Interesting. Thanks. I plotted your data next to mine. http://users.megapathdsl.net/~hmurray/time-nuts/60Hz/60Hz-2019-June.png There are actually 2 lines there. At this scale, they are on top of eachother.
My samples are roughly every 10 seconds. So there should be a 6 to 1 ratio. If you zoom in, you can see some quirks. This is June 10th. http://users.megapathdsl.net/~hmurray/time-nuts/60Hz/60Hz-2019-June-10.png This is zoomed in enough so that you can see the offset which is arbitary. Zooming in farther. http://users.megapathdsl.net/~hmurray/time-nuts/60Hz/60Hz-2019-June-10a.png Looks like I picked up at extra cycle. I see that occasionally - ballpark of once a day. This is June 25th. http://users.megapathdsl.net/~hmurray/time-nuts/60Hz/60Hz-2019-June-25.png Here is one area that caught my attention. http://users.megapathdsl.net/~hmurray/time-nuts/60Hz/60Hz-2019-June-25a.png It looks like the cycle count on one sample is off by one. I see that occasionally. I assume it's a missing lock in the Linux PPS kernel API that I'm using to read the data. I get pairs of time-stamp and count. A pulse comes through between when it has grabbed the time stamp and when it grabbed the count. The following sample looks good so it's not a extra or missing cycle. -- These are my opinions. I hate spam. _______________________________________________ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.