Synchronization networks in deployment often include chains of Telecom Boundary Clocks (T-BCs), each with established device and network time-error specifications.
This test case assessed the performance of such a chain based on the number of class D T-BCs it contained.
Class D T-BCs have only a single metric, max|TE_L|, defined, for which the accumulation is not specified. max|TE_L| may be considered as all constant Time Error (cTE), all dynamic Time Error low-pass filtered (dTE_L), or a combination of the two. Considering all of the time error as cTE gives a limit of  N · max|TE_L|, i.e.,  N · 5ns, with  N being the number of T-BCs in the chain, since the full allowance is linearly summed.
If considered only as dTE_L, the limit could be calculated as  √(N ∙ max|TE_L|²), i.e,  √(N ∙ 5ns²).

In this test, the Microchip TimeProvider 4500 acted as the Telecom Grandmaster (T-GM), and eleven Class D T-BCs were included. Based on the formula, treating all time error as cTE, the limit was 55 ns.
The T-GM distributed PTP and SyncE downstream, with all T-BCs locked. Each T-BC was individually connected to a frequency/phase analyzer, either the Calnex Paragon-neo, Paragon neo-A, or Keysight Time Sync Analyzer.
This approach allowed us to observe both the total time error at the end of the chain and the time error contributed by each T-BC.
At the end of the chain, we measured a max|TE_L| of 52.313 ns, meaning that we were within the above-mentioned limit of 55 ns.
Additionally, we measured the relative cTE between T-BC-1 and T-BC-11 (36.981 ns), as well as the relative low-pass-filtered dynamic Time Error (dTE_L) between T-BC-1 and T-BC-11 (4.397 ns).
Analysing the 37 ns of cTE relative to T-BC-1 and T-BC-11 results in an average of around 3.5ns of cTE per T-BC in the chain. As per ITU-T G.8273.2, a class D T-BC is given a max|TE_L| of 5 ns; since max|TE_L| is the sum of cTE and dTE_L, that would leave us with around 1.5 ns of dTE_L per T-BC. The accumulated dTE_L would then be, according to the formula above, √(11 ∙ 1.5 ns²) = 4.97 ns. Comparing this to the relative dTE_L measured between T-BC-1 and T-BC-11 (4.4 ns), we are within that limit. Combining our measured cTE and dTE yields 42 ns of max|TE_L| throughout the chain. An additional TE budget of around 10 ns can be attributed to a time-error contribution from the T-GM,  resulting in the 52.313 ns measured at the end of the chain.

Table 76: Time Error Accumulation and Calculation for Class D Telecom Boundary Clocks - Devices Under Test