Not on the eastern grid, but I had hook my picpet ac mains logger back to the southwest grid if that would be of any help.

On Nov 15, 2013, at 7:51 PM, Bill Dailey docdailey@gmail.com wrote:

would probably be an interesting comparison.  I am working with a guy on
the eastern grid part now.

You arent using python for processing on the pc are you?  If so, I would be
interested in your script.  I am trying to verify I am not a just a little
off with mine.

Bill

On Sat, Nov 16, 2013 at 1:55 AM, Kevin M. Rosenberg kevin@rosenberg.netwrote:

--
Doc

Bill Dailey
KXØO

Here is a real-time map of mains frequency
http://fnetpublic.utk.edu/gradientmap.html

A common way to measure this is to connect a 9 volt AC plug-in power supply
to the DCD pin of a serial port and let the PPS system log and time stamp
each cycle.  You can do it no more hardware than the transformer and any
PC-like computer.  The computer can bedding other tasks at the same time,
like serving files or whatever.

On Fri, Nov 15, 2013 at 6:51 PM, Bill Dailey docdailey@gmail.com wrote:

--

Chris Albertson
Redondo Beach, California

Your method tosses out a lot of data.  You can't see transients.  Ideally
rather then record a 1 second average you'd record the time of EVERY zero
crossing.  It sounds like a lot of data but not really.  You only record
32 bits 60 times each second.  That is 240 bytes per second.

On Fri, Nov 15, 2013 at 6:51 PM, Bill Dailey docdailey@gmail.com wrote:

--

Chris Albertson
Redondo Beach, California

On 11/16/2013 08:55 AM, Kevin M. Rosenberg wrote:

They are separated, so they are not phase coherent. Texas is it's own
grid too.

Cheers,
Magnus

Understood.. I have it every 6 cycles now to sync up with his
synchrophaser.  I wrote the script so I can specify the number of cycles I
average.  Right now it is at 6 because his measures are every 0.1s.

On Sat, Nov 16, 2013 at 2:52 AM, Chris Albertson
albertson.chris@gmail.comwrote:

--
Doc

Bill Dailey
KXØO

rough comparison... I didnt have my interval right for this set.

https://www.dropbox.com/s/1gi5tbf96yop5hz/stonercompare.JPG

On Sat, Nov 16, 2013 at 3:02 AM, Bill Dailey docdailey@gmail.com wrote:

--
Doc

Bill Dailey
KXØO

On 11/16/2013 09:52 AM, Chris Albertson wrote:

But you want it filtered to avoid the transients. Those are really not
that interesting when you measure the grid.

Also, if you use the event trigger method you probably want to use an
amplifier to increase the slew-rate such that noise does not convert
into time jitter.

Cheers,
Magnus

Random noise always converts into time jitter.  It doesn't
matter how much you amplify the input signal, noise can push
the detected zero crossing wherever it wants to.

-Chuck Harris

Magnus Danielson wrote:

The signal is 120 volts.  You hardly need to amplify it.  Clip it with a
diode to +- 9 volts so as not to blow up your serial port.  But I'd use a
transformer for safety. The zero crossing detectors are built into the
RS232 interface.    You take advantage of the RS232 spec which has a DCD
pin input of about +-9 volts that is already set up to find a leading edge
of a pulse and cause a very low latency interrupt.  The system software
already will capture the time all inside a kernel level interrupt handler.

The jitter turns out to be on the order of a single digit microseconds.
Good enough for measuring a 60Hz signal.

I guess if you want to see transients depends on the purpose of the
experiment.  Are you looking at local AC power quality or wanting to
measure the grid.  The grid is well monitored, just use FNET and you get
real-time data for all of North America.  I think the reason for measuring
it yourself is to see local power quality and things load switching inside
your own building, that's transients.

The other way to measure AC with zero added equipment is to treat it as an
audio signal and after reducing it to 1 volt run it into an audio interface
And then use FFT.  This will let you see very small spikes and noise.  It
depends again on your purpose for doing this.

On Sat, Nov 16, 2013 at 1:18 AM, Magnus Danielson <
magnus@rubidium.dyndns.org> wrote:

--

Chris Albertson
Redondo Beach, California

My purpose is to do it with a picpet.  That's it.  So, that eliminates a bunch of the options.  I can decouple the measurements from the pc clock that way.

Doc

Sent from mobile

On 11/16/2013 03:13 PM, Chuck Harris wrote:

You are missing that you can alter how much trigger jitter you get.

Cheers,
Magnus

Doc,

I measure mains time & frequency with a picPET all the time. In fact that's one of the reasons I designed it. If you're having any trouble contact me by email.

/tvb (i5s)

No trouble.  Easy.  I love it.  Keeping track of the rolling counters was a hack because I am so far removed from serious programming.

Sent from mobile

No, I meant the purpose of the whole thing.  Why are you measuring power
frequency?  Not why are you using a PIC.    How will the data be used, what
is the question driving the measurement?

On Sat, Nov 16, 2013 at 11:23 AM, Bill Dailey docdailey@gmail.com wrote:

--

Chris Albertson
Redondo Beach, California

On Sat, Nov 16, 2013 at 09:26:10AM -0800, Chris Albertson wrote:

I tried this a slightly different way. Since there is mains noise
everywhere, I made a small loop and connected it to a 3.5mm jack
and then plugged that into the mic socket on a sound card. You can
get lots of (slightly noisy) samples per second. I took chunks of
this data and took the Fourier transform to find the dominant
frequency:

http://www.maths.tcd.ie/~dwmalone/time/leap2012/#mains

but I guess you could filter it and count crossings too?

David.

There is no higher purpose actually.  I just fiddle.  This is how I relax.

On Sat, Nov 16, 2013 at 3:47 PM, Chris Albertson
albertson.chris@gmail.comwrote:

--
Doc

Bill Dailey
KXØO

Only if the noise in question comes from the trigger electronics.

In the case of a 60Hz mains derived signal, most of the noise is
going to be riding on the signal, and will be amplified with your
gain stage.  A -delta amplitude noise signal will be zero when the
sine wave signal is at +delta.  That will be sometime other than
the actual zero crossing... which is phase noise.

-Chuck Harris

Magnus Danielson wrote:

Chuck wrote:

The potential evils of bandpass filters in a timing chain are well
known, but as long as you can accept the delay of a filter (or
correct for it, which should be trivial with a PIC or other uC), you
may be much further ahead with a noisy signal like the AC mains if
you use a sharp bandpass filter on the incoming 60 Hz then amplify &
clip the signal to increase the slew rate.  Active filters with fast,
quiet op-amps should do the job well.  For the lowest jitter, a
Collins-style multi-stage zero cross detector may be helpful.

Best regards,

Charles

Charles, et al.

I think we agree. Just to clarify...

I rely on no hardware and no software filters when I use a time-stamping counter such as a sub-nanosecond Pendulum CNT-9x or sub-microsecond picPET. An electrical zero-crossing happens when it happens. If you "filter" you're just trying to change history: spikes are spikes; noise is noise; history is history. Deal with it. Record it, don't filter it away.

The beauty of the time-stamping method is that you capture any and all positive zero-crossings. If there is "noise" all it does is create unexpected and obvious artificial too-early or too-late samples -- which are trivial to analyze or eliminate in software.

Some call them "outliers" and ignore them. This is correct. However, if one "filters" or "averages" them, you give validity they may not deserve. Bogus data should be eliminated by logic, not attenuated with pseudo-analog filtering.

You can either focus on the signal, or the noise. That's two separate plots. An extraneous time-stamp happens to me a couple times a month; they are easy to spot and ignore. Similarly, a couple times a year I might miss a 60 Hz sample; these are also easy to spot and repair. For best time & frequency results, never "divide by =60"; instead "decimate by ~1 second".

/tvb

----- Original Message -----
From: "Charles Steinmetz" csteinmetz@yandex.com
To: "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Saturday, November 16, 2013 4:54 PM
Subject: Re: [time-nuts] Mains frequency

Tom,

On 11/17/2013 03:02 AM, Tom Van Baak wrote:

Standard wide-bandwidth counters isn't really ideal for signals like this.

When you measure the mains signal, nominally 60 Hz in this case, spikes
etc. is noise which is local and not of interest when comparing over a
large area. Inter-area oscillations have much slower properties.
If you go the time-stamping way, you should remove such noise.

Removing or padding over it by logic will mean dropping data, which is
not helpful.

Turns out that professional gear for this does not do time-stamping in
this regard. Rather, they I-Q demodulate the signal with a reference
signal at the nominal rate, low-pass filter it and pay attention to
details of filtering like group-delay and compensation thereof. It's a
rather wise approach for the type of conditions you have.

Cheers,
Magnus

Again, why are you measuring the AC line?  I'd think maybe to measure the
noise that is on it.  The fundamental freq. changes second by second.
It's not a clean 60Hz my any means.  The rate of frequency change is one
thing you'd like to measure

I was just watching a minute ago and can see a 0.01Hz/second drift.  It is
likely MUCH worse as what I was watching is filtered over second

On Sat, Nov 16, 2013 at 4:54 PM, Charles Steinmetz csteinmetz@yandex.comwrote:

--

Chris Albertson
Redondo Beach, California

On 11/17/2013 03:33 AM, Chris Albertson wrote:

The phase of it tells you a lot about the powergrid, especially when you
collect data over the full power-grid and at sufficiently high rate.

Cheers,
Magnus

A time-stamping counter never drops data; instead you apply logic to drop noise when necessary.

I'd be happy to compare "professional gear" with a time-stamping counter. I have two or three years of data to show 60 Hz time-stamping works perfectly. Not to say your professional gear might also. But a zero-crossing counter never misses a cycle; you can set the s/w bandwidth just as you can set an IQ hardware bandwidth.  Both assume some reasonable limit of mains df/f/dt. You can either do it with a fancy $100 to $1000 reference signal generator + PLL or FLL + IQ detector + professional box -- or with a $1 PIC and $0 s/w.

/tvb

Chris,

No, forget the noise (it's actually quite clean: look at it sometime).

We measure mains because we can.

We also measure it because millions of wall-clocks are based on mains frequency; it was the original "GPSDO".

We measure it because its phase plot, frequency histogram, and ADEV plot are really quite interesting.

We measure it because Seattle, WA (tvb) and New Mexico (Kevin) are both on the same grid and mutually agree to 10 microseconds (!) over an hour even though they can both wander by many seconds relative to UTC. It's a textbook example of common view time transfer. See also:
http://leapsecond.com/pages/mains/
http://leapsecond.com/pages/ac-detect/

You too can join the mains party. Measure it with your own method, or a fancy TrueTime time/frequency deviation meter (TFDM) or use something simple like a picPET (http://www.leapsecond.com/pic/picpet.htm) or Arduino or even a NTP/Linux/serial DCD pin hack.

/tvb

For a non Time-Nuts application, I needed a narrow bandpass filter that would provide essentially zero phase shift (no more than 10 or 20uS was desired) over a frequency range of 55 to 65 Hz while providing useful reduction of the harmonics, particularly in the range of 400 to 1kHz. This was to filter the line voltage. I needed an output that was  a clean sine wave but with essentially no phase or amplitude error compared to the input signal, even when the input signal changed in frequency.
Of course, there is no "conventional" filter topology that will provide zero phase shift over a range of frequency, you will be lucky to find one that provides zero phase shift at ONE frequency. I knew the switched capacitor filter could be set to provide zero phase shift near the center frequency, but the phase shift would change rapidly when the input frequency changed, like any other filter.
I came across the LTC1060 and I found out that when driving its clock from a PLL that multiplies the line frequency by the proper factor (50 or 100 for the LTC1060), a single potentiometer can then be used to adjust the phase shift to zero nominally, and the phase shift remains below a few uS over the range I was interested in.

Being a switched capacitor filter, it can be made as narrow as desired without affecting the other characteristics too much.

Reading this thread, I just realized that this filter would be very useful to filter the 60Hz before feeding a counter for the purpose of measuring the line frequency and phase without being too much affected by noise and other disturbances.

Didier KO4BB

Charles Steinmetz csteinmetz@yandex.com wrote:

--
Sent from my Motorola Droid Razr 4G LTE wireless tracker while I do other things.

Does an ac transformer hurt me?  I was looking for that dang megohm page when I started this.  Couldn't find it so I used a transformer.

Doc

Sent from my iPad

On Nov 16, 2013, at 9:17 PM, "Tom Van Baak" tvb@LeapSecond.com wrote:

"Turns out that professional gear for this does not do time-stamping in
this regard. Rather, they I-Q demodulate the signal with a reference
signal at the nominal rate, low-pass filter it and pay attention to
details of filtering like group-delay and compensation thereof."

It makes sense to me. That way, you use the entire signal instead of just the zero crossing.
You know that the signal is generated as a sine wave. Therefore, essentially you synthesize a local sine wave that is a best fit to the input signal and you measure the zero crossings of the synthesized signal.

Didier KO4BB

Magnus Danielson magnus@rubidium.dyndns.org wrote:

--
Sent from my Motorola Droid Razr 4G LTE wireless tracker while I do other things.

tvb wrote:

Well, it depends on what one wants to investigate.  The "naked"
history one captures with no filtering may not be the cleanest
history available of the phenomenon under investigation.  Except in
unusual circumstances, mains voltage is generated by massive rotating
machinery -- so anything fast that happens on your incoming mains
voltage is not a reflection of the grid frequency.  If what you want
to know is the grid frequency over time (vector sum of the rotational
velocity of the various generators on the grid, as seen from your
location), a filtered and limited signal may (probably will) provide
the best assessment.  Note that local zero crossings are only a proxy
for grid frequency to begin with -- and not a very good one,
specifically because of the high noise level.  Of course, you can
always filter in software if you time-stamp each zero cross in all
its naked glory, but removing the noise prior to time-stamping is
often preferable to digitally processing a noisy capture.

Put another way, the massive rotating machinery that generates the
mains voltage can only change the zero cross of the grid by a tiny
amount from one cycle to the next.  If a data capture method shows
cycle-to-cycle jitter that is significantly greater than this amount,
the increase cannot be due to the generators, it can only be due to
noise.  If one's interest is the grid frequency, removing this noise
prior to time-stamping can only help.

Note that I'm not talking about a filter Q in the millions -- I'd
probably be inclined to use a linear-phase filter with several Hz
bandwidth, after a more rigorous analysis of the application.

Agreed.  If you are investigating incidental noise on the mains
rather than the grid frequency, then the signal you capture needs to
be at least as broadband as the noise in which you are interested.

Since I do not use the actual local mains zero crossings for anything
(other than electronically switching loads on at zero voltage and off
at zero current, where absolute timing is irrelevant), I'm not sure
why one might be interested in characterizing them.  OTOH, since I do
have equipment that responds to the grid frequency, I can see
practical utility in characterizing that.  Hence my suggestion to filter.

Best regards,

Charles

This resonates with me somewhat since I used to run nuclear power plants and operate the actual turbines.  It does seem that the time interval measurements have much more jitter than I would expect.  I suspect the thousands of turbines phase locked may introduce all kinds of very subtle variations.  I do know when you put a submarine turbine on shore power (grid). You no longer have to control speed... The grid does that for you.

Sent from my iPad

Hi

Let’s hope Santa does not bring you an X10 power line based remote control system for Christmas….

Bob

On Nov 18, 2013, at 5:15 PM, Tom Van Baak tvb@LeapSecond.com wrote:

Tom,

Don't confuse everybody with facts, we had a good thread going :)

The PicPet is holding its own very well in that application.

I am surprised at the effect of the laptop supply. I would have certainly expected effect on phase noise (smaller taus), but not that close to the carrier. Do you know if it is a power factor corrected supply or not?

Didier KO4BB

Tom Van Baak tvb@LeapSecond.com wrote:

--
Sent from my Motorola Droid Razr 4G LTE wireless tracker while I do other things.

tom,

nice plots.  how do you figure out what the contribution of variability vs
noise? In other words there is a differential between the "ideal" and the
actual a dev curves... is there a way to tease out how much nose contribute
to that differential?  It does seem to me that there should be far less
short term variability (< 100s) than there appears to be.  Clearly in the
very short tau (< 0.1 s) the picPET can't tease that out but as the curves
diverge, how much of that is noise? between say 0.1s and 100s?  Being a
power plant operator I would say quite a bit although I am rethinking that
some due to the way the turbines push and pull each other.  I can envision
some fine whole grid oscillations due to that push and pull.

bill

On Mon, Nov 18, 2013 at 4:15 PM, Tom Van Baak tvb@leapsecond.com wrote:

--
Doc

Bill Dailey
KXØO

Hi

An “ideal” curve would go to the bottom of the scale as soon as the plot started. Anything that shows on the ADEV curve is by definition noise. The slope of the ADEV curve can help you determine what sort of noise it is. The slope(s) on an modified ADEV curve can do that slightly better.

Bob

On Nov 18, 2013, at 8:03 PM, Bill Dailey docdailey@gmail.com wrote:

I meant ideal at the noise floor of the picPET (i.e in this case the
generated 60Hz).

On Mon, Nov 18, 2013 at 8:08 PM, Bob Camp lists@rtty.us wrote:

--
Doc

Bill Dailey
KXØO

The power supply contribution is interesting.  This might have been a useful
tool when a year ago I was playing with some very large inverters on a
microgrid.  I had one inverter as master (in UF mode) and two others as
grid-connected slaves in PQ mode.  The first slave would come online just fine
yet when the second was synched the entire microgrid would go unstable.  It was
noise at the zero crossings but not enough to see on a scope.

With all the distributed generation coming online I would be wary of relying on
those zero crossings.

Peter

On 11/18/2013 5:15 PM, Tom Van Baak wrote:

Hi

There is no way to come up with the noise floor of the picPET from that plot. In fact coming up with the floor of a single channel device like the picPET is not all that easy. First you need an ideal noise free sine wave signal …. I’ve spent more than a few hours on that particular project with other list members involved as well. As always we kept it off list to keep from offending those who place a high value on their bandwidth.

Bob

On Nov 18, 2013, at 9:11 PM, Bill Dailey docdailey@gmail.com wrote: