Measuring Jitter

[colorado.rob]

I am trying to measure the jitter of two digital signals.  I am measuring the period and capturing statistics as suggested here:

https://www.eevblog.com/forum/projects/is-my-jitter-measurement-method-correct/msg2299848/#msg2299848

The problem I am facing is that the statistics vary considerably, and especially when changing the time base on the 'scope.  I've tried enabling gating of the measurement and changing the gate period.  I cannot get a consistent result.  This is with a Siglent SDS2204X, using 2 channels (1&3) to get maximum accuracy, 2Gsps.  I am measuring a 28MHz signal and expecting to measure jitter in the 1.2ns to 200ps range.

Why does the jitter measurement vary so much based on the timebase?  What additional information would be helpful in diagnosing this and allowing me to get an accurate measurement of jitter?

[Kleinstein]

The time base can effect the scopes own jitter. Depending on the method used the trigger uncertainty can be quite a problem with the measurement. So it really depends on the detail of the implementation in the scope and how the measurement is made.
Also the type of signal can be important - a 28 MHz sine wave is different from a square wave.

[colorado.rob]

Thanks.  This is an attempt to measure two square waves generated by a Xilinx 7z020 using LVCMOS33 signalling into a PMOD port.  I suspect I need to work on proper termination to clean up the signal being measured.

One signal is a noisy fractionally divided signal.  This is fed back into a PLL on the FPGA.. The other signal is the output from the PLL.  The first has a jitter of at least 1.25ns as it is being generated by an 800Msps SERDES output.  The seconds has jitter which I suspect is about 200ps.  The signal that has the widest variance is the one with at least 1.25ns jitter.  It sometimes shows as low as 250ps, which is clearly not correct when viewing the trace on the scope.  This makes me suspect both measurements.