Accuracy of UK power grid time control?

Can't say about the UK but in the US the clocks that run only on the power grid and depend on the frequency are very accurate over a long period of time. I would say more like less than a minuit or less over a year period of time if the clock its self is up to it. During periods of peak loads the nominal 60 hz may go down a cycle or so and when the load is taken away the power company will raise the frequency long enough to get the clocks back in time.

In other words the power company keeps the frequency to an average of exectally 60 hz over a long period of time. Much more stable than any clock you could build that depends on a quartz chip.

Couldn't find anything on the UK grid, but the US grid short-term is generally 10 milliHz or better and (probably) averages a few milliseconds per year long-term. Take care to accommodate local line noise because it can trip digital counters; even cheap consumer digital clocks sometimes don't use line frequency directly for this reason, but use a loosely coupled PLL instead.

Used to be the case that frequency was governed to certain tolerances and number of cycles turned out in a 24 period was mandated to be exactly

50*24*3600, with time keeping in mind.

Not sure what the position is since privitisation.

Tim

During periods of high demand, the frequency will drop slightly. To compensate, the frequency will later be increased so that there are the same number of cycles per day, so that the long term accuracy of synchronous electric clocks is very good. If you don't have a seconds hand, no-one will even notice the short term inaccuracies.

A quick look at the graphs suggests that the maximum deviation is about 0.2 / 50 (2.5 seconds in 10 minutes), the period of low frequency lasts in the order of ten minutes, and corrections are made immediately after the dip.

--
Alan J. Wylie                                          http://www.wylie.me.uk/
"Perfection [in design] is achieved not when there is nothing left to add,
but rather when there is nothing left to take away."
  -- Antoine de Saint-Exupery

300s/30Ms = 10ppm. 300s/15Ms = 20ppm. That's not too taxing. IIRC, maxim/dallas do some chips that may suit, for way, way less than 40 quid.

My understanding is that the frequency is allowed to fluctuate by a small amount (don't know how much) but that, over a 24 hour period, exactly the right number of cycles will be delivered. So your clock should at least be exactly right once per day.

--
Cheers,
Roger
______
Please reply to newsgroup.
Reply address IS valid, but not regularly monitored.

if the hands are balanced it should not make much difference how long they are

As a matter of interest...when you see those pictures of the control room, there are two clocks near top right. One is grid driven, the other is 'crystal' driven (probably rather better than that, these days). They

*ought* to read the same...

--
The information contained in this post is copyright the
poster, and specifically may not be published in, or used by
Avenue Supplies,   http://avenuesupplies.co.uk

uses

that

books,

for

to 50

determine a

every

grid,

much

Chris,

OK it's going back a few years, but when the CEGB had their National Grid Control Centre at Park ST London SE1 the number of cycles per day was very accurately ensured to be correct (A pair of Ferranti Argus

500 Process Control computers each had an ultra accurate crystal clocks in them feeding displays in the control room) and the Control Enginners could let the frequency drift a tad hour by hour but had to get it right over 24. They dispersed the control to various regional centres (Winnersh, St Albans and three others I cannot remember but I think that the principle remains the same.

AWEM (who in a past life occassionally sweated blood over those computers!)

It really depends where you are - if you are in a rural area with above ground lines and lots of ice storms, hurricanes, etc. then not too good. In an urban area with buried lines, power may go uninterrupted for years at a time.

If it's critical that the clock not stop you can put it on a UPS battery backup.

Hi all,

I'm thinking of building an electronic clock control circuit which uses the 50 Hz mains frequency for time keeping. The reason for this is that the clock dial is rather large, so probably wouldn't run for long on battery power, and I don't fancy spending £40 buying a programmable quartz oscillator chip.

Despite doing some research online and in electrical engineering books, I can't find a figure for the accuracy of the time keeping of the UK power grid. Perhaps this is because there aren't official bounds set for the time error - one of my electrical engineering books says it is a legal requirement that the frequency be kept "as close as possible to 50 Hz" - but even if this is the case it should be possible to determine a typical error figure.

From my point of view I'd regard an acceptable error as 5 minutes every

6 months. Does anyone know the typical time error seen on the UK grid, or where I might find this information? Any suggestions would be much appreciated.

Best wishes,

Chris Tidy

It's weird. Frequency is allowed to wobble a bit, but it has to average out very accurately over 24 hours or so, because of the number of clocks in service.

Other people have given the rules that are applied (i.e. that the total number of cycles in 24 hours is constant), but the reason is for efficienty of electricty transfer. It is very important that all generators run in sync, so as to minimize transmission losses in the National Grid, AFAIUI.

Paul

I probably should have mentioned. If this is outside. You can get GPS, for about that price, and if it's a clockface, you can mount the antenna there, and have it work just fine.

Well, you got the basic idea right. But it never drifts down 1 cycle/second. Very rarely drops even a tenth of a cycle.

Indeed. Part of the standard equipment in the old days was a special 'crystal oven' with tightly controlled temperature. By regulating the temperature of the crystal inside, the accuracy its vibrations was improved.

An old 'urban lengend' was that the first quartz watches were calibrated assuming the temperature of the crystal was going to be controlled by the body heat of the wearer. And that leaving your watch on the dresser over the week-end would cause it to slow down slightly. Don't know if it is really true, but it's a nice story.

daestrom

In the 'old' days; mains powered clocks (squirrel cage, synchronous) were common. Many mantel shelves had a two pin round outlet fixed above them for the sole purposes of plugging in an electric clock. Although the _instantaneous_ frequency stability of the mains supply is (relatively) wide; the aggregation of time periods (inverse of frequency) over the course of the day results in a discrepancy that was/is tiny. IIRC, power station control rooms had a red finger which was slaved to a pendulum/GPO time source while the same 'meter' showed the integration of the 50Hz periods - aka 'Electric Time'. It was a responsibility of the Control Room staff to keep 'Electric Time' consonant with 'GPO Time': this could be achieved by altering the instantaneous frequency (speeding up/ slowing down ).

However; as Argos flogs Rugby-synchronised wrist watches for ten-quid-odd, why bother? PS, the 'Klick' wristwatch not only adapts to GMT/BST changes but it functions as a depth-gauge: - if it stops working I must be deeper than fifty feet - but at least I'll have the facility to observe TOD !

--

Brian

There's nothing "urban legend" about that. To this day, most quartz watch circuits are not temperature compensated (and obviously there is not enough power available to put the crystal in an "oven" in a wristwatch) so their timekeeping will vary slightly based on temperature. Typically a manufacturer will pick some midpoint between body temperature and room temperature (I have seen 31C used as at typical #) and use that as the temperature at which their movements are calibrated at the factory. If the actual operating conditions vary from that temperature, the watch will drift slightly from the calibrated rate, but it's no big deal.

The older consumer quartz wrist watches were a little temperature sensitive. Wearing them could effect their accuracy a little. Whether or not it went faster or slower, depended on the temperature coefficiency of the particular quartz crystal and components in the watch and how the local oscillator in the watch was designed. The manufactures of the higher end watches tried to have their calibration set up for the watch to be at the average body surface temperature for approximately 14 hours per day, and at room temperature for approximately 10 hours per day.

The newer watches are improved to a great extent. Many of the new watches will keep an average time of better than about 5 to 8 seconds per month. There are some very expensive models where they will guarantee an average accuracy of better than 2 to 5 seconds per month. The older watches going back more than about 8 to 10 years ago were usually not much better than 15 to 20 seconds per month.

--

JANA _____

Well, you got the basic idea right. But it never drifts down 1 cycle/second. Very rarely drops even a tenth of a cycle.

Indeed. Part of the standard equipment in the old days was a special 'crystal oven' with tightly controlled temperature. By regulating the temperature of the crystal inside, the accuracy its vibrations was improved.

An old 'urban lengend' was that the first quartz watches were calibrated assuming the temperature of the crystal was going to be controlled by the body heat of the wearer. And that leaving your watch on the dresser over the week-end would cause it to slow down slightly. Don't know if it is really true, but it's a nice story.

daestrom

For any type of accuracy, depending on the power grid is not a dependable option! There can be noise, interference, and power dips. On the short term, the power grid will be accurate, but over the long term, it can be out by a fair amount.

I have an electric clock that is dependent on the power grid for its timing. This clock can be a few minutes or a few seconds out at times.This is not from the power company being off frequency. It is most likely caused by power dips, or noise in the power delivery.

The accuracy of the power companies timing is very accuracy. It has to be in order to keep synchronized to other systems that they are working with. They have many generating facilities that are working together, therefore all their systems have to be synchronized to within fractions of a degree on the line frequency. From what I am told, here in North America, their margin of error is within about

1 second per month.

The problem when connecting a device at home to the power line, such as a clock, the clock can jump off time very easily. It is susceptible to interference, and power disturbance through their distribution. I was also told this by an engineer from our local power company.

Normally, during peak conditions, they allow the frequency to drop very slightly. It may be as much as 0.5 to 1 Hertz or so. During peak conditions, they will give it back. This means that all the power generation systems working together must also drop, and increase by the same amount.

The clocks that I have that are referenced to the AC line, are not dependable for accurate time. Over a few weeks, I found the quartz clocks to be more accurate. I can check them with the NIST broadcast.

Here in North America many stores are selling quartz clocks with built in time receivers. These are getting the time reference from the NIST. The user only has to put in the approximate time to within about 30 minutes. After 12 hours, the clock will be exactly on time, as long as it can receive the NIST signal. The clock automatically checks in with the NIST about every 12 hours. If the NIST signal is unavailable, the clock will keep time to the spec of any consumer type quartz clock. This is about 5 to 10 seconds per month.

Another thought would be to get a low cost GPS that can accept an external antenna and external power supply. This will give you the most accurate possible time for home use.

You can also go to the NIST site with your computer, and get very accurate time. It should be within about 100 to 200 ms. Going from memory, I believe it will tell you its error to your computer. By hitting the reset key, you can sometimes get it down to within a few milliseconds.

--

JANA
_____


"Christopher Tidy"  wrote in message
news:443FBF5B.4090101@cantabgold.net...
Hi all,

I'm thinking of building an electronic clock control circuit which uses
the 50 Hz mains frequency for time keeping. The reason for this is that
the clock dial is rather large, so probably wouldn't run for long on
battery power, and I don't fancy spending £40 buying a programmable
quartz oscillator chip.

Despite doing some research online and in electrical engineering books,
I can't find a figure for the accuracy of the time keeping of the UK
power grid. Perhaps this is because there aren't official bounds set for
the time error - one of my electrical engineering books says it is a
legal requirement that the frequency be kept "as close as possible to 50
Hz" - but even if this is the case it should be possible to determine a
typical error figure.

 From my point of view I'd regard an acceptable error as 5 minutes every
6 months. Does anyone know the typical time error seen on the UK grid,
or where I might find this information? Any suggestions would be much
appreciated.

Best wishes,

Chris Tidy