OF A SIGNAL

This invention relates to apparatus for, and a method of, detecting a cyclic variation of magnitude of a signal. The invention finds especial application in detecting the number of rotations of a rotating component of a commodity consumption registering meter.

In the field of commodity consumption registering meters, for example electricity, water or gas consumption registering meters, remote meter reading is desirable which requires an electrical signal which is related to the rate of consumption of the commodity. However, most gas, water and even electricity consumption registering meters do not have such a signal since the majority of these meters are mechanical. A feature that is common to these meters, however, is some form of mechanical indicator whose rate of rotation is related to the rate of consumption of the commodity such as, for example, cyclometer registers, dial registers or, in the case of an electricity consumption registering meter, a Ferraris disc.

It has been proposed to provide an optical pick-up arrangement which is attached to the consumption meter to convert this mechanical rotation into an electrical signal, by directing light onto the rotating component and detecting light reflected from it. The optical pick-up arrangement produces an electrical signal whose magnitude cyclically varies with time due to variations in the reflectance ofthe rotating component, e.g. due to a mark provided on the component. The period of the cyclic variation is equal to the

period of rotation of the component which depends upon the rate of consumption ofthe

commodity. To log the number of units of commodity consumed requires reliable detection of each cycle of the signal and it is known to set a magnitude threshold level which, when exceeded by the signal, indicates that one rotation of the component has occurred. However the signal magnitude, which depends upon the reflectance of the rotating component, will be different for each meter and the magnitude threshold level therefore has to be manually set for each individual meter which is time consuming and expensive. In addition the magnitude threshold level needs to be periodically reset as the components in the optical pick-up arrangement change with time and/or the surface of the rotating component degrades, further adding to the cost. A need exists, therefore, for apparatus which can automatically and reliably detect a cyclic variation of magnitude in a signal.

According to the present invention there is provided apparatus for detecting a cyclic variation of magnitude of a signal, the apparatus comprising: means for dividing the range of variation of magnitude of the signal, over at least one cycle of the signal, into a plurality of bands; means for counting the number of peaks and troughs of the signal magnitude occurring within each band over at least one cycle of the signal; means for setting a magnitude threshold level within a band with relatively few peaks and troughs and means for producing an output signal indicating the occurrence of the variation when the input signal magnitude crosses said magnitude threshold level.

In one particular application ofthe present invention the apparatus is used in combination

with means for producing the cyclically varying signal, the means comprising a light source arranged to direct light onto a rotating component and a light detector for detecting

light reflected from the component to generate said signal.

In a preferred arrangement the apparatus comprises means for setting a further magnitude threshold level in a different band with relatively few peaks and troughs and in which the means for producing an output signal produces an output signal when the signal magnitude crosses said magnitude threshold level and only subsequently produces an output signal when the signal magnitude crosses said magnitude threshold level after having crossed and recrossed said further magnitude threshold level. Such apparatus is less susceptible to noise, that is, the means for producing an output signal is not triggered if the signal magnitude oscillates around a given threshold value. With such an airangement it is preferred that the means for counting counts a peak or trough only if a change in signal magnitude of more than the width of a band occurs after the peak or trough.

In such applications the rotating component is typically the Ferraris disc of an electricity consumption registering meter.

According to another aspect of the invention there is provided a method of detecting a cyclic variation of magnitude of a signal comprising: dividing the range of variation of magnitude ofthe signal, over at least one cycle of the signal, into a plurality of bands;

counting the number of peaks and troughs of the signal magnitude occurring within each band over at least one cycle of the signal;

setting a magnitude threshold level within a band with relatively few peaks and troughs; and

generating an output signal indicating the occurrance of the variation when the signal magnitude crosses said magnitude threshold.

In a preferred embodiment the method further comprises setting a further magnitude threshold level in a different band with relatively few peaks and troughs and producing an output signal when the signal magnitude crosses said magnitude threshold level after having crossed and recrossed said further magnitude threshold level. Conveniently a peak or trough is counted only if a change of signal magnitude of more than the width of a band occurs after the peak or trough.

An embodiment of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

Figure 1 is a diagram of an electricity consumption meter, an optical pick-up and an apparatus in accordance with the invention;

Figure 2 shows (a) an input signal for the apparatus in Figure 1, and (b) a sampled and digitised form ofthe signal in (a);

Figure 3 is an enlarged portion of the plot in Figure 2(b);

Figure 4 illustrates a method for detecting peaks and troughs within the input signal of Figure 2; and

Figure 5 illustrates an alternative method for detecting peaks and troughs within the input signal of Figure 2.

Referring to Figure 1, there is shown an electricity consumption meter 21 of a type known in the art. The meter 21 illustrated is a Ferraris-type induction meter and includes a Ferraris disc 22 whose rate of rotation is proportional to the rate of consumption of electricity. A mark 23 on the edge of the disc 22 enables rotation of the disc to be observed. The total number of units of electricity consumed is indicated by a plurality of cyclometer registers 24 which are visible through a window 25 in the front ofthe meter 21.

An optical pick-up 26 is provided for producing an electrical signal 27 as the disc 22 rotates. The optical pick-up 26 comprises a light source 28 and photo detector 29. The light source 28 is arranged to direct light 30 onto the edge of the disc 22 and reflected light 31 from the edge is sensed by the photo detector 29.

Rotation ofthe disc 22 causes the mark 23 to modulate the light 31 reflected back to the photo detector 29 which produces an electrical signal 27 whose magnitude cyclically

varies with time in a manner as shown in Figure 2(a). Referring to Figure 2(a), there is shown a typical plot of input signal magnitude versus time for two complete rotations of the disc 22. A peak 40 in the input signal magnitude occurs each time the mark 23 passes the optical pick-up 26. In the example illustrated the disc 22 is rotating at a constant rate and the period of time between the peaks 40 is therefore constant though this is not necessary for operation of the apparatus 33 in accordance with the invention.

Referring to Figure 1 the input signal 27 is applied to an input 32 of the apparatus 33 according to the invention which detects cyclic variation in the input signal 27 and produces an output signal 34 for each rotation of the disc 22. The apparatus 33 comprises an analogue to digital converter 35 which samples and digitises the input signal 27 to produce a sequence of digital data sample values as shown in Figure 2(b). Referring to Figure 1, a microprocessor 36 processes these digital data sample values to determine the number of rotations of the disc 22 by detecting the number of peaks 40 using the method now described.

Referring to Figure 3, typical data for one complete signal cycle of Figure 2(b) is shown. The processor 36 is configured to determine the range, V _R ^^ _E , of variation of magnitude of the input signal 27 for at least one rotation of the disc 22, that is over at least one complete cycle of the input signal 27. The processor 36 divides at least this range, V _RANGE* ^{t0 a} plurality of magnitude bands, bands numbered 1-11 in Figure 3. It should be noted that the number of bands is arbitrary and depends upon the particular application to which the invention is being applied. Accordingly the invention should not be

construed as being Umited to eleven bands only. The processor 36 counts the number of peaks and troughs occurring within each magnitude band over at least one cycle of the input signal 27 and assigns an activity level to each band as indicated on the right hand side of Figure 3. For example, in band 1 two troughs occur and the activity is designated as 2, in band 2 one peak and two troughs occur and the activity is recorded as 3 and so on for the other magnitude bands. The method for determining the activity in each band is described below.

Having determined the relative level of activity within each band the processor 36 checks for bands in which no activity is recorded, that is bands in which no peaks and troughs occur. In Figure 3 this corresponds to bands 4-9. The processor 36 sets two magnitude threshold levels Va,, and V,,,* as indicated in Figure 2(b), in these bands of no activity.

In operation of the apparatus 33, when the input signal 27 crosses the first magnitude threshold level, V _ftl , the processor 36 continues monitoring the input signal magnitude until the input signal magnitude crosses the second magnitude threshold level, V „ _£ , whereupon it generates an output signal 34 indicating that the mark is passing the optical pick-up 26. No further output signal 34 is then generated until the input signal magnitude subsequently falls below the two magnitude threshold levels, V _ftl and V^, and recrosses them.

To make the apparatus 33 less susceptible to noise, that is reduce the likelihood of the processor 36 being falsely triggered if the input signal magnitude is oscillating between

the magnitude threshold levels, the processor 36 sets the magnitude threshold levels, V _Λ , and V _ft2 , as widely apart as possible in bands of no activity. Accordingly in the example

illustrated in Figure 3 the lower magnitude threshold level V _ώ , is set at the bottom of the lowest magnitude band of no activity, that is band 4, and the upper magnitude threshold level V _lh2 set at the top of the highest band of no activity, that is band 9. Whilst it is desirable to set the magnitude threshold levels as widely apart as possible they must not be set at levels which are too close to the mimmum or maximum signal magnitude as this could result in the apparatus failing to detect a peak 40 if the input signal magmtude were to become offset slightly such that the magnitude threshold levels were no longer crossed. Accordingly in an alternative arrangement the lower magnitude threshold level V _m , is set in the middle of the lowest band of no activity and the upper magnitude threshold level V^ set in the middle of the highest band of no activity.

Once the processor 36 has determined the position ofthe magnitude threshold levels, V _ftI and V^, it calculates what proportion of the maximum signal magnitude they represent and stores these calculated proportion values. These proportion values are used by the processor 36 to periodically reset the position of the magnitude threshold levels to their optimum position even though the condition of the disc 22 or sensitivity of the optical pick-up 26 may degrade over a period of time. The processor 36 resets the magnitude threshold levels by measuring the maximum signal magmtude and sets the magnitude threshold levels according to the stored proportion values. In an alternative arrangement the processor 36 is configured to periodically reset the threshold levels by measuring the activity in each band using the method described above.

Referring to Figure 4 the method by which the processor 36 detects peaks and troughs is illustrated. At time t, the input signal magnitude is in band 1 and the processor 36 stores the band number 1. The input signal magmtude increases such that it lies in band 2 at

time t ₂ As a change of band has occurred the processor 36 also stores the current band number 2. The processor 36 at any given time maintains a record of the band numbers for the previous two band changes. At time t ₃ the input signal magmtude has increased into band 6 and the processor 36 compares the current band number with the previous two band changes, that is 2 and 1. The fact that the previous two band numbers were both lower indicates that no peaks or troughs have occurred since the previous two band changes. As the current band number 6 is different from the band number for the previous band change the processor 36 updates the record of the band number for the previous two band changes, that is it is updated to 6 and 2. At time t ₄ the signal magnitude remains in band 6 and the processor 36 compares this band number with the previous two band changes (6 and 2). Since there is no change in band number from the previous band change this indicates that no peaks or troughs have occurred. As no band number change has occurred the processor 36 does not update the record of the band numbers for the previous two band changes. At time tj the signal magnitude falls to band 5. As the band number 6 for the previous band change is higher than the current band 5 and the band number 2 for the band change prior to that was lower, this indicates that a peak in input signal magmtude occurred in band 6. The processor 36 increments the activity count for band 6 and updates the record ofthe band numbers for the previous two band changes. Similarly the processor also detects a peak in band 2, which occurs around time t ₇ and increments the activity count for this band.

The processor 36 also detects troughs using the current band number and previous two band changes. For example at t ₇ the signal magnitude lies in band 2 whilst the previous two band changes were band 1 and band 5 respectively, indicating that a trough occurred in band 1 and the activity count for the band is incremented. Likewise a further trough in band 1, which occurs around time t _g , is detected.

By comparing the current band number with the band numbers for the previous two band changes the processor is able to determine the number of peaks and troughs and the bands within which they occur. Thus for the portion of input signal illustrated in Figure 4 the activity count for bands 1 to 6 will be 2, 1, 0, 0, 0 and 1 respectively.

It will be appreciated that the method described will only detect peaks and troughs when the input signal passes into and out of a band. The method will not detect peaks and troughs which occur wholly within the band, as indicated in Figure 4 at times 1 ₄ and t ,, since such peaks and troughs do not result in a band number change. Failure to detect these peaks and troughs does not result in false triggering of the apparatus 33 since the two magnitude thresholds are set in different bands.

It is possible that the input signal magnitude could oscillate around the boundary between two adjacent magmtude bands which would result in a high activity count for those magmtude bands. As a result no magnitude thresholds would be set in these magnitude bands. However as described above the use of two thresholds ensures that the apparatus 33 will only detect a change in input signal magnitude which crosses and

re-crosses both magnitude thresholds V _Λ1 and V _m> that is a change of magnitude of at least V _dtf -V _t ø,, which difference will be at least one magnitude band since the magnitude

thresholds are set in different magmtude bands. Consequently when using two magnitude thresholds it is only necessary to register peaks and troughs if a change of signal magnitude of more than the width of a magnitude band occurs between consecutive peaks

and troughs, since smaller changes will not falsely trigger the apparatus 33. In an alternative arrangement the processor 36 is configured to determine the activity in each magnitude band using the following method which only registers peaks and troughs if a change in signal magnitude of more than width of a magnitude band occurs between peaks or troughs.

As described above the processor 36 utilises the current band number with the band numbers for the previous two band changes to determine if any peaks or troughs have occurred in the signal magnitude, however the method of updating the record of the band number for the previous two band changes is modified as will now be described by way of reference with Figure 5.

Referring to Figure 5, at time t* the signal magnitude lies in band 6 and the band numbers for the previous two band changes are bands 3 and 1 respectively. Since the previous two band numbers are lower than the current band number this indicates that no peaks or troughs have occurred. As the current band number 6 is greater than the previous band numbers 3, 1 the processor 36 updates the record of the band numbers for the previous two band changes with the current band number, that is it updates it to 6 and 3.

At time t ₄ the signal magnitude still remains in band 6 indicating that no peaks or troughs have occurred. Since no change of band number has occurred the record for the previous two band changes is not updated.

At time t _j the magnitude of the input signal has fallen to band 5. The processor compares the current band number 5 with the previous band number 6 and establishes that there has been a fall in band number. Whenever a fall in band number occurs the processor 36, in applying its test to establish whether any peaks or troughs have occurred, uses an effective band number rather than the current band number. The effective band number is one greater than the current band number; in the example therefore the effective band number is 6. The effective band number is compared with the previous two band changes being 6 and 3 respectively to establish that no peaks or troughs have occurred. Since the effective band number 6 is the same as the previous band number, the record is not updated.

At time t^ the current band number is 4. The processor 36 compares this number with the band number of the previous band change, that is 6 and determines that a fall in band number has occurred. Accordingly the processor 36 calculates an effective band number, which is one greater than the cunent value, that is 5 and compares this value with the numbers of the previous two band changes 6 and 3 and determines that a peak occurred in band 6. Since the effective band number 5 is different to the previous recorded band number 6 the processor 36 updates the record with the effective band number, such that the record now contains the band numbers 5 and 6.

At time t ₇ the signal magnitude has risen into band 5. Since there is no change of band number compared with the previously recorded band number, this indicates that no peaks

or troughs have occurred and the record is not updated. Although a trough did in fact occur in band 4, as can be seen in Figure 5, this is not registered by the processor 36 as

the change of magnitude since the trough at time tg is less than the width of a magnitude band.

At time t ₈ the signal magnitude falls to band 4. Since a fall has occurred the processor 36 calculates an effective band number 5 and compares this with the band numbers for the previous two band changes 5 and 6. As the effective band number 5 is the same as the previous recorded band number, this indicates that no peaks or troughs have occurred and that there is no need to update the record of band changes. Again even though a peak actually occurs in band 5 this is not registered by the processor 36 since the change of signal magnitude since the peak at time V, is less than the width of a magmtude band.

At time , the signal magmtude falls to band 1. Since the current band number is lower than the previous recorded band number 4 an effective band number is calculated to determine that no peaks or troughs have occurred. Since the effective band number is different from the previous band number the processor 36 updates the record with the effective band number such that the record contains the band numbers 2 and 5.

Finally at time t ₁₀ the signal magmtude rises into band 3. The processor 36 on detecting

a rise in band number compares the current band number with the previous two to establish that a trough has occurred. Since the effective band number is stored for the previous band change as opposed to the actual band number, the trough actually occurs in a band which is one lower than that recorded, that is in band 1. Accordingly the processor 36 increments the activity count for band 1.

It will be appreciated therefore that by using this method peaks or troughs are only registered if a change in signal magnitude of more than the width of a magnitude band occurs between peaks and troughs.

In summary the method comprises checking if the current band number is lower than the band number for the previous band change. If it is, an effective band number, which is one greater than the current band number, is compared with the band numbers of the previous two band changes to determine if any peaks or troughs have occurred. Wherever a trough is detected the activity count for the band number which is one lower than the previous recorded band number is incremented. It will be appreciated, however, that other methods of registering a peak or trough, only if a change in signal magnitude of more than the width of a magnitude band occurs after the peak or trough, can be employed.

As stated above the magnitude threshold levels V„,,and V^are set within bands where no peaks or troughs occur. If no bands with no activity are found, or only a single band of no activity is found, the processor 36 reduces the width of the bands, that is it increases

the number of bands and repeats the procedure of determining the activity count to try to find magnitude bands with no activity. If the processor 36 is still unable to find any bands with no activity it sets the magnitude threshold levels in the magmtude bands with the lowest activity.

If the processor 36 finds a number of magnitude bands with no peaks and troughs it is configured to determine if any of the magnitude bands are adjacent to each other forming a group of magnitude bands of no activity. The processor 36 sets the threshold levels, V _Λ ι and V^, in the widest two group of adjacent magnitude bands of no activity.

Whilst in the example above two magnitude threshold levels V^, and V _Λ2 are described, it has been found that a single magmtude threshold works as well when the signal peaks 40 are well defined as is the case for a clean disc 22 with a clearly defined mark 23. When only a single magnitude threshold level is used the processor 36 sets it in a magnitude band or group of magnitude bands with no activity using the method described above.

Of course variations may be made without departing from the scope of the invention. Thus, whilst in the embodiment ofthe invention described the input signal is sampled and digitised and all processing carried out using digital processing, it is also envisaged that analogue circuitry could be used. Further it will be appreciated that other algorithms can be used to detect for the presence of peaks and troughs.

It has been found that the apparatus is able to detect the cyclic variation in the signal even for a disc with no deliberately applied mark, since the disc always has some features on its edge, such as scratches or dirt, which produce a cyclically varying signal. Accordingly it will be appreciated that although the example described has been for detecting rotation of a Ferraris disc of an electricity consumption meter the invention is equally suited for detecting cyclic variation in any signal irrespective of how the signal is generated.