Tiza Group

Home / Tools / Time synchronisation

Time synchronisation: PTP, IRIG-B and NTP

Choose how the time reaches the IED and for which function. The tool draws the chain from the antenna to the IED, adds up the error of each device, compares it with the accuracy the function requires and computes how long it holds after losing GPS.

Example cases
Time synchronisation: PTP, IRIG-B and NTP

Only the basic data; the rest is taken with typical values.

ALearn

E1Function and method
µs
Hz
E2PTP
ns

Estimate: use the figure of your manufacturer.

ns
ns
ns
ns
E3IRIG-B
m
ns

Estimate: use the figure of your clock.

µs

Estimate: use the figure of your manufacturer. With modulated IRIG-B (B12x) the error is much larger than with B00x.

E4NTP
ms

The one the NTP client measures.

µs

Estimate: time stamping of the IED and drift of its clock between polls.

µs

Estimate: use the figure of your server.

E5Loss of GPS
ns/s

Linear estimate: holdover figure of the manufacturer (1 ns/s = 1 ppb); temperature and ageing make it grow.

s

Use a decimal point. The calculation runs in your browser and updates as you type.

Total error against the requirement

—

Best class reached

—

Holdover to the limit

—

PTP network budget

—

From the GPS clock to the IED: error of each device
Above, the devices the time passes through with the error each one adds. Below, the accumulated error on a logarithmic scale: in red, where it already exceeds what the function requires.
Error after losing GPS
Without GPS the clock drifts and the error grows. The red dot marks when it exceeds the limit.

Error budget

What it is for

Each function requires a different time accuracy: an event recorder is satisfied with one millisecond and a merging unit needs one microsecond. This tool adds up the error of each device between the GPS clock and the IED, tells which class the chain reaches and how long it holds when the satellite signal is lost.

How it is calculated

Source: IEC/IEEE 61850-9-3:2016, cl. 7

PTP

e = eGM + N · eTC or BC + Δasym / 2 + eIED

Worst case: the errors add up. The profile aims at less than 1 µs after about 15 transparent clocks or 3 boundary clocks.

Source: IRIG Standard 200

IRIG-B

e = eclock + L / (Vp · c) + edecoding

If the IED compensates the cable delay, the cable term disappears.

Source: RFC 5905

NTP

e ≤ eserver + δ / 2 + eclient

δ is the round-trip delay the client measures.

Source: own derivation

Holdover and synchrophasors

tholdover = (erequired − e) / drift; tTVE 1 % = 2 · asin(0.005) / (2π f)

At 60 Hz, a time error of 26.5 µs already gives 1 % TVE even if the magnitude is perfect.

Limits of the calculation

  • It adds up the errors in the worst case; in practice part of them cancel out.
  • The oscillator drift is taken as constant; a real oscillator changes with temperature and ageing.
  • It does not model redundant grandmasters or the change of master.
  • It does not replace the measurement of the time in your devices. The example cases are illustrative.

Frequently asked questions

Why is NTP not suitable for sampled values?

Because its error is bounded by half the round-trip delay of the network, which on a LAN is tenths of a millisecond. Merging units need 1 µs: only PTP with the power profile or a 1PPS signal reaches it.

How long does a merging unit hold without GPS?

As long as its oscillator takes to drift from the current error to the limit. With 450 ns of margin and an oscillator drifting 10 ns per second, that is 45 s. The holdover figure is declared by the merging unit manufacturer.

Should the IRIG-B cable be compensated?

For events, no: a few hundred nanoseconds do not matter against 1 ms. For synchrophasors or line differential, yes: 150 m of coaxial already delay about 0.76 µs.

Related guides

Other tools

All tools
WhatsApp