[time-nuts] PTP Grandmaster Clock using Macro Sniffing

lincoln lincoln at ampmonkeys.com
Thu Mar 24 08:28:04 EDT 2016


Hello,
   First full disclosure, I work for a PTP vendor. 
   A master is not that difficult. The protocol is well defined and there are a few open source stacks available. More and more PHYs have time stamping built in though each manufacturer has their own method for getting time stamps out of the PHY. Time stamping also can be done in the the ethernet MAC as well with a small FPGA. this is what we do. 

The hard part, even for large companies, is getting a good slave servo.  From an operator stand point just sync is not the goal, it is to pass traffic.  The loading of a carriers network can be quite large and the networks are big, staggeringly big.  China Unicom has over 1M clocks in their network. Now they have full on path support (transparent switches) and gateway clocks (node with GNSS as primary ref with PTP as backup) To see what kind of packet delay variation (PDV) slaves must cope with look at ITU G.8261 test case 12 through 17.  Thees test cases try to put together to see what the worst PDV is for a 10 switch between master and slave telco network. While G.8261 is specifically for transfer of frequency it has become the defacto test suite, common phase limits for 3g would be 1.5 us, lte 5g tdd is around 400 to 500 ns.

Now how good dose you slave have to be? If your slave will not see, say 80% load, or packet re-routing (floor jumps),  or have to cope with more than a few small switches, a relatively simple servo with a time constant of 100 to 300 seconds will work fine and get you +- 100 ns across a number of nodes. 

Link
   
On Mar 23, 2016, at 12:36 PM, Vlad <time at patoka.org> wrote:

> 
> 
> Hello,
> 
> Today I woke up with an idea to search the resources about PTP Grandmaster Clock building.
> 
> However, its appeared, only commercial solutions is available now. Its pretty common to see several NTP implementation using various HW (GPS is most frequent now)
> But almost no "home-brew" projects for PTP GrandMaster clocks. I would assume its too complex or just "geek toy" which has no sense to build.
> 
> Also, I am interesting if anybody tried  Macro Sniffing (sensing and slaving to external macrocell broadcasts) for the time synchronization.
> 
> 
> -- 
> WBW,
> 
> V.P.
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.



More information about the time-nuts mailing list