Sequence-of-events records, disturbance records and sampled values are only useful if every device has the same time. A grandmaster that lost GPS, a wrong UTC offset, a rogue device winning the master selection, an NTP server that is itself unsynchronised or a Windows PC polling once every nine hours each put events in the wrong order — usually unnoticed until an incident is investigated.
Standards followed
IEC/IEEE 61850-9-3 (PTP power utility profile: layer 2, peer delay, 1 s messages, domain 0)IEEE 1588-2019 (PTP messages, clock class, best master clock algorithm)RFC 5905 (NTPv4: stratum, leap indicator, offset θ and delay δ)IEC 61850-5 (time performance classes T0–T5)
What it can do
- PTP: grandmaster candidates and the one the best master clock algorithm selects; clock class changes (GPS lost → holdover); an unsynchronised device winning by priority1; currentUtcOffset; timeTraceable; other domains; UDP transport; end-to-end Delay_Req in a peer-delay network; Sync interval and lost Syncs; two-step Syncs without Follow_Up; Pdelay requests not answered; the capture computer's offset from PTP time.
- NTP: each server's stratum, reference, leap indicator and root dispersion; Kiss-o'-Death; unanswered requests; offset θ = ((T2 − T1) + (T3 − T4)) / 2 and delay δ = (T4 − T1) − (T3 − T2) per request.
- w32tm /query /status and /stripchart: source (including local CMOS clock), stratum, root dispersion, poll interval, offsets.
- Time source comparison: every measured or entered offset against classes T0 (10 ms) to T5 (1 µs).
Example problem
The S13 samples: a PTP capture, an NTP capture from HMI1 and its w32tm output.
- PTP: GPS1 goes into holdover (class 7); a device announcing priority1 64 with class 248 would win the selection and announces UTC offset 36 s; one Sync without Follow_Up; a device in domain 93 and one using Delay_Req.
- NTP: HMI1 is 850 ms ahead of GPS1; the second configured server 10.13.1.1 is unsynchronised (stratum 16).
- w32tm: root dispersion 7.8 s and a 9-hour poll interval explain the drift.
Fix the priority1 of the rogue device, remove the unsynchronised NTP server, shorten the Windows poll interval.
Capture PTP on a port of a PTP-aware switch, and NTP on the client computer itself (the offset uses the capture time as the client's receive time). Files are read in your browser.