[time-nuts] 60 Hz graphs
hmurray at megapathdsl.net
Mon Sep 12 06:22:04 UTC 2011
Here is a typical 2 day view.
Note that there are 3 interesting events visible on the frequency graph.
More on those later.
I think this is a generator dropping off-line, or something interesting at
the main power production level.
That happens at about 4 hours on the above pair of graphs. On the offset
graph, it's just a tiny glitch. It's much easier to spot in the frequency
Here is another similar event:
It took 0.07 hours (4 minutes) for the frequency to get back to normal.
During that time, the system lost 13 cycles or over 0.2 seconds.
Here is a typical extra cycle event. This one is actually 2 cycles.
It's about 16 hours on the above graphs.
Here is another extra cycle:
This is about 27.5 hours on the above graphs.
Here is a time step on the machine collecting the data.
I was downloading a huge file. That added a lot of queuing delay on my DSL
link. After a while, ntpd decided the delayed version was the correct time
and shifted to it which produced a 1/4 second offset. After another hour the
download finished and ntpd shifted back to the correct time.
1/4 second at 60 Hz is 15 cycles. 1/4 second out of 10 seconds at 60 Hz is
[I have good time locally. I left that laptop running off network time
servers so I could see things like this and/or other glitches I didn't
This looks like a glitch in the PPS data collection code. I think it got the
new time-stamp with the old count.
My guess would be a missing lock or disable interrupts. I've only seen one
of these over roughly 2 months. The extra cycle events are more like 1 per
These are my opinions, not necessarily my employer's. I hate spam.
More information about the time-nuts