DE2630959C2 - Kilowatt hour meter with static measuring mechanism - Google Patents

Description

The invention is based on a kilowatt hour meter with a static measuring mechanism for three-phase networks according to the preamble of claim 1.

Known currently used kilowatt hour meters with static measuring mechanisms work with analog ones Signal processing methods, preferably using the time division method with subsequent current-frequency conversion. These methods are unsuitable for an economical three-phase meter because they are analog Signal processing precise and therefore expensive components must be used, their temporal inconsistencies the measurement result is additionally falsified.

The energy supply companies expect that a kilowatt hour meter can be left uncontrolled in the network for many years and that such a device will remain functional for twenty to thirty years.
Known methods for energy meters with static measuring units use scanning devices, which ever

Phase need their own analog / digital converter (DE-AS 22 45 426). With such measuring devices, zero point errors and linearity deviations of the analog-digital converter falsify the measurement results with increasing use of the devices.

Another method (DE-OS 22 50 298) uses a thermoelectric conversion of electrical quantities on, from which the consumed power is determined by periodic sampling of voltage differences and a constant current source. A long-term constancy of the measurement accuracy without readjustments is included driving this vet is not guaranteed.

The invention is based on the object of a method for a kilowatt hour meter with a static To create a measuring mechanism that meets the requirements for measuring accuracy and long-term constancy an automatic zero point and calibration correction of the measuring mechanism.

In the case of a generic kilowatt hour meter, this object is achieved by the features of claim 1 solved It is based essentially on the fact that the measured value processing is digital as far as possible, thus error and drift-free, that the influence of the components or circuit parts of the counter, which process the measured values analogously, on the measurement result through an automatic zero point and calibration correction a remaining linearity error is eliminated that a multiplexing method and, as an essential characteristic of the invention, a microcomputer for the control and the measured value processing, one or two Analog / Digkal converters are placed in front of them.

The invention is explained in more detail below with the aid of the drawing, which consists of the figures F i g. 1 a two-channel multiplex system that F i g. 2 a single-channel multiplex system that F i g. 3 a single-channel multiplex system in which all input variables are present as current signals that F i g. 4 shows an overview of the signal sampling possibilities for calculating the instantaneous power, and the F i g. 5 the scanning sequence of a voltage and a current to form the instantaneous power.

Since a computer can only process digital values, the three voltage-Z-current pairs of the three phases of the three-phase network are converted into corresponding digital values. It should be noted that receive the temporal assignment of the voltage and the current of a phase to record the active power must stay. In order to adhere to the specified error limits, a maximum of one error of a few arc minutes occur; at 50 Hz this corresponds to a time error of a few microseconds.

If you use two analog / digital converters and a two-channel multiplexer, you can get one Avoid timing errors at the expense of effort (Fig. 1).

Advantageously, however, an analog-to-digital converter and a single-channel multiplexer are initially used six inputs are used. B. the six measurands in the following order of that Multiplexer applied to the analog / digital converter, converted one after the other into digital values and fed to the computer for further processing:

35 U _R -1R-Us-Is-Ut-It

Between the sampling time z. B. the voltage Ur and the sampling time of the associated current Ir, occurs through the multiplexer! a time offset, the amount of which depends on the multiplex frequency

Da dun. ·! the multiplexing of each measured variable is sampled, its level is essentially determined by the harmonics which are still to be detected. If anger jmmen that higher than the tenth harmonic are no longer important, where would an A btastfrequenz of 1 kHz suffice. If you take 2 kHz, you need at least 16 kHz & 60 µs, including the calibration and calibration cycle. In this case there is a line of 60 μβ between the sampling of a voltage and the associated current; this falsifies the measurement of the active power in an impermissible manner

If the multiplexer, the analog / digital converter and the computer are so fast that the Multiplexfreqaenz can be increased so much that the remaining time error is negligible, then none further action to be taken. To the computer, multiplying two numbers is in a few Microseconds problematic; this requirement can practically only be met by parallel multiplication.

This means that using a faster computer is an unnecessary increase in the cost of the device represents. In the following, therefore, some methods are explained with which the slow scan resulting time errors can be largely compensated so that a computer is used for processing the measured values significantly lower performance can be used

The first method provides analog delay elements that are inserted into the feed lines to the multiplexer in the each later sampled three measured variables are looped in. If tfem is assumed to be a numerical value of 60 μβ in the previous example, then such a delay is difficult to produce and over to keep constant for long periods of time. At higher multiplex frequencies, if the time error is already small is, however, this method appears to be quite useful in order to largely compensate for this remaining error.

According to the invention it is proposed that the computer first always store a few samples before he evaluates them. These stored values can be used in various ways for interpolation or extrapolation calculations.

In the simplest case, by changing the scanning sequence:

Ur-Ih-Uh-Us-Is-Us-Ut-It-Ut

create a very precise intern olation option. In this case the multiplexer does not need more inputs, i. H. including the zero point and the

Calibration correction eight inputs. However, the number of clocks per multiplexer cycle is increased. The computer can now use the two voltage samples to determine the associated current value on both sides over time , calculate the missing voltage value by linear interpolation and then compare this with the Multiply the relevant current value to determine the instantaneous power. Another sampling order is:

Ur-Ur-Ir- Us- Us- Is- Ut- U ₇ - h

From which the computer calculates the voltage value associated with the current value by extrapolation. In order to avoid increasing the number of clocks per multiplexer cycle, one can also be extended in time Realize extensive interpolation. In this case, the same multiplexer cycles no longer follow one another, but the three voltages first alternate:

Ur-Us-Ut

15th

and in the next multiplexer pass the three streams: Ir-Is-It

scanned. The computer now interpolates the voltage value that is temporally associated with a current value from the two voltage values taken from the preceding and the following multiplexer cycle. If the multiplex frequency is high enough, the resulting interpolation error remains sufficient small. In principle, it does not matter whether the voltage or current values are determined by interpolation will. In the case of the method with double sampling of the voltages or the currents, however, it is expedient. to sample the currents more often, since the mains voltage is generally quite good sinusoidal, while the current,

z. B. by thyristor-controlled loads, can be heavily distorted, so that it is due to the more frequent sampling is detected.

With interpolation, it is also possible to calculate the mean value at the level of the instantaneous output.

take, d. H. z. B. in the scanning order

U-I-U

you can first multiply the first voltage value by the current value, then the second and over average the two products.

FIG. 5 shows how - without taking into account the multiplexer of the three phases - a voltage u (t) and an associated current i (t) are sampled. The number of samples is denoted by η. The individual samples are multiplied with one another as follows:

Pm (t) = - (Ual'i + / | U ₂ + U ₂ h + ■■■ i (n-2) U („- t) ■ I _n ) Inserting the function values results in general:

45 50

" ^U ' ^{φ) =} " 7 i? ^l " ^{sin (ö} ° ^{v Tif)]} ' f ^sin (^ ^{0 v Tm + φ +} " V ^- ) J

- [/ sin (ωονΤ ^ + φ- - ^) J >.

£ [ύ sin («*, ν TJ] - [/ sin

The result for the real measurement error of the power is:

free = I COS (ZT * " ZfM) - 1 I

The error depends neither on the modulation nor on the phase shift between voltage and current, but only on the number of samples per period. However, the period and the number of samples per period are constants; the measurement error can therefore be calibrated once and thus ω is omitted. For a measurement error of 1% there are at least 44 samples per period (20 ms); this leads to a sampling frequency of 22 kHz. It does not matter whether the sample values of voltage and current are obtained in one and the same multiplexer cycle or in alternating multiplexer cycles; only the time interval is decisive for the measurement error.

With alternating cycles, the number of clocks in a sampling cycle is reduced to four, with the Order z. B. looks like this:

Ur-Us— Ut- zero ... Ir-h — h— reference voltage ... Ur- l / s — ί / τ — zero ... Ir — Isusv /.

Theoretically it is possible, at the moment of scanning a quantity, e.g. B. a voltage, the associated Value of the other quantity, /. B. a stream, to be buffered in an analog manner, so that a temporally related Couple is available for product formation. In practice, this is difficult because of the kilowatt-hour meters the measuring error remains constant over the entire measuring range relative to the respective measured value, so that very small values still have to be recorded precisely. The interference from the switching impulses would already be at 5 Using an analog memory can significantly falsify small stored values.

After the problems that occur with single-channel multiplexing have been solved according to the invention, what remains is Fej>; r source mainly the analog / digital converter, because the subsequent digital processing by the computer is to be regarded as error-free.

In each individual - or with alternating multiplexer cycles z. B. in every second multiplexer cycle - a zero point is obtained in which the input of the analog / digital converter via the multiplexer Zero potential is switched. Does the analog / digital converter indicate a digital value other than zero the computer, it detects the deviation and saves the value in order to then receive each further from To correct the digital value delivered to the analog / digital converter before the evaluation. Because of the rapid Zeroing clocks also result in relatively fast zero point migrations of the analog / digital converter settled. This means that there is no zero point error over the entire service life of the device.

The input of the A nalog / digital converter connected to a reference voltage via the multiplexer, the setpoint value in the memory of the computer is included. If the annlog / digital converter delivers a setpoint that deviates from this setpoint Digital value, this is how the computer determines the deviation, saves the correction factor and then corrects it each delivered value after having previously taken the zero point error into account. This eliminates the Calibration or slope errors over the entire service life of the device.

The remaining error is the linearity deviation of the analog-digital converter-converter characteristic curve, the however, it can always be kept sufficiently small, since it is practically only dependent on the precision of a resistor divider chain depends.

Because of the requirement for constant measurement errors in the entire measurement range, it is not advisable to work with a linear analog / digital converter-converter characteristic that requires a more complex analog / digital converter whose accuracy would not be exploited for large measured values. According to the invention, a non-linear stepped characteristic is therefore preferred which is such that the relative measurement error is the same at every point on the characteristic. Of course, any analog / digital converter / converter characteristic curve can be taken into account by means of a suitable program on the computer. All circuit parts of the electronic energy meter listed so far - the multiplexer M, the microcomputer MC and the analog / digital converter AD - can be monolithically integrated, including the resistors of the analog / digital converter.

With a three-phase meter, z. B. Voltage dividers and current transformers, such as ferrite pot-core current transformers with or without electronic error compensation, used. The multiplexer and the analog / digital converter process all input signals in the form of voltages or currents, whereby currents can be switched through more easily via the multiplexer; also the comparator of the analog / digital converter preferably compares currents rather than voltages.

In any case, the six input variables must be divided down so that they can be processed by the electronics can be. In the manufacture of such a counter are therefore z. B. the voltage divider ratios and to adjust the load resistances of the current transformers to their setpoints. Such adjustments require expensive adjustable components and are particularly labor-intensive.

According to the invention, this disadvantage is eliminated by the fact that, in principle, all adjustment work is dispensed with, by being replaced by appropriate programming in a memory. For this purpose, the actual actual values of each individual device in a preferably fully automatic test process Voltage divider ratios, current transformer transformation ratios (including the load resistances) and the reference voltage measured and corresponding correction values in a computer accessible Inscribed memory. A programmable read-only memory (PROM) is expediently used for this det. which is contained in the computer circuit and is automatically programmed by the test device. If a repeatedly programmable read-only memory (RePROM) is used, later Recalibrate the meter again with the same test equipment by programming in new values.

Electronically error-compensated current transformers are known and allow with small ferrite pot cores very precise current ratios. However, due to the required auxiliary amplifier, they have at the output in usually an offset DC voltage. When calculating the eye performance, there is no such thing as long Errors such as such a DC voltage component is only present at one input of the multiplier; this is to be assumed when the meter - as it was previously only customary - from voltage and current transformers is fed. When used with a direct connection, however, longer leads and thyristor-controlled Consumers definitely have a DC component of the voltage, which together with the DC offset an electronically error-compensated current transformer would result in a measurement error. It is therefore useful if the electronic error compensation is not used. Current transformers with ferrite cores show relatively large errors. When using the method described here, the automatic test device is able to display the two error curves for each of the three current transformers in a meter using the measure the entire dynamic range and store it in the programmable read-only memory (PROM). The computer is now every single measured current value delivered by the analog-digital converter with the evaluate the associated error curves. The only prerequisite for the practical use of this procedure is

that the error curves of such a current transformer over a longer period of time and also after possible strong ones Overloads remain within given tolerances.

In a further embodiment of the invention, temperature-dependent measurement errors are largely compensated for in the same way that the known and generally less sample-dependent temperature coefficients of the components not detected by the automatic zero point and calibration correction are also stored in the programmable read-only memory (PROM), with a suitable one A temperature sensor is to be provided, the output signal of which is to be fed to the computer at sufficient time intervals via the multiplexer and the analog / digital converter. With particularly high demands on the Meßgenauiglteit it is of course also possible, for. B. the temperature curve that determines the calibration

to measure the reference voltage source automatically for each individual copy and in the programmable To save read-only memory (PROM).

With the long service life of the meters, the aging-related drifts are not from the zero point and Calibration correction detected components of importance. These are mainly the input voltage dividers, the current transformers and the reference voltage source. With a well-mastered production are

the aging rates are known and could also be stored in the non-changeable read-only memory (ROM) or programmable read-only memory (PROM) of the computer. As energy meters in most If you work with a mains supply for the electronics, a time base can easily be derived from the mains frequency derive, which is set to zero in the aforementioned fully automatic test process. When calculating the energy, the computer takes into account the respective status of the time base and the aging rates to

compensation for age-related influences.

The analog / digital converter must be used with an energy meter - especially with a three-phase meter - process a very large dynamic range. Because his input from the multiplexer is constantly on others Input variables is switched, its input signal from one sample to the next, z. B. from that max. negative to max. positive value, jump. These large signal jumps make high demands

to the comparator of the analog / digital converter. The comparator needs one after each overload certain recovery time. The exact detection of one from a maximum value to one close to zero is critical jumping signal value. The known tracking analog / digital converter avoids this problem by the comparator input signal only corresponds to the difference between two samples.

In a further embodiment of the invention, the principle of the concurrent analog / digital wall

lers applicable. With each sampling of an input variable, as described, the sampling value im Computer stored for the purpose of interpolation. The analog / digital converter resumes after each sample converts the analog value into a digital value that the computer takes over. The analog / digital converter is then free again and can now be preset by the computer to the last sampled value of the input variable to be sampled before the start of the next sampling. The comparator only receives the

Difference between two consecutive samples supplied. Presetting of the analog / digital converter and the activation of the input variable to be measured are to be coordinated with one another in terms of time so that one Comparator overload is kept as small as possible.

As already mentioned, it is advantageous to supply the input variables as currents and to provide a Sirom comparator.

Compared to the usual Ferraris meter, the measurement method described here is susceptible to interference pulses due to its speed. About the behavior of the electronic energy meter that of the Ferraris meter approximate, the computer is programmed according to the invention so that the jumps between two scans are limited in height and that larger jumps are evaluated as interference pulses. Are there either to suppress such sampling values recognized as disturbed entirely, or the programmed

♦ The part exceeding the 5th maximum value is subtracted. Furthermore, a sliding mean value calculation is possible. Here the averaging follows before or after calculating the instantaneous power.

By appropriate programming of the computer it can be achieved that the often undesirable accuracy of the measurement with regard to the detection of harmonics of the network frequency is reduced and that of one Ferraris counter is approximated. For this purpose z. B. a Fourier transformation (FFT) carried out by the computer

leads.

After calculating the instantaneous power of each phase by multiplication, the result is a Digital word available, whereby with a three-phase meter the values of the three phases have to be added. An integration over time is also required to determine the energy. The energy consumed is recorded in a cumulative counter and on a display device

The well-known counting and storing roller counters with stepper motor drive are shown for this purpose usable. But it is also possible to store electronic display devices such. B. electrochromic or electrolytic displays. It is also a normal non-memorizing electronic display possible if a non-volatile memory, e.g. B. MNOS memory, is available because of the limited The number of write cycles that such a memory can carry out before it becomes unusable is only given by

write the numerical value to be saved there in the event of a recognized imminent power failure.

In any type of digital display, a jump in the smallest digit corresponds to a specific one Amount of energy consumed. These counting pulses can be obtained, for example, by the computer following Determining the total instantaneous power counts as many periods of a reference frequency as this one Corresponds to instantaneous power; these counted impulses are now directly or according to the corresponding frequency

to divide the advertisement. It is also possible to use the current value of the instantaneous power as a digital value to switch to a programmable frequency divider that constantly divides pulses from a reference frequency source by the programmed factor. The mains frequency is preferably used as the reference frequency; it is also possible to control the clock frequency of the computer via a phase locked loop (PLL)

derive from the mains frequency or use a crystal. The computer is through an speaking Make the program count down certain times; also the program contained in many computers A simple timer can be used to advantage.

For test purposes, a special test output of the computer is advantageous, at which the consumed energy • 'is available with a shorter measuring time than at the output intended for the display device.

Economical computers work with relatively low clock frequencies and therefore need, e.g. for

Multiplications, long times. To save time when multiplying, the principle of concurrent use is again used

ί Make use of the analog / digital converter, not each time using the full values of voltage and

'.; Current are multiplied with each other, but only the difference values to the previous sample values

■.:, According to the following scheme:

; U _n ■ i _n = U _n -I ■ Ai + i "- \ ■ Au

If the multiplication time for purely serial processing is too long for such an application, then always a to use your own parallel multiplier without changing the measuring principle described.

ν '' \ The immunity to interference is a particular problem with electronic energy meters. Usual Ferraris counters,

■ A can withstand extremely high peak chipping at the entrances, and in many cases they can withstand peaks of

■ '] rr.-more than 10 kV. The current input suggested as advantageous - also on the voltage side - has proven itself

In this respect, too, it is cheaper, since practically no tensions occur in the device, so that very much

| jl! me protective measures are possible.

ι »; In the case of an energy meter with a computer, special precautions must be taken against the intrusion of

yes disturbances in the electronics are hit, since a glitch in a computer doesn't just lead to wrong

ι / Ι data, but also can lead to an incorrect command. In this respect, too, the in the

';'! Patent application P 25 37 549.9 proposed single-chip microcomputer as superior because of the effects

. ; j interference becomes minimal when the entire computer is turned off. a single semiconductor crystal is integrated.

4 Malfunctions then only affect the inputs and outputs of the computer, with malfunctions on the

; Inputs only temporarily generate incorrect input data, but do not cause incorrect program execution.

■ .: chen.

; * The invention will be explained below with the aid of some figures:

"■; ■ In Fig. 1, the less advantageous because more complex solution with a two-channel multiple

<·, Xer M and two analog / digital converters ADi, AD _{2 are} shown. Each multiplexer M has five inputs as each

<;': of the two multiplexer channels and each of the two analog / digital walls! er AD \, AD ₂ with regard to their

'.'; Zero point and calibration errors must be corrected. The microcomputer MC controls both the multiple

1. xer M like the analog / digital converter AD \, ADi. For the purpose of calculating the energy he receives the

Reference _frequency f rc r supplied. The accumulated energy consumption is finally displayed on a display device. In this as in the following figures it is assumed that the microcomputer AiC also the _; : contains required memory This also applies to the electrically programmable one mentioned several times

- Memory for recording the correction curves or values. It is Z. B. a metal-nitride-oxide semiconductor memory

.: (M NOS) usable. It is of course also possible to integrate the microcomputer MC from several

■ assemble circuits; this does not change anything in the concept of the invention.

'; j Furthermore, according to the state of the art, it is possible to combine the analog / digital converter AD _U ADi

; men with the associated resistors as well as the multiplexer (s) M together with the computer

i ter on a single circuit. The entire electronic three-phase meter consists of at nn im

, j essentially consisting of three current transformers, a resistor arrangement (e.g. thin-film or thick-film technology),

w the large circuit, the display and the power supply. As described, it does not have any adjustable

; or adjustable components.

ψ In Fig. 2 is the advantageous solution with a single-channel multiplexer M and only one analog / digital

i ^; Converter AD drawn, with all input signals of the multiplexer Min in the form of voltages. the

The order in which the connections are drawn is not necessarily identical to the order in which the input signals are sampled by the multiplexer M. The computer sends the address of the desired input to the multiplexer M via a control line; the order depends on the selected program.

In Fig. 3 shows the circuit for the case that all input signals of the multiplexer M are in the form of current signals.

In Fig. 4 different ways of sampling the signals and the instantaneous power are recorded to calculate in each case; the F i g. 4a, b and c relate to the measurement of only one phase, while the F i g. 4d, e and f apply to the measurement in three phases.

The invention can be used for any number of phases, e.g. B. also for a so-called three-wire meter, in which only two measuring channels are required.

F i g. 4a shows the sampling and interpolation during voltage and current measurement in a single-phase network; the indices denote the number of the sample concerned; the voltage and current input are sampled alternately. A first partial product (instantaneous power) p \ is obtained e.g. B, by interpolating a voltage value from the two voltage samples i / o and U ₂ , which is then multiplied by the current sample value i \. Another partial product p $ arises z. B. in that a value is interpolated from the two voltage values u ₂ and U ₄ , which is then multiplied by the current value / 3. The total power P results for a certain number of samples π as:

₅ -γ multiplications are carried out here

Voltages and currents are completely equal and can therefore be interchanged.

In FIG. 4b, a different product image is described for the same case of alternating sampling of input voltage and current: partial products p _a are now formed from the existing sampling values, the voltage and the current being interpolated alternately. The service Per calculates to:

In Fig. 4c shows an example for the case in which the averaging takes place only after the partial products p " 15 have been formed. The total power P results as in FaH 4b. however, only one memory is required here for only one sample value, since no interpolation takes place or no mean value is formed before the multiplication.

Good for all three methods:

²⁰ P = - (oei ·. + ItUi + ihh + UzU + - ■■)

For the practically most important part of the measurement in a three-phase network, FIG. 4d shows the scanning of phase R, in FIG. 4e the scanning of phase 5 and in FIG. 4f the scanning of phase T is shown.

Ur-Is-Ut-Ir-Us-It

provided the arrows marked with the partial products p ″ indicate how these partial products were created 30 are.

A variety of scan sequences are possible; some typical for the process according to the invention are e.g. B .: Ur- Ir- Ur- Us- Is- U ₅ - Ut- It- Ut

35 Ur- Ir- Ur- I _s - U ₅ - Is- U ₇ - It- Ut- Ir- Ur- Ir- Us- Is- Us- It- Ut- h

Ur-Us-Ut-Ir-Is-It Ur-Is-Ut-Ir-Us-It

As mentioned, the computer determines which input signal the multiplexer M has to switch through, so that, depending on the program, any sampling sequences and any type of evaluation of the samples are possible in the same circuit

In the case of the energy meter with computer described here, the sampling according to FIG. 4a is particularly useful because a computer can add relatively quickly, but a comparatively great amount for a multiplication

needed a long time; according to FIG. 4a, however, only y multiplications are carried out

In Fig. 4a-f, for the sake of convenience, the sampling of the zero or reference potential has been omitted. Depending on the expected drift of the zero point or the calibration, it is only required that zero or so calibration potential is scanned at sufficient time intervals and corresponding correction values are formed and saved. It is therefore by no means necessary to sample the input variables in every cycle also to include the two other quantities mentioned.

In addition 5 sheets of drawings

55

Claims (1)

Patent claims: 1. Method for a kilowatt hour meter with a static measuring mechanism for three-phase networks, which consists of a multiplexer, an analog-digital converter and a computer, the multiplexer being used for cyclical scanning of the voltage and current inputs and with devices for adapting the input variables to those of the electronics processable level, a power supply iür the electronic components and a display device is provided, characterized in that,
that the instantaneous powers of the individual phases are determined and linked to a reference frequency (fnt) , to that the calculated instantaneous power is used to set a programmable frequency divider or frequency multiplier, which divides or multiplies the pulses of the reference frequency (f ^ f) and the counting pulses obtained in this way are fed to a storage display or a non-volatile memory with display device as a measure of the energy consumed will,
that for the determination and storage of a zero point error value by the computer (MC) from the ts multiplexer (M) the Nu ^ potential as well as the reference potential (U _n J) is switched to the analog-to-digital converter (AD) to determine and store a calibration error at sufficient intervals and
that the computer (MC) links all input variables fed to it in digital form via the multiplexer (M) and the analog-digital converter (AD) with the stored correction factor corresponding to the calibration error before further processing Z. The method according to claim i, characterized in that the clock frequency of the computer (MC) is used as the reference frequency (f _re r) for displaying the energy consumed and the time required for the integration is generated by a timer integrated on the computer circuit. 3. The method according to claim 1, characterized in that to compensate for the time delay single-channel sampling and analog-digital conversion of the earlier sampled value Voltage-current pair is delayed accordingly with the help of an analog delay element 4. The method according to claim 1, characterized in that a non-linearly stepped converter characteristic curve for the analog-digital converter (AD) is used, so that the relative measurement error related to the respective measured value remains constant regardless of the modulation 5. The method according to claim 4, characterized in that the analog / digital converter (AD) has a logarithmically graduated converter characteristic 6. The method according to claim 4, characterized in that the analog / digital converter (AD) has a binary stepped converter characteristic 7. The method according to claim 1, characterized in that the computer (MC) controlled analog-to-digital converter (ACf) is designed in addition to multiplexing as a concurrent analog-to-digital converter and the analog-to-digit converter before the start of sampling of a certain size is set to the previous sample of the same size, so that the comparator of the analog-digital converter only processes the difference between the new and the old sample 8. The method according to claim 1, characterized in that a for calculating the instantaneous power Parallel multiplier is used to increase the multiplication time in serial processing of measured values shorten. 9. The method according to claim 1, characterized in that when determining the deviations of the tolerances of the components in the measurement result from the nominal values of the multiplexer (M) and the analog-to-digital converter (AD) are also used by appropriate to the kilowatt-hour meter Input signals are applied and the actual values are tapped in digital form at the output of the analog-digital converter or at a suitable output of the computer (MC) , so that all curves or correction curves measured in this way reflect the individual deviations of both the multiplexer (M) and the analog -Digital converter (AD) of the target converter characteristic between 0% and 100% of the measuring range, whereby with a corresponding number of measuring points per correction curve, every deviation of the converter characteristic from its target course (linear, logarithmic, binary, etc.) except for an arbitrarily small residual error so compensated. 10. The method according to claim 1 or 9, characterized in that for each individual current transformer a correction curve for the amount and a correction curve for the missing angle are determined, stored and is taken into account in the calculation.