(Please keep the list on CC) On Thu, May 17, 2018 at 07:42:07PM +0000, Matt Middendorf wrote: > I measured the error of the clock two ways. I used the -m option on the > ptp4l software to print out messages. It prints out the offset between the > master and slave computer. The other thing I did was I used the 1PPS signal > on the two host adapters and measured the difference between the two clocks > using an interval counter. Both measures are in nanoseconds and are similar.
Ok. Great. > I did try other servo PI values, but none worked as well as the default > hardware values. I think that would indicate you are not dealing with network jitter, but rather a congestion or something similar delaying the sync packets significantly and causing large spikes in the measured offset instead of adding a random well-distributed noise. How large are the spikes? Are you running PTP on a network where switches don't support PTP? A better approach to deal with a spike might be to drop the sample instead of clamping the absolute value of the offset. The servos require periodic updates, so the filter would probably have to limit the maximum number of consecutive samples dropped. -- Miroslav Lichvar ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
