csteinmetz@yandex.com said:

What are you going to do with data from the line accurate to 1 microsecond?

--
These are my opinions.  I hate spam.

Hi

or a nanosecond …

Obviously it would depend on just what your system needed to do and the signal to noise on the 60 Hz input.

Bob

There is an interesting article in the Nov/Dec issue of Inside GNSS
describing the robust measurement of "...voltage and current phasors at
widely dispersed locations in a power grid".  A Phase Measurement Unit
measures and time stamps the voltage and current phasors "...thousands of
times per second..." to an accuracy of <1 us using GPS.  The authors discuss
several strategies for dealing with jamming and spoofing of the civil GPS
signals.  It's a good read.

See http://www.insidegnss.com/node/4281

Mike

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Hal Murray
Sent: Wednesday, December 17, 2014 3:28 AM
To: Discussion of precise time and frequency measurement
Cc: hmurray@megapathdsl.net
Subject: Re: [time-nuts] Simple AC mains zero-cross detector

csteinmetz@yandex.com said:

What are you going to do with data from the line accurate to 1 microsecond?

--
These are my opinions.  I hate spam.

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Hi

Using GPS timing for power network analysis and control dates back into the 1980’s. The guys at Quebec Hydro set up the first system I’m aware of. They had tried it with Loran-C before, but the noise in the vicinity of a major power station made that impossible. GPS being microwave helped them a lot with that part of it. They did a paper at FCS(?) on the results. Back then being able to actually get very accurate phase data over a 100’s of mile range  was a pretty novel thing. Using GPS timing for this actually pre-dates the whole CDMA / GPSDO thing.

Bob

Yes, PMUs makes it a bit different. Here is a random paper someone wrote:
https://rubidium.dyndns.org/~magnus/papers/KTH_paper1.pdf

PMUs samples at some "high" frequency, mixes down the network frequency
to base-band, filters away the mirror frequency before sub-sampling it
into the configured sample-rate. The process is being controlled by the
IEEE C37.118.1 standard, while the communication is described in IEEE
C37.118.2 but an IEC 61850 extension provides for another way of
conveying that data.

PMUs have proven themselves to outperform many of the normal frequency
and ROCOF estimators, which became evident in the 2003 NE black-out
scenario, as the SCADA data kept getting them on the wrong tracks, so
after 8 months they just scrapped the frequency readings from the
traditional equipment and just looked at the PMU data, they could sort
the events out properly in time. Turns out that the details of how a
particular vendor implements the frequency estimation and filtering can
be devastating in getting comparable frequency measures, and thus
loosing the observation precision needed to follow the aftermath properly.

The C37.118.1 has put a stringency on how filtering is to be done, as
well as how frequency and ROCOF (Rate Of Change Of Frequency) is to be
calculated. NIST has a small department focusing on the calibration of
PMUs, and is working actively with the vendors to get them improved.
Good folks and I have helped them a little with some ideas, amongst
others to do through-zero calibrators.

For other events, Digital Fault Recorders (DFR) is being used. They are
essentially memory oscilloscopes which have a more advanced trigger
adapted to go off on all "interesting" events. Today DFRs is a legal
requirement in some countries.

So, I do not completely agree that a through-zero measurement with a TIC
has all the information, and for the information you do get, you would
like to be as careful as the PMU folks about the group-delay of filters,
time-compensation of processing and filters etc. to maintain good
precision. There is reason to look at it and learn.

Cheers,
Magnus

On 12/18/2014 11:59 PM, Mike Garvey wrote:

Magnus wrote:

No, a series of time-stamped zero crossings doesn't have all of the
information in the original signal, and a small glitch that occurs
during the middle of a cycle (far away from a zero cross) could hide
and show nothing more than a slight displacement of one or two zero
crosses.  Grid-nuts can ignore such short glitches.  Utilities can't,
particularly in today's cybersecurity environment.  Horses for courses.

From my observations of the AC mains while I was testing the simple
ZCD, I would expect such hidden glitches [that are real grid-related
phenomena, not just someone starting a motor downstairs] to be very
rare.  The grid phenomena I saw typically last more than one grid
cycle and thus affect more than one zero cross, and/or are large in
magnitude and cause serious displacement of at least one zero
crossing, or several extra zero crossings.

Best regards,

Charles

all we'd be interested in?  Phase is best measured at a zero crossing as
this is the (only) phase measurement point which is independent of
amplitude.
Mike

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Charles
Steinmetz
Sent: Saturday, December 20, 2014 4:57 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Simple AC mains zero-cross detector

Magnus wrote:

No, a series of time-stamped zero crossings doesn't have all of the
information in the original signal, and a small glitch that occurs during
the middle of a cycle (far away from a zero cross) could hide and show
nothing more than a slight displacement of one or two zero crosses.
Grid-nuts can ignore such short glitches.  Utilities can't, particularly in
today's cybersecurity environment.  Horses for courses.

From my observations of the AC mains while I was testing the simple ZCD, I
would expect such hidden glitches [that are real grid-related phenomena, not
just someone starting a motor downstairs] to be very rare.  The grid
phenomena I saw typically last more than one grid cycle and thus affect more
than one zero cross, and/or are large in magnitude and cause serious
displacement of at least one zero crossing, or several extra zero crossings.

Best regards,

Charles

time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Mike wrote:

That is the primary interest (as I understand it -- I am not, myself,
a grid-nut), and the reason the "simple ZCD" circuit uses this
approach.  But grid-nuts are also interested in perturbations of the
grid voltage caused by grid sections going offline and coming back,
lightning strikes, etc., etc.  (After all, simply monitoring the ebb
and flow of the line frequency is about as interesting as watching
the tide come in and go out, so they naturally want some occasional
excitement.)  These anomalies can be detected by their effect on the
zero crossings of the mains voltage, so one data collection serves
both purposes at the time-nuts level.

While the ZCD approach is ideal for monitoring the grid
phase/frequency, and as a bonus provides timing information about
grid anomalies, it does not capture all of the information about
anomalies.  If you are a utility concerned about grid security or
making sure that "new energy" sources play nicely with the grid, you
probably want more information about anomalies than time-stamped zero
crossings provide.  Magnus described a system used by utilities to
track grid anomalies in greater detail.  My reply agreed that zero
cross detection is not the tool of choice for utilities with such
concerns, and noted the different needs of grid-nuts and utilities.

Grid-nuts are well established, and the vast majority of them use
time-stamped zero crossings as their data sets.  I was concerned that
many grid-nuts seem to use non-isolated feeds from the mains that,
while "safe enough" under normal conditions, are not preferred
practice.  I also thought that the timing relationship between the
ZCD and the actual zero cross could be improved and stabilized with a
new ZCD.  So, I designed the "simple ZCD" circuit to provide an
isolated source of very predictably timed pulses with fast edges.  I
tested it and it proved to be reliable and to have very stable timing
with respect to the line zero crossings, so I published it and
announced it on-list with the first message in this thread.

Since then, the thread has taken on a life of its own and ranged very
far from the initial, simple proposition of improved zero cross
detection.  There has been a flurry of comments mostly aimed not at
whether the "simple ZCD" is a good AC mains zero cross detector, but
more toward whether zero crossings are what grid-nuts should be
interested in in the first place.  Since I am not, myself, a
grid-nut, I cannot really speak to what grid-nuts "should" be
interested in.  I do think that time-stamped zero crossings have many
significant advantages when one is interested in comparing notes with
others, and it is comparatively easy data to collect with good
accuracy -- so, IMO, the choice of grid-nuts to settle on
time-stamped zero crossings was eminently rational.  The "simple ZCD"
has proven to be an excellent front end for such a data collection,
and is a project within the skills of anyone who knows which end of a
soldering iron to grip.  I am happy to answer any questions that
potential builders may have.

Personally, I think the thread has more than run its course and
should be laid to rest.  But if it is to continue, please accept as a
given that grid-nuts decided long ago that time-stamped zero
crossings are the appropriate data to collect, and focus on the
narrow topic of the "simple ZCD" as a means for accurately detecting
zero crossings of the AC mains.

Best regards,

Charles

Charles,

A commend regarding your ZCD. You propose to use a dual 120V primary
transformer to generate the isolated 120V AC needed by your circuit.
Unless specifically designed for that purpose, the isolation between the
two 120V primaries of a common transformer is probably not as good as the
isolation between primary and secondary, which could be a safety hazard.
Since small transformers with a 120V primary and a true 120V secondary are
hard to find, a better way would be to use two "regular step-down"
transformers back to back, like two door bell transformers: 120-24-120. You
would then get double isolation.

Didier KO4BB

On Sun, Dec 21, 2014 at 2:52 AM, Charles Steinmetz csteinmetz@yandex.com
wrote: