Precision Time Protocol

What is Precision Time Protocol?

Precision Time Protocol (PTP), defined in the IEEE 1588 standard, is a network protocol that synchronises the clocks of devices on an Ethernet network to within microseconds or nanoseconds by exchanging time stamps and measuring and correcting the transit delay through the network. PTP is used wherever NTP and its millisecond accuracy fall short: in motion control, digital substations based on IEC 61850, mobile networks, broadcasting and trading systems. In OT networks and industrial automation, PTP determines whether measurements, alarms and movements from different devices really sit on one shared timeline.

🧠 What exactly does PTP do?

PTP distributes time from a single reference clock, the grandmaster, to every other clock in the same PTP domain. The grandmaster usually takes its time from GPS or another GNSS receiver, often backed by a stable oscillator (OCXO or rubidium) that holds the time when satellite reception is lost.

PTP does three things:

• Distributes time - the grandmaster periodically sends Sync messages carrying the exact transmission time
• Measures transit delay - each receiver measures how long a message spent on the network and corrects for it
• Selects the best source - using the Best Master Clock Algorithm (BMCA), clocks automatically elect the best grandmaster and fail over when it disappears

There are three main versions. IEEE 1588-2002 (PTPv1) was the first. IEEE 1588-2008 (PTPv2) is not compatible with v1 and underpins almost every current profile. IEEE 1588-2019 (PTPv2.1) is backward compatible with v2 and adds, among other things, a High Accuracy profile (derived from CERN’s White Rabbit project, sub-nanosecond) and an Authentication TLV for message integrity. The IEEE 1588g-2022 amendment introduced the optional terms timeTransmitter and timeReceiver alongside master and slave.

🔧 How does PTP time synchronisation work?

At its core is an exchange of four time stamps between the master (time transmitter) and the slave (time receiver):

From these the path delay follows: ((t2 − t1) + (t4 − t3)) / 2, and the offset of the slave clock: ((t2 − t1) − (t4 − t3)) / 2. The slave steers its clock with a control loop that corrects both phase and frequency.

Important details:

• Symmetry assumption - PTP assumes the outbound and return paths take equally long. An asymmetry of 1 µs causes an error of 0.5 µs, and PTP itself cannot detect it.
• One-step or two-step - a two-step clock sends the precise time t1 afterwards in a separate Follow_Up message; a one-step clock writes it straight into the Sync message.
• Hardware time stamping - accuracy depends on where the time stamp is taken. In the network card or PHY you reach tens of nanoseconds; with software time stamping in the operating system you are stuck at microseconds to tens of microseconds.
• Transport - PTP runs over UDP/IP (port 319 for event messages, 320 for general messages) or directly over Ethernet layer 2 with EtherType 0x88F7.
• Announce and BMCA - clocks advertise their quality in Announce messages. The BMCA compares, in order, priority1, clockClass, clockAccuracy, the variance, priority2 and finally the unique clockIdentity.

🧱 Which clock types does PTP use?

A PTP network is only accurate if the switches take part as well. Ordinary switches introduce variable queuing delays in their buffers, which translate directly into jitter on the time.

Boundary clocks scale well in large networks because they split the message flow per segment. Transparent clocks are simpler to manage and prevent errors from accumulating through a chain of cascaded control loops.

🏭 Where is PTP used in OT and critical infrastructure?

Because every sector has different requirements, there are PTP profiles: fixed combinations of message intervals, transport, delay mechanism and domain numbers.

Concrete OT applications:

• Digital substations - sampled values from merging units must line up to roughly 1 µs, otherwise protection relays calculate with the wrong phase angles. This is the best-known reason to use PTP in SCADA and substation networks.
• Motion control and robotics - axes, PLCs, cameras and vision systems must act at exactly the same moment. EtherNet/IP uses CIP Sync for this, which is based on IEEE 1588 (see IP). ProfiNET IRT uses PTCP, a protocol derived from PTP. EtherCAT uses its own distributed clocks (below 100 ns) and can connect to an IEEE 1588 grandmaster through a gateway. OPC UA FX over TSN uses gPTP as defined in IEEE 802.1AS-2020.
• Sequence of events and historians - events from different devices only appear in the right order in a historian or log when their clocks agree to the microsecond.
• Telecoms - 5G NR with TDD requires a time error of no more than ±1.5 µs at the radio interface.
• Financial sector - MiFID II (RTS 25) requires high-frequency trading to stay within 100 µs of UTC with a granularity of 1 µs, which is hard to achieve with NTP.

🆚 What is the difference between PTP and NTP?

In practice both coexist: PTP for time-critical process equipment and NTP for servers and logging in the IT layer.

🛠️ What should you watch for when introducing PTP?

PTP only works well when every link in the chain takes part. A few practical points:

• Choose one profile per domain - devices with different profiles or domain numbers do not synchronise with each other. IEC/IEEE 61850-9-3 uses domain 0 by default and IEEE C37.238-2017 uses domain 254; both profiles can coexist on the same network.
• Check hardware time stamping on every hop - a single switch without boundary or transparent clock support in the path can dictate the accuracy of the whole segment.
• Allow for asymmetry - unequal fibre lengths, media converters or different link speeds in each direction cause a fixed error that you can only compensate with a measured correction value.
• Plan holdover and redundancy - define how long the grandmaster stays within the required accuracy without GNSS, and keep a second grandmaster with a different time source ready.
• Measure rather than assume - use the offset and path delay values that end devices and switches report (on Linux, for example, through linuxptp with ptp4l and phc2sys) and record them as a baseline at commissioning.

🔐 How do you secure PTP?

PTPv2 has no authentication by default. Anyone who can transmit on the network can announce themselves as a better grandmaster with a favourable priority1 or clockClass and shift the time of every device. There are also delay attacks: a man-in-the-middle holds back messages in one direction, creating an asymmetry that remains invisible even with cryptographic integrity checks. The GNSS source of the grandmaster is another weak spot (see GNSS Interference).

Practical measures:

1. Segment timing traffic into its own VLAN or zone in line with network segmentation and IEC 62443.
2. Block the master role on edge ports - many industrial switches offer a master-only or slave-only port setting, so an end device can never become grandmaster.
3. Use redundant grandmasters with a holdover oscillator and diverse time sources.
4. Monitor offset, path delay and grandmaster changes; a sudden change or jump is a warning sign.
5. Deploy the Authentication TLV from IEEE 1588-2019 where equipment supports it.

❓ Frequently asked questions

What is the difference between PTP and IEEE 1588?

PTP is the name of the protocol and IEEE 1588 is the standard in which Precision Time Protocol is defined. The terms are used interchangeably; the versions are called IEEE 1588-2008 (PTPv2) and IEEE 1588-2019 (PTPv2.1).

How accurate is Precision Time Protocol?

With hardware time stamping and PTP-aware switches, PTP typically achieves sub-microsecond accuracy, often tens of nanoseconds. The High Accuracy profile in IEEE 1588-2019 even allows sub-nanosecond accuracy, whereas software time stamping usually stalls at microseconds.

Do I need special switches for PTP?

PTP works over ordinary switches, but then variable buffer delays determine the accuracy. For sub-microsecond accuracy you need switches that act as a boundary clock or transparent clock, plus end devices with hardware time stamping.

What happens if the PTP grandmaster fails?

The Best Master Clock Algorithm automatically elects the next best clock as the new grandmaster. If only GNSS reception is lost, the grandmaster enters holdover and its oscillator keeps time with limited drift.

Can PTP fully replace NTP?

PTP can replace NTP, but in most organisations that is neither necessary nor sensible. NTP remains suitable for servers and workstations, while PTP is used for equipment that needs microsecond precision, such as merging units, drives and 5G base stations.

Is PTP vulnerable to cyberattacks?

Yes, PTPv2 has no authentication by default, so an attacker can pose as the grandmaster or delay messages. Segmentation, master-only ports, redundant grandmasters and monitoring of offset jumps reduce that risk.

📌 In summary

Precision Time Protocol (PTP, IEEE 1588) is the standard for accurate time synchronisation in OT, energy and telecom networks. It measures network transit delay with four time stamps, elects the best grandmaster automatically and achieves sub-microsecond accuracy with hardware time stamping and boundary or transparent clocks. Profiles such as IEC/IEEE 61850-9-3, IEEE 802.1AS and ITU-T G.8275.1 tailor the protocol to each sector. Because PTP is unsecured by default, grandmasters and timing traffic deserve the same protection as any other critical OT asset.

PTP (IEEE 1588) keeps every device in an industrial network living on the same microsecond; its accuracy stands or falls with hardware time stamping, PTP-aware switches and a well-protected grandmaster.