Chain Ring of Class D Boundary Clocks

In real-world telecom deployments, it is typical to have a chain ring of Class D Boundary Clocks, enabling precise and stable Time Synchronization over multiple hops, even in the events of a failover or holdover.
For this test case, we had two Microchip TimeProvider® Telecom Grandmasters (T-GMs ), T-GM -A and T-GM -B, with a 10 nanosecond (ns) offset configured between them. We also had six Boundary Clocks (BCs) set up in a chain ring, three of which are measured by the Calnex Paragon-Neo PAM4 and the Keysight Time Sync Analyzer.
All BCs had local priorities set to T-GM -A, and T-GM -B was therefore used as the backup T-GM.

Initially, T-GM-A provided PTP and SyncE downstream. The measurement was started, and after a short period, the GNSS connection to T-GM -A was cut, causing a Failover of all BCs onto the PTP path of T-GM-B prompted by the change in clockClass from 6 to 7 from T-GM-A.

After 200 seconds, the time measurement was restarted, and once again, after a short period, the GNSS connection to T-GM-B was cut, causing all T-BCs to enter PTP Holdover. The measurement was left running for 200 seconds.
Following this, the time error measurement was restarted once more, and after 30 seconds, the GNSS connection to T-GM -B was restored, causing all T-BCs to lock back onto T-GM-B on PTP after some time. Once all BCs were locked back onto the PTPpath of T-GM-B, the measurement was left running for 300 more seconds.
Finally, the time error measurement was restarted, and after 30 seconds, the GNSS connection to T-GM -A was restored. After the lock-acquisition period, all BCs locked back onto the PTP path of T-GM-A. The measurement was then left running for 300 seconds before being stopped.
   
During all of these procedures, the network passed the requirement defined for clock accuracy class level 6B as per ITU-T G.8271, even during the failover and holdover periods.