What is PTP or Precision Time Protocol?

Precision Time Protocol or PTP (as it is also known), can be fully defined within IEEE standard 1588.  It is a highly accurate and scalable time source that can efficiently service a wide range of equipment in a network. PTP has the potential to synchronize the time of a terminal device to within 2-3*10-9 seconds (nanosecond) of a master time source. Such timing accuracy can be required to understand the order of events in high speed environments. For example, the order of banking transactions or the sequence of circuit breaker trips in a power distribution system.

The ultimate time source is designated as the grand master clock (GMC) and is expected to be at least a stratum 11 clock device. A Network terminal device will generally have a time master (counter) synchronized directly or indirectly by the GMC.

PTP Network Topology

PTP tools can be classified as a Grand Master Clock (GMC), a Master Clock (MC), Boundary Clock (BC), Transparent Clock (TC) or a slave device. In the diagram below, it is attempted to explain the place where each PTP element is generally used in a network.

PTP Elements of Protocol in a network diagram

 

The GMC is generally a primary time device synchronized to GPS or an atomic clock oscillator.  In either case these become Stratum 1 devices. The distinction between the GMC and MC is generally that a GMC is never a slave to an MC or other PTP clock.  Whereas, a MC may be a slave to a GMC or another MC.

Best Master Clock Algorithm

When multiple MCs exist on a network, an algorithm known as the Best Master Clock Algorithm (BMCA and present in every MC) is used to determine which of the MCs will produce the most accurate time service to the network. Having several MCs on a network is common and provides failsafe network timing functions.  IEEE1588 V2 specifies the messaging protocol of the packet data and how the data is handled. The requirements of the BMCA are also found in this specification. These details are out of the scope of this paper.  However, it is the next layer of information to fully understand how PTP manages time and achieves such high accuracy and reliability.

How PTP manages time and achieves high accuracy and reliability

LANs generally are comprised of routers, switches, and hubs facilitating the use of many devices on the network. Routers, switches, and hubs are devices of varying complexity offering interoperability between fibre, fibre and wi-fi transport channels.

A countless number of data packets move through these devices.  They are transferred from source nodes to destination nodes.  Within them the routing information embedded in the packets. Consequently, transport of any one packet from the input through routing or switching to the destination device will be delayed.

The delay is a variable dependent on the store-and-forward methods used (if any), network traffic volume at the time, traffic shaping functions applied and the priority assigned to packet classes. Therefore, the time embedded in a PTP packet will be old compared to the real-time at the exit of the GMC; a latency. Each node through which the time packets travel to get to the destination adds latency. To maintain time

accuracy throughout the network these latencies must be measured and corrected at each node until the time packet arrives at a terminal device.

PTP Network Aware

Consequently, care must be given to the selection of any router, switch or hub used in a PTP network. Without PTP aware devices, latencies are additive at each node.  This can result in large time disparities between the actual time and the time that is embedded in the PTP packet when it arrives. Further, latencies vary as described earlier, so the total latency will have significant jitter. The real time will not be known at the desired accuracy.  In the limit nanosecond (10-9) accuracy at the terminal device can be achieved using PTP aware devices throughout the network. Keep in mind that this accuracy is relative to the stratum 1 transfer standard.  A switch or hub can be a transparent clock. In this case, it slaves to the MC or GMC on the network and appears as a MC to the output nodes.

Drift Rate of the Time Master

Failures such as GMC disconnect, MC loss of lock, network interference, transparent clock (and switch, hub, router) can interrupt the messaging between the GMC or MC and those slaving to the MC on the network. If time accuracy is important, the GMC, MC and terminal devices should have a means to maintain time accuracy during an outage (unlocked state). During an unlocked state if the frequency driving the time master is exactly matching that of the MC clock to which it was slaved, no time accuracy would be lost. However, crystal aging and temperature variation will progressively drive the output frequency from perfection. The rate at which it

drives off (drift) will result in a cumulative time error between the time master and real time.  The rate at which this occurs is known as the drift rate of the time master.

Time Master

Time is kept by setting a counter to the current time of day (TOD). This counter is the time master of the device. In any device that synchronizes time or synchronizes to a time master, will have a time master of some variety.  The block diagram below sketches out the elements to generate a precise TOD.

Time Master Diagram

The output of the crystal oscillator may be phase locked.  Therefore, disciplined by a GPS receiver or a land-based atomic clock. This frequency drives a counter. The counter is designed such that each increment represents an element of time (resolution). For example, to operate at a 1microsecond resolution, the counter would count one million increments between 1-second intervals. Some systems, may use a crystal oscillator as the frequency source when unlocked to GPS.  Some systems may use an atomic clock (rubidium, cesium) which is synchronized to GPS periodically. Disciplined oscillators and atomic clocks are used when unlocked to limit the drift of time from the transfer standard while unlocked.

Drift Error

Drift error is specified as time error/hour or time error/year.  In general, the exact time is set in two parts.

 

  1. The year, days, hours, minutes, and seconds of the day and
  2. the fractional second count is set to zero at the 1-second mark.

 

The 1-second mark is a reference within the GPS signal and is the GPS transfer standard of time. The GPS 1-second mark, referred to as the one pulse/second (1PPS) signal.  It is used to indicate when the fractional count should be a zero value (zero rollover). However, when the fractional second is reset to zero at each 1-second mark, it is called a jam synchronization (jam sync). Jam Sync can cause discontinuities in the fractional time count either positive (jump) or negative time (not valid). These discontinuities can misalign a critical series of events.

The size of a time jump (plus or minus) depends on the

  1. difference between the clock frequencies of the atomic clock and the oscillator driving the time master and
  2. the time resolution of the counter (e.g., nanoseconds/count)
    3. and how many counts away from the zero rollover it was at the time of jam sync.

 

Alternative Approach to Correct Absolute Error

Another approach is to discipline the oscillator.  It is adjusted to generate exactly the correct number of counts between one-second intervals. If this is accomplished over the next one-second interval, there is no jam sync event and fractional second counting is monotonic and sequential; no jumps and no invalid times.

ITS equipment uses oven controlled crystal oscillators (OXCO) that have native drift rates between 10 and 30 parts per billion. The output of this oscillator is further disciplined.  So that the output frequency is managed to provide a fixed number of counts between 1PPS signals while locked to a time reference (e.g., GPS, IRIG B, PTP master clock). While locked, resolution of the time master is one second divided by the disciplined clock frequency. The accuracy will be ±1 count. In one of ITS’s systems, the oscillator is disciplined to a generate 40 million counts each second yielding a resolution of 25 nanoseconds and an accuracy of ±25 nanoseconds.

Accuracy and ITS Products

This accuracy will be fixed while locked to the transfer standard. Disciplining the output clock to a fixed number of counts between 1-second marks serves to calibrate out any absolute frequency error of the crystal. A crystal will have an absolute error.  This is a combination of the frequency error at the time of manufacture, aging error, and temperature variation. Using an OXCO minimises temperature variation error, but manufacturing and aging errors remain. These errors should be compensated out to minimise unlocked time drift.

In ITS systems, the disciplined frequency output is dynamically adjusted second-to-second using the 1-second mark of the transfer standard.  Therefore, resulting in an unlocked drift that is 5-8 times better than the OXCO would produce. In the ITS 6055C and NetVIDxs products, the time master has an unlocked drift at less than 5 microseconds/hour (less than 2 parts/billion).  This exceeds the stability of that defined for Stratum 22.

The ITS discipline system can achieve as little as 5 microseconds per hour of drift.  Without resorting to a high cost atomic clock source. At this low drift rate, a 10 minute outage will result in a time error of less than 1 microsecond, which is excellent for many applications.

ITS Precision Time Protocol or PTP Devices

ITS offers a family of PTP devices that can serve as grand master and master clocks. ITS’s time synchronize products  use a PTP MC as a time reference.  In addition to GPS, IRIG B and 1PPS including their Janus and NetVIDxs product lines. Some of their products bridge the gap between systems that need IRIG B synchronization and the desire to synchronize all your systems to a single PTP GMC or MC.

  1. There is a Stratum 0 which is generally the time at the GPS satellite itself or at an atomic clock designated as a world time master by appropriate controlling bodies.
  2. For an explanation of the Stratum levels (1-4) refer to an article, What’s the Difference Between the Four Stratum Levels by Rob Butkowski (March 9, 2019).

 

This article was kindly provided by our friends at Instrumentation Technology Systems (ITS) If you want to learn more about their product lines or PTP systems, contact ITS at [email protected], www.ITSamerica.com, or by telephone at +1 818-886-2034 (USA).

 

However, if you reside in Australia or New Zealand.  Contact Metromatics as we represent ITS in this region and would be glad to assist you.  Contact us.

Leave a Reply

Your email address will not be published. Required fields are marked *