[ntp:questions] IEEE 1588 (PTP) at the nanosecond level?

Brian Inglis Brian.Inglis at SystematicSw.ab.ca
Wed Mar 19 11:41:29 UTC 2014

On 2014-03-19 03:30, Martin Burnicki wrote:
> Brian Inglis wrote:
>> On 2014-03-18 02:59, Martin Burnicki wrote:
>>> All depends on how accurate and precise you can get your timestamps,
>>> and this
>>> is probably easier with network packet timestampers at both sides of a
>>> cable
>>> than with a wireless time transfer method like GPS which usually
>>> suffers from
>>> delays which can't easily be measured, like ionospheric delays.
>>> And yes, I know that this can be improved if you receive 2 GPS
>>> frequencies
>>> instead of only the L1. ;-)
>> Problem with PTP accuracy is how do you set and discipline your GM?
>> With NTP probably.
> Or from a GPS receiver.
>> NIST site has a paper on TWSTFT stating that some of the delays cancel and
>> others can be measured or estimated so that on commercial sat channels they
>> can get uncertainties between two stations down to 0.1-1ns, which is better
>> than they can get with common view even using L1+L2 freq and P code.
>> BIPM monthly national TAI/UTC comparisons show uncertainties up to 10-20ns.
> The original topic was if it's possible to get nanosecond accuracy over a
> network connection, i.e. from a PTP grandmaster to a PTP client, and this
> is basically a matter of accurate PTP packet timestamping on *all* involved
> network nodes.

Given the above stats, unlikely to be able to get 1ns, maybe UTC within 10ns?

Precise GHz clock GbE over dedicated calibrated cables to get ns precision,
with accuracy limited by source? PTP not ATP!

> Another question is which accuracy relative to UTC you can get into the
> grandmaster, which depends on the basic accuracy of the GPS receiver,
> and how you get this accuracy into your PTP grandmaster
> (I'm not talking about a standard PC getting the time from a GPS receiver
> via a 1 PPS slope triggereing an IRQ with associated latencies).

Something like a Thunderbolt GPSDO feeding data, PPS, and 10MHz clock to a
chip executing instructions at some clock multiple and handling interrupts
in a deterministic time to feed the PTP GM?

> At the single nanosecond accuracy level it would also be important to *which*
> local realizations UTC(k) you are referring, UTC(NIST), UTC(USNO), UTC(PTB), ...

The source would need to provided by or calibrated against the ref, at some cost!
Unless at that level you just call it UTC(GPS), UTC(LCL), or UTC(MB)!

Take care. Thanks, Brian Inglis

More information about the questions mailing list