[time-nuts] Re: Is this amount of measurement errors to be excepted when measuring with a small frequency difference?

[Erik Kaashoek]

Hi Tom,
I did the test as you described for measuring the phase difference.
When using a sinus signals there is a gradual shift back and forth over 
1000s in the measured phase residue of the order of 3e-10 but no sudden 
jumps.
When measuring the frequency there is an extremely narrow window in the 
gradual phase shift where the frequency jumps to a value that is 
proportional to the difference in frequency. The problem with measuring 
this is that it is only is really visible with a gate time of less then 
0.1s and most pronounced with a 0.02s gate time and a frequency 
difference of 0.01Hz. A frequency difference of 0.001Hz disappears in 
the noise. Larger frequency differences change so quickly that the 
effect is much shorter than the gate time so it again becomes invisible.

What I learn from this is that I should check if sudden peaks in the 
frequency measurement could originate from a small, but no too small, 
frequency difference and  a matching short gate time. Not a big problem, 
just to be aware of.
An example of these spikes can be found here: 
http://athome.kaashoek.com/time-nuts/Spikes.png
The pink trace has the smallest frequency difference and therefore the 
smallest spikes.
The blue trace has a bigger frequency difference but as soon as I change 
the gate time to 0.1 s the spikes disappear

The variations in the residue of the  measured phase with the shifting 
sinus input phase can be seen in this plot.
http://athome.kaashoek.com/time-nuts/Phase_shift_0.001Hz.png
Blue is the picotest counter with 0.02 s gate time. Maybe it is caused 
by small gradual changes in the effective trigger level.
Pink is my DIY counter which shows much more pulling. Expected because 
part is still discrete logic and long connecting wires.

Thanks for all the help.
Erik.

On 20-7-2022 20:04, Tom Van Baak via time-nuts wrote:
> The test setup that I use is a very stable external 10 MHz reference 
> and a very stable input of 10 MHz plus, say, 1e-10. This creates a 
> growing "calibrated" phase difference of 100 ps per second, so over a 
> run lasting 1000 seconds you have covered an entire 100 ns period of 
> the clocks.