What Is Precision Time Protocol (PTP)?

Precision Time Protocol (PTP) is a network protocol that keeps the clocks in many devices locked to one reference. It is defined in IEEE 1588, first published in 2002. The 2008 revision, PTP version 2, is what almost all equipment runs today, and the 2019 revision (sometimes called v2.1) added security and management features while staying compatible with it.

PTP exists because many systems need to agree on time far more closely than the Network Time Protocol (NTP) can manage. NTP typically keeps computers within a millisecond or so of each other on a local network. PTP, with timestamping done in hardware at the network interface, typically reaches sub-microsecond accuracy, and specialised profiles go to tens of nanoseconds. That is tight enough to trigger cameras at the same instant, time-stamp electrical faults across a substation or line up motion axes on separate controllers.

In conversation it is usually just PTP, IEEE 1588 or 1588v2. Its automotive and TSN variant, gPTP, is covered below.

PTP measures two things: how far a device’s clock is from the reference (the offset), and how long messages take to cross the network (the path delay). A short exchange of timestamped messages gives both. The master sends a Sync message and notes the exact send time, t1, which it delivers in the Sync itself or in a Follow_Up message. The slave notes when Sync arrived, t2. The slave then sends a Delay_Req at t3, and the master replies with the arrival time, t4, in a Delay_Resp.

From those four times the slave works out the mean path delay, ((t2 − t1) + (t4 − t3)) ÷ 2, and its offset, ((t2 − t1) − (t4 − t3)) ÷ 2, then steers its clock to remove the offset. The exchange repeats, typically once a second or faster. The maths assumes the path takes the same time in both directions, which is why asymmetric links and software timestamping limit accuracy.

Before any of this, the best master clock algorithm (BMCA) compares the Announce messages from every candidate clock, using priority, clock class and accuracy, and elects the grandmaster. If the grandmaster fails, the next best clock takes over automatically.

The standard defines several roles. Which ones your switches support decides how accurate the network can be.

IEEE 1588 is a toolkit. Industries pick a profile that fixes the options, message rates and transport for their use, and devices must support the same profile to work together.

gPTP, the generalised PTP of IEEE 802.1AS, is the one most often asked about. It is a tightly defined subset: every device in the path must be time-aware, only peer-to-peer delay measurement is used and it runs directly over Ethernet. That makes behaviour predictable, which is why it is the clock beneath time-sensitive networking. Our guide to time-sensitive networking covers the rest of the TSN family.

PTP is now built into much industrial hardware, often without being noticed. Multi-camera machine vision and robot guidance use it to capture frames from several cameras at the same instant; our white paper on synchronising machine vision with IEEE 1588 walks through that case. Substations time-stamp sampled values and fault records with the power profile. Motion control, test rigs, broadcast and mobile networks rely on it too.

Accuracy depends on every link in the chain, so the hardware matters. In our range, the multi-port PoE LAN cards among our Ethernet and PoE frame grabbers list IEEE 1588 PTP, and several add IEEE 802.1AS, so an industrial PC can time-stamp camera traffic in hardware. Selected managed industrial switches list IEEE 1588 v2 PTP with the 1588 default, ITU-T G.8265.1 and G.8275.1 profiles, and IEEE 1588 builds of our board-level embedded Ethernet switches carry PTP inside vehicles and sensor rigs.

We supply PTP-capable network cards and switches inside complete industrial systems, and as standard we configure the PTP profile, clock roles and timestamping before the system ships.