Rigol dg1022z clock accuracy and stability measured using LBE-1420 GPSdo

[bb1]

Rigol dg1022z is waveform generator and frequency counter.
Both the generator and the frequency counter use the same clock.

Rigol datasheet specifies the clock accuracy as "±1ppm of the settings, 18℃ to 28℃".
The actual accuracy can be measured by providing GPSdo 10MHz signal to the input of the dg1022z counter.
In my case the measurement result was 10.0000030 MHz, which means 0.3 ppm error.

It also was noticed that this error remains unchanged in time.
This means that it can be removed by calibration.

Rigol provides dg1022z calibration instructions in the file "DG1000Z Calibration Manual.pdf".
These instructions suggest using more accurate counter to make adjustments to dg1022z clock.

Instead I compared dg1022z output frequency with the 10 MHz GPSdo output on an oscilloscope screen.
I used dg1022z up and down soft calibration buttons to stop dg1022z waveform moving relative to GPSdo waveform.

The scope screenshot of the result is shown on the attached rscope_10secPersist_LEO_dg10227_10ns.png.
GPSdo 10 Mhz signal is shown by yellow trace, and dg1022z waveform is shown by blue.
The scope persistance was set to 10 sec.
We see that the dg1022z signal moved by less than 10 ns in 10 sec,
which means dg1022z calibrated clock is accurate within about 1 part per billion (1ppb).

Longer time stability can be checked by using PC to capture array of the dg1022z counter values with 10 Mhz input from the GPSdo.
As dg1022z counter makes most accurate frequency measurements with Gate Time of 10.73 sec, the PC captured the counted frequency each 12 seconds,
which means 300 measuremnts per hour. Two attached plots show measured frequency values as a function of time.
Frequency values on the plots are shown by filled circles. The connecting line is just an eye aid.

First hour measurements starting with cold dg1022z show measured frequency drop by about 8 ppb at
freqDrift_firstHour.png with maximum error of about 9 ppb.

Frequency drift for the next 2 hours, visible at freqDrift_secondAndThirsHour.png, shows dg1022z clock frequency fluctuations of about +-1 ppb.
Here the average frequency error is less than 1 ppb.

Using more data, I would estimate the dg1022z clock uncertainty to be within 10 ppb.

[bb1]

scope screenshot