[time-nuts] 60hz west coast electric grid gone berserk

Hal Murray hmurray at megapathdsl.net
Sat Sep 10 13:09:35 EDT 2011

albertson.chris at gmail.com said:
> You really don't even need software as the Linux PPS project ships a test
> program that logs the time stamps.

That (tries to) print a line for each pulse.  At 60 Hz, that's a lot of 
printing.  It might work reasonably well if you include an external divide by 
60 or 100 or 1000 or some handy number in that range.

> That said, I found an even easier way to monitor AC.  I have an battery
> backup box that keeps the PC running for about 10 minutes after a power
> failure.  The box connects to the computer via USB and some of the data
> passed over the USB cable are measurements of the AC mains power, line
> frequency and voltage.  This gets logged. 

I have an APC unit.  It's got both USB and a serial port.  Only one works at 
a time.  I have code working on the serial port but didn't get it to work via 
USB.  I forget why.

The serial port has commands for reading the lowest/highest line voltage 
since the last time you asked.  That makes it easy to catch short glitches.

You can also get the temperature.


The topic from a month or two ago that started my interest in this area was a 
proposal by the US power industry to stop providing accurate time.

The line frequency from my APC unit isn't useful for tracking time drift.  
Normally, I only see 60.00 and 60.25  I found one sample of 60.50 last month. 
 Mostly, it's boring so I haven't been paying attention to it.

In contrast, with the PPS capture path, it's easy to see frequency variations 
in the 0.01 Hz range.

I think the counter part of the API is the key.  It's really hard to beat 
actually counting cycles if that's what you want to investigate.

In my case, you do have to look carefully to check for extra counts.  I 
assume they are glitches on the line, but I'm not yet setup to capture the 

I see roughly 1 a day.  I'm logging data every 10 seconds so an extra count 
shows up as the frequency off by 0.1 Hz.  It's pretty obvious to spot by eye. 
 I haven't tried to automate it.

These are my opinions, not necessarily my employer's.  I hate spam.

More information about the time-nuts mailing list