> That being the case, why this consensus of what time source to use for what tau?
> I recall reading on TICC, in time interval mode, anything that's reasonably good is good enough.  I'm aware TI mode and Freq mode is entirely different, but it is the same in fact that measurement is made for very short time span AT A TIME.
> I'm still trying to wrap my small head around this.  
> 
> --------------------------------------- 
> (Mr.) Taka Kamiya
> KB4EMF / ex JF2DKG
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.