[time-nuts] Recommendations for Mains Power Monitor / Logger

Tom Van Baak tvb at LeapSecond.com
Sun Mar 11 12:32:02 UTC 2018


Bob,

Correct, measuring mains frequency to a couple of digits is not hard. What makes an interesting challenge is to monitor mains, "kitchen clock", phase drift. And to do it with cycle accuracy; no slips. Note that to measure down to 1 cycle over 1 day is 0.2 ppm. Over a month, 6 ppb, and over a year, 5e-10. So the numbers add up and you see why we use atomic standards or GPS or even NTP as a long-term reference for this.

Your measurement system needs to have short- and long-term stability ~10x better than:
    http://leapsecond.com/pic/mains-adev-mdev-gnuplot-g4.png

Again, that's not asking a lot. But it makes a really fun project. Much of what you ever need to know about time & frequency metrology can be done by a student with $10 in parts and a 60 Hz outlet.

/tvb

p.s. Yes, it's very early here on the west coast, but I had to check how badly my WWVB clocks handled DST a few hours ago.

----- Original Message ----- 
From: "Bob kb8tq" <kb8tq at n1k.org>
To: "Discussion of precise time and frequency measurement" <time-nuts at febo.com>
Sent: Sunday, March 11, 2018 4:53 AM
Subject: Re: [time-nuts] Recommendations for Mains Power Monitor / Logger


Hi

So, how good is “good enough?”. My first attempt ran a counter with a 1 us period resolution. 
(remember, it was tube based …). That turned out to be major overkill in terms of line frequency
measurement. 60.123 Hz is doing pretty well in terms of line frequency. Even to get that level, you 
will be doing a bit of filtering (or you are  just watching the last two digits pop around randomly). 

Your typical time base in a PC is good to a few hundred ppm. That’s giving you an error in the 
fourth digit of your measurement. With a bit of luck, your sound card timebase may be 5X 
more accurate than your system clock. (or it may be worse …) it depends a bit on how fancy
your audio setup is. 

Adding NTP to your PC will correct for any long term errors. In a rational environment it should 
get you into the “few ppm” range short term and zero error long term. 

A GPS gizmo will get you into the parts per billion (or better) range. It might be 100’s of ppb, but it’s
still *way* better than your CPU clock. The usual auction sites have lots of candidates in the sub $50
range.There are also places that are happy to sell you shields with GPS devices on them.

A fancier yet solution is a GPSDO. We are well into overkill at this point. The advantage to using
one is that it may be the time / frequency standard for your entire lab setup. You are up in the 
$100 to $500 range for most of them. They will get you into 10’s or 100’s of parts per trillion. 

There are indeed *lots* of different time sources you could use. The number of alternatives is 
*much* larger than what’s on the list above.

Bob

> On Mar 10, 2018, at 11:46 PM, Tom Van Baak <tvb at leapsecond.com> wrote:
> 
>> I've done some Googling and have found any number of designs.
> 
> Pat,
> 
> 1) Safety. I usually use a low voltage step-down transformer. This gives isolation and safety. Anything from 3 VAC to 24 VAC is fine.
> 
> 2) Trigger. There are dozens of schematics on the web for capturing the zero-crossing of a low-voltage sine wave. You can easily go overboard on this. Or just keep it simple and feed the signal through a resistor directly into a microprocessor input. The internal clamping diodes do their thing. A Schmitt trigger input is helpful but not necessary depending on how your software makes the measurement.
> 
> 3) Timebase. Given the long-term accuracy of mains (seconds a day, seconds a year) you don't need an atomic timebase. If you collect data for a couple of days any old XO will be fine. If you plan to collect data for months you may want a OCXO. Most of us just use cheap GPS receivers.
> 
> 4) Measurement. There are many ways to measure the signal. You can measure frequency directly, as with a frequency counter. You get nice data but it may not be perfect long-term due to dead time or gating effects in the counter.
> 
> So what most of us do is measure phase (time error) instead. One way is to make time interval measurements from a given mains cycle to a GPS 1PPS tick or vice versa, from each GPS/1PPS tick to the very next mains cycle. Either way you get about sample per second. If you're in search of perfection it gets a bit tricky when the two signals are in a coincidence zone.
> 
> The other approach is not to use a frequency or time interval counter at all. Instead you timestamp each cycle, or every 60th cycle. Unix-like systems have this capability. See Hal's posting. I use a picPET, a PIC microcontroller that takes snapshots of a free-running decimal counter driven by a 10 MHz timebase (OCXO or GPSDO).
> 
> The advantage of the timestamp method is that you don't ever miss samples, you can time every cycle (if you want), or throw away all but one sample per second or per 10 seconds or per minute, etc. And best of all, timestamping avoids the hassles of the coincidence zone.
> 
> 5) CPU. A plain microcontroller, or Arduino, or R-Pi can be used. Or if you're on Windows and have a native or USB serial port try this simple tool as a demo:
> 
>    http://leapsecond.com/tools/pctsc.exe
>    http://leapsecond.com/tools/pctsc.c
> 
> 6) An assortment of mains links:
> 
> http://leapsecond.com/pages/mains/
> http://leapsecond.com/pages/mains-cv/
> http://wwwhome.cs.utwente.nl/~ptdeboer/misc/mains.html
> http://leapsecond.com/pages/mains/mains-adev-mdev-gnuplot-g4.png
> http://leapsecond.com/pages/tec/mains-clock-ani.gif
> http://leapsecond.com/pages/ac-detect/
> http://leapsecond.com/pic/picpet.htm
> http://leapsecond.com/pic/pp06.htm
> 
> 7) Final comments.
> 
> It is tempting to worry about the design, as they are so many out there on the web. Which is best? What are the pitfalls? What about noise immunity? What about precision and accuracy? My recommendation is not to over-think this. Just throw something together and see what you've got. Most of the work is with handling the data you get, doing the math, making plots, etc. If after the first day you see odd-looking 16 ms jumps in your data then you know you need to pay more attention to trigger level or noise issues.
> 
> 8) A sound idea.
> 
> We need someone to try out the sound card method. Send the isolated low voltage AC into the L channel and a GPS 1PPS into the R channel. "The rest is just software." Note that because you have access to the entire sine wave there's a lot you can do with this method besides making charts of time drift or frequency deviation from the zero-crossings.
> 
> For an even cheaper solution, forget the GPS receiver and the R channel -- since the PC (if running NTP) already knows the correct time. And skip the AC transformer too -- instead just hang a foot of wire off the L channel input. There's mains hum everywhere. It would be the one time in your life where the ever-present audio hum actually has a good use.
> 
> /tvb
> 
> ----- Original Message ----- 
> From: "Patrick Murphy" <fgdhrtey at gmail.com>
> To: <time-nuts at febo.com>
> Sent: Saturday, March 10, 2018 2:53 PM
> Subject: [time-nuts] Recommendations for Mains Power Monitor / Logger
> 
> 
> All this talk of varying mains power frequency aberrations has me
> curious what is happening in my own back yard here in Tulsa in the
> USA. Can some recommend a reasonable "introductory level" solution for
> this? (As a fledgling Time-Nut, those two words were hard to say.😀)
> At the least I would like to watch voltage and frequency, with a
> configurable monitoring and logging interval. I can provide precise
> timing as needed for synchronization and time-stamping. Expanded
> ability to also monitor amperage, various power factors, etc is a plus
> but not required at this point.
> 
> I've done some Googling and have found any number of designs. What I
> can't tell is how well they work. I am pretty handy with my hands and
> do not at all mind a DIY solution.
> 
> So what do the Oracles say?
> 
> Thanks!
> 
> -Pat
> _______________________________________________
> 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.

_______________________________________________
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.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mains-adev-mdev-gnuplot-g4.png
Type: image/png
Size: 178623 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20180311/11ebf9ef/attachment.png>


More information about the Time-nuts_lists.febo.com mailing list