METHOD OF AND APPARATUS FOR LEARNING THE PHASE ERROR OR TIMING DELAYS WITHIN A CURRENT TRANSDUCER AND POWER MEASUREMENT APPARATUS INCLUDING CURRENT TRANSDUCER ERROR CORRECTION

Abstract

Current transducers are widely used in current measuring systems. They provide good isolation between the supply voltage and the measurement equipment. However they can introduce small phase errors which can become significant sources of error if the current to a load is out of phase with the supply voltage for the load. This disclosure discusses a robust measurement apparatus and method that can be used in situ to monitor for and correct phase errors.

Claims

1. A method of estimating phase measurement errors in measurements of current or voltage, the method comprising: providing an input signal having a fundamental frequency to an input of a current transducer; receiving an output signal from the current transducer, and analyzing the output signal to determine a phase difference compared to the input signal; wherein the method further comprises applying a correction to the estimate of phase measurement error compensate for a finite rate of change of the input signal.

2. A method as claimed in claim 1, in which the input signal is a repeating signal.

3. A method as claimed in claim 1, in which the input signal is a stepwise approximation to a continuous signal.

4. A method as claimed in claim 3, in which the continuous signal is one of a sinusoid, a triangle wave, a square wave with smoothed transitions, and a bandwidth limited noise source.

5. A method as claimed in claim 2, in which the input signal is a square wave like signal, and the correction comprises making a compensation for the finite rate of change of edges of the square wave like signal when transitioning between high and low values.

6. A method as claimed in claim 2, in which the current transducer comprises a current transformer, and the repeating input signal approximates a slew rate limited square wave or a charge rate limited square wave.

7. A method as claimed in claim 6, in which the slew rate or charge rate limited square wave transitions between first and second values, and where if a transition from the first value to the second values commences at a time T1, and finishes at a time T2, an estimate at a time to reach a midpoint is made to proceed a first phase correction.

8. A method as claimed in claim 6, in which the time to reach the midpoint is formed by starting a counter or timer at the beginning of the transition and stopping it when the end-point value is reached.

9. A method as claimed in claim 7, further comprising adding a second phase correction to account for non-linearly of the rate of transition between the first and second values.

10. A method as claimed in claim 9, in which the second correction is estimated or measured at manufacture and stored in memory.

11. A method as claimed in claim 1, in which the transducer is a current transformer and the phase difference after applying the correction signal represents a phase shift resulting from the current transformer.

12. A method of estimating power consumption comprising measuring the potential at a first conductor, measuring the current flowing in the first conductor, applying a phase correction to the measurement of current using the method of claim 1, and multiplying the potential and current measurements to estimate power.

13. A method as claimed in claim 12, in which the correction to the measured current signal comprises estimating and applying a time shift to be applied to sampled values of the current compared to sampled values of voltage.

14. A method as claimed in claim 6 in which the slew rate or charge rate limited square wave has a substantially equal mark-space ratio.

15. A method as claimed in claim 1 in which the phase difference is estimated using a ITT or a Geortzel algorithm, or a phase detector circuit.

16. A method as claimed in claim 1 in which the input signal has a predetermined slew rate or transition time and is formed by a digital to analog converter and the corrections required to account for the finite rate of change of the input signal are known because the slew rate or transition time of the input signal is predetermined.

17. An apparatus comprising means for performing the method of claim 1.

18. An apparatus of estimating phase shifts in measurements of current, the apparatus comprising: a signal generator for providing an input signal to an input of a current transducer, and a phase or time shift comparator for receiving an output signal from the current transducer and analyzing the output signal to determine a phase or time difference compared to the input signal, and the phase comparator is arranged to apply a phase correction to compensate for a finite rate of change of the input signal.

19. An apparatus as claimed in claim 18 further comprising a circuit for measuring the transition time in the input signal due to a finite rate of change of the input signal,

20. An apparatus as claimed in claim 18, in which the signal generator is a square wave generator.

21. An apparatus as claimed in claim 20 in which the measurement circuit estimates the midpoint of the square wave transition and provides a timing signal to the phase comparator.

22. An apparatus as claimed in claim 18, in which the input signal is an approximation to a square wave but having ramp like transitions, the input signal is formed by a digital to analog converter such that the duration of the ramp like transitions are known or predetermined, and a correction value to account for the duration of the ramp like transition is known or predetermined.

23. A power meter including an apparatus as claimed in claim 18.

24. A power meter as claimed in claim 23 further including an interface for sending data back to a network operator, where the data includes estimates of performance of the power meter and/or information about load and voltage conditions at the power meter.

25. A power meter as claimed in claim 23 further arranged to account for harmonic signals in the current when calculating the power drawn.

26. A power meter as claimed in claim 23 in which the signal generator generates a repeating signal having a fundamental frequency and the frequency of the signal generator is adjustable.

27. A method of estimating phase measurement errors in measurements of current, the method comprising: providing a repeating input signal having a fundamental frequency to an input of a current transformer, where the repeating signal has nominally linear rising and falling edges; receiving an output signal from the current transducer, and analyzing the output signal to determine a phase difference compared to the input signal; wherein the method further comprises applying a correction to the estimate of phase measurement error compensate for a finite rate of change of the edges of the input signal, and the fundamental frequency of the input signal is adjustable such that the phase measurement error can be estimated at one or more frequencies.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Embodiments of the present disclosure will now be described, by way of non-limiting example only, with reference to the accompanying Figures, in which:

[0027] FIG. 1 schematically illustrates the data acquisition channels of an electronic watt-hour meter;

[0028] FIG. 2 schematically illustrates a current transformer;

[0029] FIG. 3 shows the circuit diagram representation of the device shown in FIG. 2;

[0030] FIG. 4 is a graph showing the percentage error in power measurement for a 1 degree and 2 degree error in estimating the current to voltage phase shift θ for various power factors;

[0031] FIG. 5 is an equivalent circuit for the current transformer of FIG. 2;

[0032] FIG. 6 is a simplified equivalent circuit;

[0033] FIG. 7 schematically illustrates first embodiment of the present disclosure;

[0034] FIG. 8 schematically illustrates a second embodiment of the present disclosure;

[0035] FIG. 9 schematically illustrates the waveform of a drive signal used within embodiments of the present disclosure;

[0036] FIG. 10 illustrates one embodiment of a circuit for detecting the time for the square wave to transition between high and low currents;

[0037] FIGS. 11a to 11c illustrate waveforms within the circuit of FIG. 10;

[0038] FIG. 12 is a circuit diagram of a current modulator that may be used in an embodiment of this disclosure;

[0039] FIG. 13 is a circuit diagram of a Watt-hour meter in accordance with the teachings of this disclosure;

[0040] FIG. 14 is a flow diagram of a method of estimating phase measurement error;

[0041] FIG. 15 is a flow diagram of a method of correcting for phase measurement error;

[0042] FIGS. 16a to 16d show how the ideal square-wave waveform may become distorted;

[0043] FIG. 17 is a schematic diagram of a measurement system to indicate how delays may build up; and

[0044] FIG. 18 is a schematic diagram of a further embodiment of this disclosure.

DESCRIPTION OF SOME EMBODIMENTS OF THE DISCLOSURE

[0045] It is often desirable to measure electrical parameters, such as voltage supplied to a load and/or current supplied to a load. In order to provide a more accurate assessment of power drawn, based on actual voltage and current as opposed to assuming a nominal voltage and a sinusoidal load current, it is becoming known to use digital meters.

[0046] FIG. 1 shows, in schematic form, the major components in the data acquisition channels for a digital power meter. Power from a power supply such as a mains supply is provided to a load. The load may be a single phase load or a multiple phase load. For the or each phase, the load voltage is measured by a suitable voltage sensor 2 and the line current for the or each phase is measured by a suitable current sensor 3.

[0047] The output from the voltage sensor 2 (or sensors in a multiphase system) may be passed through a filter 4 to limit the signal bandwidth to an appropriate range, for example to avoid aliasing, and then to an analog to digital converter 5. The analog to digital converter (ADC) may be associated with a programmable gain amplifier. Similarly the output from the or each current sensor 3 may be passed through a suitable filter 6 and the digitized by an ADC 7 which may be associated with a programmable gain amplifier.

[0048] The voltage sensor is often a potential divider so the response of the potential divider should be rapid, i.e. it does not introduce a processing delay or phase error. The anti-alias filter will introduce a phase delay, and although this should be well determined, tolerances during manufacture mean that its delay is unlikely to be precisely known as its cut-off frequency will only be approximately known. The ADC will also introduce a delay, but as this is a digital component its performance is determined by system clocks.

[0049] For the current measurement channel similar comments to those above apply to the filter 6 and ADC 7. However, the current measurement sensor may introduce a further phase error/delay depending on the current measurement technology used. Shunt resistors introduce no phase delay, but have the disadvantages of having to be placed into the supply path. On the other hand current transformers can be placed around a conductor in situ and have excellent isolation properties. They do however introduce phase delays.

[0050] In order to set the present invention in context, it is useful to consider the operation of a current transformer. FIG. 2 schematically illustrates the components of a current transformer. In essence, a conductor 10 carrying an alternating current to be measured acts as the primary winding of the current transformer. A secondary winding 12 is magnetically coupled to the primary winding 10. The secondary winding 12 may be wound around the primary winding 10 or may be wound around a core 14 which magnetically couples to the primary winding 10. By their very nature, current transformers provide good isolation between the conductor 10 and the secondary winding 12. They also exhibit minimal influence on the conductor 10 and if the core 14 can be split, then the current transformer can be inserted around the primary conductor 10 without disrupting it.

[0051] The effective turns ratio between the primary and secondary is normally specified by the ratio of current flowing through the primary to the current output by the secondary. A transformer having a ratio of 1000 to 1 would output 1 amp from the secondary for every 1000 amps flowing through the primary. The transformers may be tapped to allow the measurement circuitry following the transformer to operate over a larger current range. A physical device shown in FIG. 2 can be represented by the circuit diagram of FIG. 3.

[0052] As noted earlier, a consumer needs to be accurately charged for the amount of power they use. Overcharging is not acceptable to regulating authorities and undercharging represents a potentially large loss of revenue. One significant problem is that the effect of even small phase errors can lead to large errors in the measurement of the amount of power consumed.

[0053] As noted earlier, the power consumed is a function not only of the voltage and the current, but also of the phase θ between the voltage and the current.

[0054] It is known that due to the inductance and resistance within the current transformer, the current transformer itself introduces a phase error. Thus the measured power may be represented as P.sub.meas where

P.sub.meas=V*(I*K.sub.i)*Cos(θ+α).  eqn 2

[0055] Whereas the actual power is

P.sub.actual=V*(I*K.sub.i)*Cos θ.  eqn 3

[0056] Where K.sub.i represents a scaling factor for the current transformer and α represents a phase error introduced by the current transformer.

[0057] If we focus only on errors introduced by the phase error, then the error can be represented by

Error=(P.sub.actual−P.sub.meas)/P.sub.actual=1−(Cos(θ+α)/Cos θ)  eqn 4

[0058] As a result, when the power factor is high (close to unity) the influence of phase error on the measurement is modest or insignificant. However, as the power factor decreases then the influence of the phase error increases significantly.

[0059] A graph showing the power measurement error as a function of power factor for the one degree phase error and a two degrees phase error is shown in FIG. 4. It can be seen that for power factors of 1 (the voltage and current are in phase) then a 2 degree error in phase measurement is not a problem. However if a load has a power factor of 0.6 (θ=53 degrees) then a 2 degree error in phase measurement manifests as a 5% error in power measurement. It is therefore desirable to accurately characterize the phase error of a current transformer, and indeed also of the signal processing chain associated with the current transformer.

[0060] A problem with current transformers is that there response is potentially quite complex. FIG. 5 is an equivalent circuit diagram of a current transformer. The primary winding resistance is designated Rp, the primary winding inductance is Lp, the secondary winding resistance is Rs, the secondary winding inductance is Ls and the burden resistor is Zb. Zm is the magnetizing impedance of the transformer (for example the transformer may have a magnetic core). We can generally ignore Rp and Lp and express the variables that occur on the primary side of the transformer in terms of equivalent quantities as the secondary winding, then the circuit can be represented as show in FIG. 6. Ī′=Īp/α where α is the transformer ratio; Z′m=Zm/α.sup.2 and Īm′=Īm/α. The angle between Ī′ and Ib is the phase error introduced by the current transformer, We can note that Ī′=Ī′m+Īb; Ē2=Ī′mZ′m; and Ē2=Īb(Rs+jωLs)+Ib(rb+jωxb) where ω is angular frequency in radians per second. This can be used to give insight into the fact that the phase error changes with the size of the burden resistor. However, it also tells us that the relative sizes of the real and imaginary parts of the impedance also change with frequency. Whilst the mains frequency is generally steady at a known frequency, e.g. 50 Hz, 60 Hz, 400 Hz (aircraft), loads such as inverters can be a source of higher order harmonics which should be taken into account if an accurate assessment of power supplied is to be achieved. The magnetizing impedance can change both with frequency and load current.

[0061] It is beneficial to be able to test the response of the current transformer. This could be done by providing a very pure sinusoidal signal as a perturbation to the current through the current transformer and then performing a frequency extraction of that signal (generally by use of Fourier analysis). This requires cost and effort to be expended on the signal source and computational cost to perform an FFT analysis. It would be beneficial to use less expensive signal sources, such as slew rate limited square wave generators. These are simple to produce, for example by use of a digital inverters in a ring, or by toggling a logic gate in response to a counter/timer or a signal from a data processor implementing a numerically controlled square wave oscillator as one of its tasks. The signal need not have a 50-50 mark space ratio and this can further simplify the circuits that generate it. Similarly the slew rate in a voltage or current increasing (pull up) direction does not have to match the slew rate in a voltage or current decreasing (pull down) direction. Other performance limitations will be discussed later.

[0062] FIG. 7 schematically illustrates a current measurement apparatus in accordance with a first embodiment of the present disclosure. A conductor 20 enables current flow to occur between a first node 22 and a second node 24. The first node 22 may be connected to an alternating current supply, and the second node 24 may be connected to a load. However in some circumstances the load 24 may both consume energy and may also supply energy. Thus, node 24 might represent a domestic dwelling which generally consumes energy, but would also have photovoltaic panels such that the dwelling can supply energy back to the electrical supply network represented by node 22 when the photovoltaic panels are operational to produce more energy than the dwelling requires. The current passing through the conductor 20 can be measured by a current transformer 30. The current measurement circuit of FIG. 7 may also be in association with a voltage measurement circuit 32 such that the actual power being delivered from node 22 to node 24 can be measured, for example for billing purposes, by a measuring circuit 50.

[0063] Watt hour meters used for electrically measurement for billing purposes are typically specified to be accurate to within 0.5% or 1%. It can therefore be seen that even a modest phase error, being less than 1 degree, is unacceptable even for power factors of approximately 0.9. Domestic residences may have a power factor different to unity because of the use of fluorescent lamps, washing machines, induction heating ovens and so forth. Industrial premises are more likely to have large inductive loads but similarly are more likely to have installed power factor correction devices in order to mitigate their energy bills.

[0064] Nevertheless, it can be seen that in order to comply with the accuracy standards required of watt hour meters, it is highly desirable to compensate for any phase errors in the current measurement transformer 30. In the arrangement shown in FIG. 7 a current modulation circuit 60 is provided so as to modulate the current drawn from the node 22 in a known manner. The current modulation circuit 60 may be directly connected to the conductor 20 and may periodically switch between drawing a first known current level and a second known current level, one of the levels may be a zero current flow. This switching information is provided to the measurement circuit 50 which is responsive to the output of the current transducer 30 and which can compare the times at which the modulated current 40 changes to observations of that change as made by the current transformer 30 in order to estimate a phase error of the current transformer 30. The timing of the changes in the modulated current may be controlled by a controller 62 which can also supply timing data to the measurement circuit 50. In order to simplify the circuit estimates of phase measurement error may be limited to being performed in one of the voltage half cycles of the mains supply.

[0065] FIG. 8 illustrates an alternative arrangement to that shown in FIG. 7 where the current modulation circuit 60 is not directly connected to the conductor 20, but instead drives the modulated current through a further conductor 64 which passes through the current transformer 30 adjacent to the conductor 20. This arrangement ensures galvanic isolation between the conductor 20 and the modulated current driver circuit 60. Otherwise the operation of the circuit is similar to that of the circuit described with respect to FIG. 7.

[0066] In order to facilitate implementation of a low cost and reliable modulated current drive circuit 60, the modulated current drive circuit 60 provides a square wave current. The square wave current is schematically illustrated in FIG. 9. The square wave current may, for example, be achieved by selectively switching on an off a current source, or by placing a current source in association with a current steering circuit, as will be described later. However, no matter what approach is taken the parasitic capacitive and inductive components associated with the drive circuit 60 and the conductor 20 or conductor 64 will be such that the current does not instantaneously transition between a first value 70 and a second value 72. Instead, when the modulated current drive circuit switches between the first value 70 and the second value 72, then although the switch instruction may result in the current change being initiated at time T1, the current by virtue of being slew rate limited will not reach the second value 72 until time T2. The time difference T2−T1 has a material effect on a subsequent estimation of phase measurement error from the transformer. Similarly, when the square wave transitions from the second value 72 to the first value 70, although the transition may start at time T3, it does not finish until time T4. Additionally if the drive circuit 70 includes an active circuit such as an amplifier, the active circuit/amplifier may have a finite gain bandwidth or slew rate limit that effects the circuit response in a substantive way. The response of the drive circuit 70 may therefore vary as a function of temperature, or may vary over its lifetime or indeed may vary with manufacturing variations. Therefore the actual form of the drive signal may not be accurately known.

[0067] The inventors realized that any phase measurements estimated as a result of applying the nominal square wave drive signal to the current in the conductor 20 or flowing in the measurement conductor 62 needs to account for the time taken to transition the square wave between the first value 70 and the second value 72 and any estimates of phase change need to be made with reference to an appropriate value, such as the mid-point of the transition, i.e. ½(T1+T2) and ½(T3+T4), and not the nominal start times T1 and T3. Furthermore, applying this correction means that the performance required of the square wave generator is not so critical so smaller and less power hungry devices can be used.

[0068] The duration of the transition, for example from T1 to T2, can be estimated by starting a counter at T1, and stopping the counter when it is determined that the second current value 72 has been reached at time T2. The value of the count held in the counter can then be converted into a time offset and supplied as a corrected transition signal to the measurement circuit 50.

[0069] The correction for slew rate limiting may be performed using an estimation circuit 80 as shown in FIG. 10. The estimation circuit 80 may, for example, include a relatively low value resistor 90 inserted in the current flow path to and from the current modulator circuit 60 so as to measure the modulated current. The voltage developed across the resistor 90 may be DC blocked and amplified by an amplifier having a moderate high pass filter response and then filtered by a further high pass filter 94. The output of the filter 94 selects the edges of the square wave. The voltage can then be detected by rising detector and falling edge detectors 96 and 98 respectively to start and stop a counter timer so as to accurately measure the time for the square wave to transition between its first and second current values, and therefore to accurately estimate the mid-point of the square wave transition and to provide this information to the measurement circuit 50 such that it can correctly account for the phase angle between the voltage and current supplied from node 22 to node 24 in order to correctly identify the amount of energy used by equipment connected to node 24.

[0070] The signals in the circuit of FIG. 10 are shown in shown in greater detail in FIGS. 11a to 11c. FIG. 11a shows the voltage developed across the resistor 90 as the current from the current modulator changes from a relatively high value to a low one. The voltage across the resistor can be high pass filtered to remove the DC component amplified by the amplifier 92 to get a pulse like shape as shown in FIG. 11b. This is then passed though the high pass filter 94 to identify the edges 100 and 102 as shown in FIG. 11c which can detected by the edge detectors 96 and 98 to start and stop a clock. Although this functionality was described in the analog domain, the same result can be achieved by digitizing the voltage across the resistor and analyzing the digital sampled to look for the edges of the current transitions.

[0071] The current flow could be bi-polar (i.e. both positive and negative) or it may be unipolar only. Unipolar is easier to achieve as this can be done by a current mirror as shown in FIG. 12. The current mirror 120 is well known to the person skilled in the art and the current flowing in transistor 122 is copied by transistor 124, subject to a scaling factor to modulate the current in the conductor 20 or 64. The current in transistor 122 can be formed by taking the voltage output of a counter and converting it to a current by passing it through resistor 132. The counter is conveniently a divide by 2 counter as this acts to clean an incoming clock into a square wave having an even mark-space ratio. Alternatively the current mirror could be driven by a digital to analog converter under the control of the controller 62.

[0072] Having formed the perturbing current and identified the mid-point of each transition, those mid points can be compared to corresponding changes in the current measured by the current transformer to determine how much phase error the current transformer introduces.

[0073] FIG. 13 illustrates a further embodiment of a power meter 150 associated with a first supply conductor 152 which extends between a first supply node S1 and first load node L1. A second supply conductor 152 extends between a second supply node S2 and a second load node L2. The second conductor can be a live conductor is a single phase supply, but the teachings here are extensible to, for example, 3 phase supplies.

[0074] A current transformer 160 has a coil that couples with the second supply conductor 164 and also with an excitation current generated by a phase error measurement circuit 170. The current at the output of the current transformer is converted to a voltage by a burden resistor 172 and the voltage across the resistor 122 is digitized by an analog to digital converter 174. The output of the analog to digital converter 174 is a steam of samples I.sub.S where S is an index and S varies as a function of time.

[0075] A potential divider formed by a resistors 182 and 184 extends between the conductors 152 and 154 so as to measure the voltage between the conductors. Typically resistor 184 is much smaller than resistor 182. The voltage across the resistor 184 is digitized by an analog to digital converter 184. It will be assumed that the transfer functions on the potential divider is known, but the teachings of WO2014/072733 can be used to determine the transfer function and are incorporated herein by reference. Similarly the transfer characteristic of the current transformer can be assumed to be known, but if it needs to be determined then reader is referred to the teachings of WO2013/038176, the teachings of which are incorporated by reference.

[0076] The outputs of the analog to digital converter 184 is a series of samples Vs. Assuming the current samples and voltage samples relate to substantially the same moment in time (i.e. the temporal separation between then is zero or very small compared to the period of the mains waveform), then the power drawn the load can be represented as:

[00001]$\begin{array}{cc}P=\frac{1}{N}.Math.\underset{S=1}{\overset{N}{.Math.}}.Math..Math.I_S.Math.V_S& \mathrm{eqn}.Math..Math.5\end{array}$

[0077] A processor 190 receives the samples I.sub.s and V.sub.s and can process them to, amongst other things, calculate the power being drawn and keep a sum of the energy consumed. The processor can also examine the series of samples to provide other services, such as looking for disturbances, excess loads, evidence of tampering and so on which might be of interest to an energy supplier. The processor may output the result of its calculations by way of a user interface 192 for example in the form of a display, and/or by way of wireless or wired data connections 194 and 196.

[0078] It can be seen, when considering a sinusoid that a phase measurement error equates to shifting the sinusoid in time. Thus, in the digital domain for a pure sinusoid the sample value I.sub.S is a displaced version of what it should have been, and if the phase measurement error is known the sample value can be moved by an amount of time that corresponds to the phase measurement error and then used in the calculation of power set out in eqn 5. Where the current signal is a superposition of sinusoids at different frequencies then the designer either has the choice to either just use a single time shift to compensate for the most significant component or to examine the phase error as a function of frequency, and then extract the individual contribution of one or more of the significant frequency components, time shift them back to their correct positions and then calculate the power consumption. If phase angle data is required, the phase angle can be determined by a phase detector circuit, or by use of FFT or Geortzel algorithms. In fact, from reference to the generalized situation of FIG. 1, it can be seen that both the current measurement signal and the voltage measurement signal can be subject to phase shifts. The teachings of this disclosure can be used to apply a correction for phase measurement errors and displacements to both the voltage and current measurements such that they can be brought into correct temporal alignment.

[0079] The phase error at a particular frequency may be examined by generating a measurement signal from the phase error measurement circuit 170 at that particular frequency in accordance with the teachings set out hereinbefore. FIG. 14 shows a flow chart for characterizing the phase error at multiple frequencies. The process starts at step 200. Control passes to step 210 where a counter/register is initialized to a value N to set the first frequency F(N) to be investigated. From here control is passed to step 220 where the current modulation circuit is arranged to supply a modulated current at frequency F(N) such that the result of such modulation can be captured in the output sequence from the analog to digital converter 174. Control passes to step 230 where a test is made to see if the phase measurement error needs to be determined at other frequencies. If yes then control is passed to step 240 where the value N is changed to represent another frequency and control returned to step 220 such that the phase measurement error is determined at another frequency. If step 230 determines that another measurement of phase error is not required control is passed to step 250 which waits until another update of phase measurement error is scheduled before passing control back to step 210.

[0080] The estimates of phase measure error may be used to correct phase measurements immediately or stored for later use. FIG. 15 is a flow chart illustrating how the phase measurement errors may be used. Step 280 obtains a phase measurement error at step 280, for example from a value stored in memory as a result of executing the flow chart shown in FIG. 13, and at step 290 used this to calculate a time correction that is used to move the signal I.sub.S from S=TR, where T is time from an arbitrary system time at which S=0 and R is the sample rate to I.sub.S′=I.sub.S+(φR/(360F)) where I.sub.S′ is a corrected sample number of the sample I.sub.S, φ is the phase measurement error is degrees, R is the sample rate and F is the frequency of the signal or signal component to which the correction is being applied. The correction is applied at step 300. From step 300 phase measurement error correction may optionally be performed at other frequencies by setting a new frequency at step 310 and then returning the process sequence to step 280.

[0081] Hitherto it has been assumed that the slew rate limited transitions are linear, but this need not be the case. Various distortions can affect the shape of an ideal square wave, an example of which is shown in FIG. 16a.

[0082] A first form of distortion which has already been considered is slew rate limiting, where the ideal instantaneous rising and falling transitions 320 and 322 of FIG. 16a are delivered as slower transitions. FIG. 16b shows example shapes of slew-rate limited square waves where the rising edge is represented by waveforms 330, 332 and 334 which exhibit fast, medium and slow slew rates, respectively. Similarly the falling edge is represented by relatively fast transition 340, a medium speed transition 342 and a relatively slow transition 344.

[0083] There is no reason to assume that the slew rates for the rising and falling transitions will be the same. Thus the square wave like input waveform may have an asymmetric slew rate limited form as shown in FIG. 16c.

[0084] Slew rate limiting is not the only form of distortion which could affect the input waveform. The on-resistance of transistors may combine with parasitic capacitance to give rise to rising and falling edges that asymptote towards their target value as shown by transition 350 in FIG. 16d, in the style of charging or discharging a capacitor via a resistor. Similarly parasitic inductance may interact with the parasitic capacitance to introduce overshoot, also shown in FIG. 16d.

[0085] The techniques disclosed herein can be used to estimate a corrected effective rising edge and falling edge time for the square wave, with the timings to be set to, for example, a 50% of the voltage transition threshold. However other values can also be selected.

[0086] As noted with respect to FIG. 1, components such as the anti-aliasing filter the ADC and the programmable gain amplifier can introduce delay. This observation can be further generalized as shown in FIG. 17.

[0087] In FIG. 17 a rising edge transition is instructed by a digital instruction at time 400 and a falling edge transition is instructed at time 402. These instructions are provided to the signal generator 404. The signal generator may be a simple logic circuit, such as a D type flip flop with its Qbar output connected to its data input, or it could be more complex, such as a DAC. However, the signal generator can be assumed to introduce some delay and have a limited transition speed such that the effective transitions should be placed at new times 410a and 412a. The output from the signal generator passes through a driver 420 which adds further delay and/or slew rate and bandwidth limitation such that the effective transitions should now be placed at times 410h and 412b. The current sensor 3 adds a further delay such that the effective transitions as measured are now at times 410c and 412c. By the time the signal has passed through the filter 6 the effective transitions have moved to times 410d and 412d. By the time the digitization on the ADC 7 has been completed the effective transitions have become delayed to times 410e and 412e.

[0088] The relative amounts of each additional delay are deliberately not drawn to scale. It should simply be noted then in the current signal path and the voltage signal path each input signal used for measurement purposes can be subject to the sum of the delays, and that correction may need to be applied to the voltage and the current measurement channels.

[0089] As noted before, the signal generator could be a DAC, and hence the input signal can be given any desired shape and since the shape of the input signal is known the same shape can be looked for in the output of the ADC 7 to acquire an estimate of propagation delay. Thus the DAC could be driven to generate classical waveforms such as square waves, step wise approximations of triangle waves, step wise approximates of sinusoids and so on.

[0090] In an alternative approach the DAC could also generate random or pseudo-random test sequences which would look like noise but which could still be recovered from the output of the ADC 7 so as to allow a delay to be estimated. Auto-correlation techniques can be used to achieve this as they are computationally robust and reasonably easy to perform. This would characterize the time delay though the system, which could then be converted to a phase delay for a given frequency.

[0091] The arrangement of FIG. 17 where the signal generator 404 is a DAC also allows for the possibility of using the DAC to generate a known approximation to a slew rate limited square wave. However now the transition rate of the rising and falling edges can be determined by the digital circuit driving the DAC and these rates can be chosen to be comfortably within the bandwidth and slew rate capabilities of the buffer/driver 420. Now the time correction discussed herein before with respect to FIG. 9 can be provided as a preset number based on the effective transition time, e.g. from T1 to T2 as defined by the control word sequence supplied to the DAC. A similar deterministic as opposed to measurement based approach can be used for other signal profiles where the characteristics of the signal output by the DAC in terms of speed of voltage transitions are selected such that they do not approach the limits of fidelity of downstream driver circuitry called upon to introduce the input signal into the measure apparatus.

[0092] However, such an observation can be further extended as shown in FIG. 18.

[0093] Here a signal may be generated from a signal source that we do not control in a deterministic manner. The signal could be from a low quality (including an exceptionally low quality) oscillator and driver, a filtered noise source or a random number generator driving a DAC. However a copy of the reference/input current is digitized by an analog to digital converter 450, which may be a separate device or could be provided by the ADC 7 operating in a time multiplexed manner, and the digitized input signal used to characterize the response of the current transducer and the output from the current transducer can then be compared, and cross-correlated to find the delay. Using the ADC 7 in a time multiplexed manner to acquire a copy of the input signal can be advantageous and the delay introduced by the filter 6 and PGA/ADC 7 can be made common to both signal chains, thereby effectively mitigating the effects of these delays.

[0094] In the embodiments described with respect to FIGS. 7 and 8 a controller 60 provided a timing signal to the power meter 50. However the power meter 50 may also be connected to an optional second current sink to provide a signal Iref as shown in FIG. 12 that tracks accurately with when the transistor 124 switches, thereby providing a direct measurement of the start of a transition.

[0095] The circuit can be used on a single phase as shown in the Figures or on split phase systems such as those in the USA or Japan, or on 3 phase systems as commonly found in large installations. In 3 phase systems 3 current transformers would be used, one for each of the phases and a neutral point may be connected to a return line to account for phase imbalance.

[0096] The circuit can be used in many applications where measurement of AC signals are desired, and can be used in domestic, industrial, aeronautical and medical fields (this is not a limiting list). The apparatus and method described herein can be used “in situ” but can also be used by component manufacturers and installers to perform testing can calibration during manufacture and/or installation of transducers and meters. The meter may include communication capability (as this is becoming commonplace) to enable it to report on power consumption. This capability may be leveraged to report the phase error as well for network monitoring purposes, to identify uncompensated loads or to monitor the performance of the meters to enable faults or degradation in performance to be identified and scheduled for rectification and/or compensation or correction to be applied to a customer's bill to avoid overcharging and hence intervention by a regulatory authority pending repair or replacement of the meter and/or current transducer either alone or in combination.

[0097] The claims presented herein are in single dependency format suitable for filing at the USPTO, but it is to be appreciated that any claim may (and is expected to) depend on any preceding claim of the same type unless that is clearly technically infeasible.