WRS Low Jitter board measurement results
- Additional hardware can improve jitter by two orders of magnitude -
Chantal van Tour and Jeroen Koelemeij have made new measurements of the performance of the White Rabbit switch [1] with Mattia Rizzi’s additional Low jitter daughterboard (LJD) [2] integrated and enhanced even more with another clean-up oscillator. They describe their incredibly good results in the article “Sub -Nanosecond Time Accuracy and Frequency Distribution” [3]:
??The LJD improves the stability by about a factor of five, in good
agreement with results earlier obtained by Rizzi et al. The LJD
implementation improves the root mean-square (rms) phase jitter,
integrated over the range 1 Hz – 100 kHz, from 13 ps to 2.9 ps.
…
While the LJD implementation already significantly improves the default
WR Slave phase noise in the low-frequency range, the PLO (clean-up
oscillator) leads to a further suppression of noise by 30 to 40 dB over
the frequency range 10 Hz – 100 kHz… As a result, the rms jitter of
the 10 MHz output is as low as 0.18 ps over the range 1 Hz – 100 kHz, an
improvement by nearly two orders of magnitude over default WR.
…
The results presented here suggest that the improved WR system may in
certain cases provide a good alternative for hydrogen masers. This may
be true in particular for situations in which the observational phase
coherence is limited by atmospheric conditions, rather than by
local-oscillator stability. Another advantage of a WR-based frequency
distribution system is that it also distributes phase. In other words,
all nodes in the WR network will have the same phase, which may allow
reducing the ‘search window’ in initial fringe searches.??
Note that the results with the additional clean-up oscillator require the White Rabbit Grandmaster (GM) to be locked to a sufficiently stable oscillator, such as a rubidium or cesium clock, or a hydrogen maser. The clean-up oscillator will not be able to track the free-running clock of a WR master, which is too unstable. With only the Low jitter daughterboard it will work fine on a free-running GM.
We will discuss on how to proceed to make this hardware available via commercial partners, likely as special low-jitter versions of the White Rabbit Switch.
[1] https://www.ohwr.org/project/wr-switch-hw/wiki
[2] https://www.ohwr.org/project/wrs-low-jitter/wiki
[3] https://library.nrao.edu/public/memos/ngvla/NGVLA_22.pdf