[time-nuts] 60 Hz power glitch, US West coast (Silicon Valley)

Tim Shoppa tshoppa at gmail.com
Wed Feb 5 21:13:46 EST 2014


Hal is graphing seconds of offset and seeing 5 seconds worth of shift in
one day. Worrying about phase shift across the transformer changing with
temperature, is like rearranging deck chairs on the titanic :-)

Tim N3QE


On Wed, Feb 5, 2014 at 5:40 PM, Tom Harris <celephicus at gmail.com> wrote:

> For your setup measuring mains there will be a large phase difference
> across the transformer. This is due to very many physical properties of the
> materials, the largest being the magnetic succeptability of the core. Now,
> this does show a slight temperature dependance. So how do you know that you
> are not getting a slow variation in the phase showing up as a frequency
> shift, since you are measuring such tiny variations. I know that the
> transformer is probably in thermal equilibrium with it's surroundings, so
> is at a steady temperature, but this problem  (of getting an accurate idea
> of mains frequency & phase) has exercised me over the years. I currently
> use an opto and voltage reference to get mains frequency, phase & and
> voltage (computed by lookup table from pulse width) which I found was more
> stable than a transformer. And cheaper as well, since this is for a
> commercial product.
>
> I'm just surprised that you get such results with a cheap transformer.
>
> Just remembered, we got a tiny change in phase shift across a transformer
> due to its orientation, we could turn it 90 Deg and get a tiny change (less
> than a milliradian), we never got to the bottom of it, maybe the Earth's
> magnetic field?
>
>
> Tom Harris <celephicus at gmail.com>
>
>
> On 6 February 2014 04:39, Hal Murray <hmurray at megapathdsl.net> wrote:
>
> >
> > jimmydburr at gmail.com said:
> > > Interesting.. I'm assuming the green graph is actual voltage and the
> red
> > > graph is..?
> >
> > The green is the frequency as measured over the last 10 seconds.
> >
> > The red is the long term clock offset in cycles relative to what it would
> > be
> > if the frequency was exactly 60 Hz.  It's the error you would see if you
> > looked at a clock that was tracking the power line.  The 0 point is
> > arbitrary
> > since I can't see the reference clock the power system is using.  For
> those
> > graphs, I used the start of the day/file as 0.
> >
> >
> > > I've never done any mains monitoring/measuring and was wondering,
> what's
> > > your equipment setup?
> >
> > It's simple.  The hardware is an AC wall wart and a couple of resistors
> as
> > a
> > divider connected to a modem control pin.  I forget which one.  It's the
> > one
> > that ntpd expects to use with a PPS input.
> >
> > There was a discussion on that topic here a year or 3 ago.  It's in the
> > archives, but I couldn't find it with a quick look.
> >
> > The software is a simple python hack.  It runs on Linux.
> >   http://www.megapathdsl.net/~hmurray/time-nuts/60Hz/pps.py
> >
> > Linux has a back door to the PPS info.  Things like
> > /sys/class/pps/pps0/assert give text like this:
> >   1391619268.999925084#1125070
> > The number left of the # is the time of the last PPS.  The number to the
> > right is the pulse count.  The software above just waits 10 seconds,
> grabs
> > another sample, and writes a line of text to a log file and switches to a
> > new
> > file every day.  It's 1/2 megabyte per day.
> >
> > If you have FreeBSD or NetBSD rather than Linux, it shouldn't be too hard
> > to
> > use the same API as ntpd uses.  I don't know how PPS works on Windows.
> >
> > Another approach would be to feed it into the audio input and scan for
> zero
> > crossings.  I captured the raw binary for a while when I was chasing some
> > noise glitches.  It's a lot of data.
> >
> >
> > --
> > These are my opinions.  I hate spam.
> >
> >
> >
> > _______________________________________________
> > 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.
> >
> _______________________________________________
> 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