GB2432432A

GB2432432A - Correcting timekeeper with oscillating signal

Info

Publication number: GB2432432A
Application number: GB0523363A
Authority: GB
Inventors: John Stuart Calvert
Original assignee: Polymeters Response International Ltd
Current assignee: Secure Meters UK Ltd
Priority date: 2005-11-16
Filing date: 2005-11-16
Publication date: 2007-05-23
Anticipated expiration: 2025-11-16
Also published as: GB0523363D0; GB2432432B; AU2006236020B2; AU2006236020A1

Abstract

A timekeeping apparatus, suitable for use with a utility meter, comprising a first clock which is preferably a crystal based real time clock, and a means for determining a correction based on an oscillating signal, which is preferably a frequency regulated mains supply. The correction is applied to the time of the first clock. The oscillating signal may be a frequency regulated alternating voltage from a mains supply. A suppression in the short term component in the applied correction may be provided by a proportional feedback of the error.

Description

TIMEKEEPING APPARATUS

The present invention relates to a timekeeping apparatus, in particular for use in a utility meter, and to a method for correcting the time in a clock.

In a typical utility meter, the internal time kept by the meter and used in processing functions is kept on the meter's real time clock which is usually based on a crystal oscillator. Where the crystal is used to keep the time, the short term accuracy is good, but the time can drift over longer periods, resulting in poor long term accuracy. The drift will increase over time in an unbounded manner and needs to be reset in order to maintain the accurate time.

Another way of monitoring the time in a utility meter is to use a mains clock which is the preferable method in the United Kingdom for example. This type of method operates by counting the number of mains frequency cycles assuming that each cycle has a time period equal to the exact time period of the nominal supply frequency. This is possible in such cases where the electrical grid frequency is controlled by the generating companies to produce a fixed number of cycles over an extended time period (usually a day); such count of cycles being equal to that would be produced by a perfectly stable nominal frequency system in the extended time period. A drawback of this system is that whereas this type of system may have good and bounded long term accuracy, the short term accuracy will be poor. Moreover, in the absence of mains power, the clock will also be switched off and have to be maintained by other non volatile means during this period.

Some current systems base their time primarily on a mains clock and revert to the crystal oscillator clock when power fails. However, this type of combination results in an inaccurate time in the short term due to perturbations of the grid frequency. It is therefore desirable to provide a meter with improved accuracy in both the short term and the long term.

The present invention provides a timekeeping apparatus comprising a first clock, a second clock and means for periodically updating the time of the first clock on the basis of the time of the second clock. The second clock may be derived from an external oscillating signal. The oscillating signal is typically a frequency regulated alternating voltage source, such as a mains supply.

Furthermore, the present invention provides a means of suppressing the short term component of the second clock in the applied correction such that the clock so derived has property of high short term accuracy of the first clock and high long term accuracy of the second clock signal.

Another aspect of the invention provides a utility meter incorporating the above described timekeeping apparatus. One of the clocks may be a battery backed crystal controlled clock with high short term accuracy and unbounded long term accuracy, and the second clock may have poorer short term accuracy than the first clock, but bounded long term accuracy. Particularly, in the case of a utility meter, the second clock may be derived from a mains frequency counter.

In order that the present invention be more readily understood, embodiments thereof will be described with reference to the accompanying drawings, in which: Fig. 1 shows a timing graph comparing the proportioned and integral feedback signals applied to a real time clock signal A using a perfect mains frequency signal; Fig. 2 shows a timing graph of a real time clock correction signal when the proportional and integral feedback signals are combined and applied to the real time clock signal A; Fig. 3 shows a timing graph for an imperfect main frequency signal wherein the correction signal uses proportional and integral feedback.

In a preferred embodiment of the invention, the timekeeping apparatus comprises a crystal based real time clock which keeps the internal time of a utility meter. Such a clock is powered by a battery and considered to have good short term accuracy but unbounded long term accuracy. The error will increase over a long period of time because of the steady monotonic drift (e.g. 0.5 second per day).

A second clock in the form of a mains frequency clock is also provided and in contrast to the crystal based clock, it has a large short term error, but the error is often bounded. For example, the clock error may be +2 seconds at 8 AM in the morning, -7 seconds at 5 PM and again +1 second at 6 AM the next day.

In order to correct the drift in the crystal based internal clock, the controlled variable is made to follow the mains clock, the command variable, by means of a servo control loop. This is shown in Figs. 1 and 2. The servo control loop applies a correction to the controlled variable A, in this case the meter's internal crystal based time, in accordance with an error, which is the difference between the controlled variable (the internal crystal time) and the command variable (mains time).

As shown in Fig. 1, if the correction applied by the servo control ioop is proportional to the error, the controlled variable will be corrected more as the error gets larger. The amount of error that is needed to correct any given rate of drift of the time is proportional to the rate of drift. This means that if the drift is non-zero, the controlled variable will not equal the command variable, as shown by corrected signal Al. If the correction applied by the servo control loop is proportional to the integral of the error, any persistent error is corrected by an increasing rate of correction. Clearly, this type of behaviour is needed to eliminate the permanent error that is otherwise needed to correct a persistent drift. By a suitable choice of the gains that are applied to the proportional and integral errors in the servo control feedback, the controlled variable can be made to follow the tracked variable with a damped oscillatory error A3. The period of the oscillations is determined by the integral gain (a higher integral gain gives a shorter period), and the damping rate is controlled by the proportional gain (a higher proportional gain gives more damping).

Once the following error has been reduced to zero, the rate of drift is determined entirely by the integral error A2. The integral error can therefore be initialised according to a measured frequency error, to reduce the time needed for the servo feedback to converge on the correct value, thereby effectively anticipating the drift in the meter's internal time. The integral error can also be used to calculate the amount of drift there has been while the power is off, so that compensation for this can be applied when the power is restored.

Another embodiment is shown in Fig. 3 where the combined proportional and integral feedback is applied by an imperfect mains frequency signal. In the UK, for example, the mains supply frequency averages out to about 50Hz over a period of 8 days. This is taken into account by adapting the correction signal A4 to have a first half-period of 8 days, thus allowing time for the frequency to average out and give a more accurate correction. Oscillations can be reduced by introducing a damping factor of for example 50%.

Accordingly, the preferred way to obtain a stable clock is to use a proportional-integral servo loop with appropriately chosen gain and rate so that any errors are quickly settled to zero.

Claims

CLAIMS: 1. A time keeping apparatus comprising a first clock, and means

for determining a correction based on an oscillating signal and applying the correction to the time of the first clock.

2. The apparatus of claim 1 wherein the first clock exhibits short term accuracy of a first quality but unbounded long term drift and the oscillating signal exhibits short term accuracy poorer quality than that of the first clock but bounded long term accuracy.

3. The apparatus of claim I or 2 wherein the first clock is a crystal based real time clock and the oscillating signal is a frequency regulated alternating voltage source.

4. The apparatus of claim 1, 2 or 3 wherein the oscillating signal is derived from a frequency regulated mains supply.

5. The apparatus of claim 1, 2, 3 or 4 further comprising a means for providing a suppression of the short term component in the applied correction.

6. The apparatus as in claim 5 wherein the suppression of the short term component is achieved by means of a proportional feedback of the error.

7. The apparatus as in claim 5 wherein the suppression of the short term component is achieved by means of a proportional and integral feedback of the error.

8. A utility meter comprising the timekeeping apparatus according to any preceding claim.