Time synchronization is chosen by the accuracy the application needs. For millisecond event time-stamping, NTP or SNTP over the network or IRIG-B over cable is enough. Sampled values, synchrophasors and traveling-wave fault location need a microsecond or better, which PTP with the power profile or 1PPS provides. IRIG-B remains useful for devices without PTP.
The right question: what accuracy the application needs
No synchronization method is best for everything. Each substation application needs a different accuracy, and the method is chosen from that. IEC 61850-5 organizes it in time performance classes:
| Class | Accuracy | Use |
|---|---|---|
| T1 | ±1 ms | Event time-stamping |
| T2 | ±100 µs | Zero-crossing time-stamping and distributed synchrocheck |
| T3 | ±25 µs | Instrument transformer synchronization |
| T4 | ±4 µs | Instrument transformer synchronization |
| T5 | ±1 µs | Instrument transformer synchronization; the level for sampled values, synchrophasors and traveling waves |
For scale: at 60 Hz, an error of 1 µs equals about 0.02 degrees of angle; an error of 1 ms, more than 21 degrees. Ordering events tolerates a millisecond; comparing current samples from two merging units does not.
The three options
NTP and SNTP. They travel over the existing Ethernet network. The client exchanges messages with the server and estimates the offset assuming the outbound and return trips take the same time. On a substation network they stay in the millisecond range, which meets class T1. They need no extra cabling or special switches. The details are in IRIG-B and SNTP.
IRIG-B. A time code carried on a dedicated cable, coaxial or twisted pair, from the clock to each device. Its accuracy depends on the format (unmodulated, DC level shift, is better than amplitude modulated), the quality of the clock and the cable length. It is the traditional method for relays and recorders, and many devices still require it. Its cost lies in the cabling: one wire per device or per daisy chain of devices.
PTP (IEEE 1588) with the power profile. It travels over the network, but with two differences from NTP: each switch, acting as a transparent clock, measures and corrects the time the frame spends inside it, and each link measures its own delay. The IEC/IEEE 61850-9-3 profile targets an error below 1 µs after several hops. IEEE C37.238-2017 adds time quality information. It requires compatible switches and IEDs. See PTP IEEE 1588 power profile.
1PPS. A one-pulse-per-second signal over cable, marking the exact start of the second without giving the time of day. It is used together with another time source for the most demanding classes on devices without PTP.
Selection table by application
| Application | Accuracy needed | Usual method |
|---|---|---|
| Event and alarm time-stamping (SOE) | Class T1, millisecond | SNTP, NTP or IRIG-B |
| Oscillography records to compare line ends | Millisecond or better | IRIG-B or PTP |
| Synchrophasors (PMU) | Microsecond | PTP, or good-quality IRIG-B with 1PPS, depending on the device |
| Sampled values on a process bus | Class T5, ±1 µs | PTP with the power profile, or 1PPS |
| Line differential with an external time reference | Microsecond | GNSS at each end, with local PTP or IRIG-B |
| Traveling-wave fault location | Sub-microsecond | PTP or the device’s dedicated clock |
The table describes common practice. Each IED manufacturer’s specification states which time inputs it accepts and with what accuracy.
- Does any function need a microsecond or better (SV, PMU, traveling wave)?
- YesDo the IEDs and switches support PTP with the power profile?
- YesPTP IEC/IEEE 61850-9-3 or C37.238-2017
- NoIRIG-B with 1PPS from the clock, wired to each device
- No, only millisecond eventsDo the devices have an IRIG-B input and should time be independent of the network?
- YesIRIG-B over cable
- NoSNTP or NTP over the network
Time distribution topology
- GNSS grandmaster clock in the control house, with its antenna in a place with a clear view of the sky. At high and extra-high voltage it is commonly duplicated.
- Distribution over cable (IRIG-B, 1PPS), with amplifiers or distribution units when there are many devices, paying attention to length and terminations.
- Distribution over the network (NTP, PTP) on the station or process bus. With PTP, every switch in the path must act as a transparent or boundary clock; a switch without PTP in the middle degrades accuracy.
- Network redundancy. PTP works over PRP and HSR with clocks that recognize both paths.
Holdover
If the clock loses the GNSS signal, it continues on its internal oscillator. A plain quartz oscillator drifts faster than a rubidium or an oven-controlled quartz (OCXO). What matters is how long it can keep the accuracy required by the most demanding application. In PTP, the profile requires the clock to announce its change of state through its clock class, so that merging units and IEDs know whether the time is still reliable. The substation time synchronization guide details the classes and alarms.
Common mistakes
- Specifying PTP across the whole substation when there are only millisecond events, with a switch cost that was not needed.
- The opposite: a process bus on SNTP.
- A switch without a transparent clock in the middle of a PTP network.
- A badly placed GNSS antenna, or one without surge protection.
- Time zone and local time set differently in the IEDs, gateway and SCADA, which scrambles the event order. The symptom and its diagnosis are in loss of time synchronization.
Time synchronization with Tiza Group
We configure clocks, PTP profiles, IRIG-B and SNTP in IEDs, switches and gateways, and we test accuracy and holdover at commissioning, as part of our IEC 61850 engineering service, at medium, high and extra-high voltage, for utilities and industry, in Colombia, or elsewhere on request. If the problem is the time configuration of a relay, we review it worldwide through remote relay configuration. Write to us with the applications in your substation.
Related guides
- Substation communications by voltage level: MV, HV, EHV
- IEC 61850 transfer time: classes and measurement
- Sampled Values and GOOSE bandwidth: how to estimate it
- Substation GNSS clock: holdover, redundancy and risks
References
- IEC 61850-5: communication requirements, time performance classes.
- IEC/IEEE 61850-9-3:2016: PTP profile for power utility automation.
- IEEE C37.238-2017: PTP profile for power system applications.
- IEEE 1588: Precision Time Protocol.
- IEC/IEEE 60255-118-1: synchrophasor measurements.
- IRIG Standard 200: time code formats.
Frequently asked questions
Which is more accurate, PTP or NTP?
PTP with the power profile, because the switches measure and correct the time the frame spends inside them and each link measures its own delay; that is how it gets below one microsecond. NTP and SNTP do not correct the delay inside the switches and, on a local network, stay in the millisecond range: enough for events but not for sampled values.
Can IRIG-B be replaced by PTP in an existing substation?
Yes, if the IEDs support PTP with the right profile and the switches act as transparent clocks. During the transition it is common for the clock to deliver PTP over the network and IRIG-B over cable to the older devices, or to use PTP-to-IRIG-B converters.
What happens if the clock loses the GNSS signal?
The clock enters holdover and continues on its internal oscillator, which drifts over time. The accuracy it keeps depends on the oscillator. In PTP the clock changes its clock class to warn the devices; if it goes out of specification, functions that depend on the microsecond may be blocked.