FR2545222A1 - Method and apparatus for measuring a voltage or current by analogue-to-digital conversion - Google Patents

Abstract

Direct digital data are created by conversion of the quantity v to be measured into digital form 11 and inverse digital data are created by a similar conversion of the inverse -v of the quantity to be measured. A quantity representing the result of the measurement is calculated from the difference between a direct digital item of data and an inverse digital item of data, and a quantity representative of the error in this result is calculated from sums each consisting of a direct and an inverse digital item of data.

Description

Method and apparatus for measuring voltage or current by

analog-to-digital conversion.

  The subject of the present invention is a pro-

  yielded to measure a voltage or an intensity by ana- conversion

  digital logic. The scope of the invention

  is more particularly that of measuring voltages or inten-

  sites of small amplitudes, whatever the magnitudes re-

  presented by the voltages or intensities measured.

  Currently, the measurement of voltages or intensities of low and very low amplitudes, that is to say by

  example of voltages lower than m V, requires amplifications

  Very high gain, very low noise and no drift gain

  Play a chain of operational amplifiers in cascade

  at least the first of which are very low noise allows satis-

  do with the first two conditions Elimination of the thermal drift can be carried out by thermostating the measurement amplifiers or by measuring the temperature with precision in order to inject into the amplifier chain a compensation signal which is a function of the measured temperature. It is also known to carry out resets by periodically memorizing the zero offset after short circuit of the input to then inject a compensation signal which is a function of the stored zero offset, Generally, known systems for measuring very low voltages are delicate to carry out and therefore expensive and there is no indication of the error

  which taint the results they display.

  The object of the invention is to provide a

  measurement method for measuring voltages or currents

  very low amplitudes with inexpensive equipment.

  The object of the invention is also to provide a method making it possible not only to display the measured value, but also to know

  and possibly display the error with which it is vitiated.

  This object is achieved by a method which, according to the invention, comprises the steps which consist in generating "direct" digital data by converting the quantity to be measured (v) into digital form.

  "reverse" digital data by conversion to digital form

  rique of the inverse (-v) of the quantity to be measured, calculate the difference between a direct digital data and a reverse digital data in order to develop a quantity representing the result of the measurement, and calculate sums each comprising the addition of a direct digital data and a second reverse digital data in order to develop a quantity representing

the error on the result.

  As explained in more detail in

  thereafter, each digital datum includes a component representing

  both the value of the quantity to be measured (or its inverse), a compo-

  health representing the systematic error introduced by the measurement circuits (drift) and a component representing the random error (noise, external influence,) The difference between a direct digital data and a reverse digital data makes it possible to eliminate the component representing l 'systematic error @ is the implementation of the well known principle of "double weighing" The addition of direct digital data with reverse digital data eliminates the component representing the value of the quantity to be measured If it is also supposed that the systematic error does not vary appreciably during the calculation time of a series of different sums, it is possible from this series, to

  characterize the random error tainting the result to be displayed.

  When the quantity to be measured is continuous or varies slowly over time, direct and inverse digital data can be generated alternately, which allows the use of a single converter The result is determined by calculating the half difference between a digital data direct and

  inverse digital data generated successively Parallel-

  ment, consecutive sums each formed by adding a direct digital data and a reverse digital data generated successively are memorized and the quantity representing the error

  on the result is developed from a set of memo values

  laughs, for example by calculating the standard deviation.

  When the quantity to be measured varies significantly during the time interval separating two successive conversions, it is then necessary to use two converters, and the method comprises the steps which consist in: simultaneously generating at a first instant a first and a second direct digital data (ND 1, ND 2) by separate conversions of the quantity to be measured, then simultaneously generate at a second instant a third direct digital data (ND 3) and an inverse digital data (N II) by conversions separate from the quantity to be measured (v) and the inverse (-9) of this quantity, calculate the difference between the third direct digital data (ND 3) and the inverse digital data (N Il) and add the difference between the second and the first direct digital data (ND 2, ND 1) to develop the quantity representing the result of the measurement at the second instant, and calculate algebraic sums each comprising, on the one hand, the additi one of the third direct digital data (ND 3) and of the reverse digital data (N II) and, on the other

  share, the difference between the first and second data nume-

  direct risks (ND 1, ND 2), to work out the quantity representing the error on the resultato In one case as in the other, the inversion of

  the quantity to be measured can be achieved by means of an inverter received

  Before this quantity However, for the measurement of a quantity supplied by a sensor supplied by a direct voltage, such as a bridge or a potentiometer, it is advantageous to invert the quantity to be measured by reversing the supply voltage because the systematic errors likely to occur at the sensor are then taken into account

account.

  The object of the present invention is also to provide an apparatus making it possible to implement the method defined above. This goal is achieved by means of a device

  comprising at least one measuring rod with analog converter

  digital and in which there are further provided: an inversion device for inverting the quantity to be measured applied to the input of the measurement chain, a processing device receiving the digital data delivered by the or each converter in response to the conversion of the quantity to be measured and the inverse thereof, the processing device comprising means for calculating in particular differences each between a direct digital datum and a reverse digital datum, in order to develop the values of the results of measurement, means for calculating sums each formed in particular by the addition of a direct digital data and a reverse digital data, means for memorizing a set of calculated sums and means for calculating from the stored sums of quantities representative of the errors affecting the results produced, and a control circuit delivering signals

  of the reversing device and of or each convertor

  sister. Other particularities and advantages of the method and of the apparatus according to the invention will become apparent from the

  description given below, for information but not limitation,

  with reference to the appended drawings in which: FIG. 1 is a diagram of a first embodiment of an apparatus in accordance with the invention for the measurement of a quantity continuous or varying slowly over time; Figure 2 is a diagram of a variant of the apparatus of Figure 1 for measuring a voltage supplied by a bridge sensor; Figure 3 is a diagram of another variant of the apparatus of Figure 1 with drift compensation; and FIG. 4 is a diagram of a second embodiment of an apparatus according to the invention for measuring

  quantities varying relatively rapidly over time.

  The apparatus shown in Figure 1 includes

  a measurement chain conventionally formed by an analog converter

  digital-digital II whose input receives the signal to be converted via an optional amplification circuit 12 formed by an amplifier or several amplifiers in cascade In the example illustrated, the quantity to be converted is a voltage v, but the invention also applies to the case where this quantity

is an intensity.

  According to the invention, an inverter 13 is connected between input terminals 14, 15 between which is

  applied the voltage v to be measured, and the input of the amplifier circuit

  cation 12 Voltage v is a direct voltage or a voltage whose amplitude varies over time at a negligible speed compared to the sampling frequency to which is controlled

  the converter 11 The digital data provided by the converter

  weaver are transmitted to a processing device 16 which develops a quantity representing the result of the measurement and a quantity representing the error which is vitiated by the result The results produced are transmitted to a digital display device 170 The displayed digital values can be limited to their significant parts determined as a function of the calculated error The latter could also be displayed on a part 17 a of the display 17 A control circuit 18 supplies signals to synchronize the operation of the inverter 13, of the converter Il

and the processing device 16.

  The control circuit 18 comprises a time base 19 which supplies a clock signal H of frequency F A flip-flop 20 forming a frequency divider by 2 receives the signal H and delivers pulses HI whose rising and falling edges control the tilting of the inverter 13 between its two positions in this way, the amplification circuit 12 receives alternately at its input and for the same time intervals 1 / Fla voltage v and its inverse -v A signal H 2 suitably delayed with respect to the signal In order to avoid uncertain operation, controls the sampling of the analog voltage at the input of the

converter 11.

  The processing circuit 16 receives alternati-

  directly digital data ND when voltage v is applied to circuit 12 and inverse digital data NI when voltage -v is applied to circuit 12 Since the voltage to be measured is a DC voltage or one that varies slowly over time

  with regard to the frequency F, two data ND and NI elaborated successively

  The converter 11 can be expressed as follows: (1) ND = V + sl + al and (2) NI = -V + 452 + ca 2 V being the true numerical value of the voltage v, ksl and 452 being the systematic errors tainting the measurements at the instants tl and t 2 of formation of the data ND and NI and, al and a 2 being the errors

  random tainting the measurements at times tl and t 2.

  The errors 4 sl and 452 come in particular from thermal drifts of the amplification circuit. Insofar as the sampling frequency F has a value sufficiently

  high, the variation of the sys-

  tematic between instants tl and t 2 o From relations (1) and (2), it then comes: (3) ND NI = 2 V + Cal a 2 and (4) ND + NI = 2 sl + Cal + a 2 The ND NI difference is independent of the systematic error; half of this difference is used to provide the numerical result Vr of the measurement: (5) Vr = = +

2 2

  If we consider a series of N successive values of ND + NI (for example eight or sixteen values), we can admit that the component 2 sl of systematic error practically does not vary By supposing besides that the population of the values of random errors is of Gaussian distribution, the variance (ND + NI), or standard deviation squared, of the sum ND + NI can be written:

2 2

  (6) a 2 (ND + NI) = 2 (gal + a 2) since Cr 2 (= O Or, we also have (7) Cr 2 (al + ta 2) = 2 Mal% a 2) o we can easily characterize the random error tainting Vr,

  taking half the standard deviation of the ND + NI sums.

  Other statistical calculation methods can be used, for example by determining the extent of the sums

ND + NI.

  The processing device 16 is, for example, of the microprocessor type. Each rising edge of the signal Hi controls, on the one hand, the tilting of the inverter 13 to apply the voltage v to the circuit 12 and, on the other hand, triggers a monostable

  21 which produces an Hll signal controlling the acquisition by the

  positive processing of the corresponding value ND available at the converter output as well as the execution of a storage and calculation cycle described below Similarly, each falling edge of the signal H 1 controls, on the one hand, the tilting of the inverter 13 to apply the voltage -v to the circuit 12 and, on the other hand, triggers a monostable 22 which produces a signal H 12 controlling the acquisition of the corresponding value NI as well as

  the execution of a memorization and calculation cycle.

  The cycle triggered by the Hll signal is as follows

  acquisition of the new ND value available

  ponible; storage of ND in an RND register; calculation of Vr = 2 (ND lRNIl), lRNIl being

  the value of NI previously acquired and recorded in a re-

  RNI registrar; calculation of S (ND + RNI) storage of S in a register R 51; calculating the variance Cr of the last N calculated values 51 a Sn of the sum S (n = 8 or 16 for example), these values being available in registers R 51 to R Sn; calculation of the error tainting the value Vr, for example the confidence interval er = + 25; calculation and display of the significant part S (Vr) of Vr with possible display of er; and

  transfer of contents of registers R 51 -

  R Snl respectively in the registers R 52 a R Sn.

  A similar cycle is triggered by the

signal H 12.

  According to a variant of the process described above, a test can be carried out on the calculated value of ta to inhibit the display and trigger a light and / or sound indicator incorporated in the display device 17 when er exceeds a predetermined limit. the indicator can reflect the existence of a significant noise, or a malfunction of the converter or even a too rapid variation of the amplitude of the measured voltage. Regarding this last point, it is recalled that the possible variations of the measured must remain at a frequency much lower than the sampling frequency F, for example less than 10 4 F and preferably at 10 5 F. Although a mode has been described above

  particular embodiment of the processing circuit 16 with a micro-

  processor, this processing circuit can be of another type, for example

example in wired logic.

  The analog-digital converter 11 can also be of any type, for example ramp converter or

voltage-frequency converter.

  As for the inverter, depending on the desired switching speeds and the admissible noise level, it can be constituted by a relay, a transistor switch, or even

a mechanical inverter.

  In the device illustrated in the figure, the assembly used makes it possible to eliminate all the systematic errors occurring downstream of the inverter. In certain cases, it is possible to eliminate systematic errors occurring at the level of a sensor supplying the voltage v This is so when this voltage v is produced by a sensor using a bridge mounting or a potentiometer and the inversion relates not to the output of the sensor, but to the DC supply voltage of the latter For example, in the case of FIG. 2, the voltage v is supplied by a sensor formed by a gauge bridge 30 whose output terminals 34, 35 are connected to an analog-digital converter 31 via a amplifier 32 with high gain A processing circuit 36 analogous to the circuit 16 described above receives the digital data supplied by the converter 31 to calculate the measurement result and control the display device 37 An inverter 33 es t connected between the power supply terminals of the bridge 30 and a circuit 39 delivering to the latter a continuous supply voltage The operations of the inverter 33, the converter 31 and the processing circuit 36 are synchronized by a circuit

38.

  Thanks to the arrangement of FIG. 2, it is also possible to eliminate systematic errors originating from the sensor, for example errors due to voltages

  parasites produced by thermocouple effect at welds.

  One of the main advantages of the method and of the device in accordance with the invention consists in the possibility of measuring very low voltages without imposing very strict tolerances on the drift of the circuits used. However, in the case where the voltage to be measured v must be subjected to a amplification by means of a very large grain chain formed, for example, of at least two amplifiers 42 a, 42 b (FIG. 3), it is possible, even with a reasonably low temperature drift, that the full scale at the input of the amplification chain is traversed

  simply for a variation of a few tens of degrees UC.

  For example, if we consider an amplification chain with a gain of 105, a full scale at the input of + 102 p V with a resolution of 0.01 u V and a saturation voltage at the output of

    10 V, the full scale is covered for a temperature excursion

  erasure of 20 'C if the drift is 10 p V / C.

  In such a case, it is necessary to correct the drift, For this purpose, we take the systematic error value, s obtained after a double measurement by performing the operation Ès = 2 The numerical value% 6 calculated is processed and then converted in analog form by means of a converter 49 to be introduced as a correction voltage in the amplification chain, for example between amplifiers 42 a and 42 b The calculation of s is carried out by the processing device 46 which receives the data

  ND and NI digital signals from analog-to-digital converter 41.

  Note that it is not a question here of making a fine correction of the drift, since the systematic error is in any case eliminated by the double measurement.

  coarse correction requiring only a 49 converter

  limited lution Drift correction therefore does not require the use

expensive circuits.

  It will also be noted that, during a series of measurements with calculation of the random error, the value fis reinjected into the amplification chain must be kept constant in order not to distort this calculation, or must be permanently known for

  be taken into account in the calculation.

  We have considered above the case where the quantity to be measured is continuous or varies at a very low frequency compared to the sampling frequency. The method according to the invention can also be applied to the measurement of a quantity which varies significantly between two sampling instants, using two converters Such an arrangement is

illustrated by figure 4.

  The quantity to be measured, for example a voltage

  sion v is applied in parallel to the inputs of two measuring chains The first measuring chain includes, in series, a

  inverter 53, an amplification circuit 52 and an analog converter

  logic-numerique 51 The second includes an amplifier circuit

  tion 54 and an analog-digital converter 55 The digital data supplied by the converters is transmitted to a processing device 56 which develops a quantity representing the result Vr of the measurement and a quantity representing the error e with which this result is vitiated A circuit 58 provides the control signals for the inverter 53, the converters 51 and 55 and the processing device 56 A control circuit similar to that described with reference to FIG. 1 may be used, the

  converters 51 and 55 being controlled synchronously.

  The operation of the assembly of Figure 4 is as follows

  At the rate of the reverse command signal

  sor 13, the processing circuit alternately receives a torque

  formed by direct digital data N Dl supplied by the conver-

  weaver 51 (voltage v not inverted at the input of circuit 52) and direct digital data ND 2 supplied by the converter 55, then a couple formed by inverse digital data N Il supplied by the converter 51 (voltage v inverted at l circuit 52)

  S and direct digital data ND 3 supplied by the converter 55.

  The data ND 1 and ND 2 provided at time tl can be written: (8) ND = Vl + sl +% al and (9) ND 2 = VI + S 2 +% a 2 VI being the true numerical value of the voltage v at time tl,

  sl et al being the systematic and random errors of the first

  first measurement chain at this time, and sl and% a 2 being the systematic and random errors of the second measurement chain

also at the moment tl.

  The data N Il and ND 3 provided at the following time t 2 can be written (10) NI = -V 2 + 'sl +' al and (11) ND 3 = V 2 + 's 2 + a 2 V 2 being the true numerical value of the voltage v at time t 2,

  while% 'sl,%' s 2 and k'al,% 'a 2 are systematic errors.

  and random of the two chains at the same time.

  Systematic errors are essential

  drifts of thermal origin in the amplifier circuits

  cation of measurement chains At sampling frequencies which

  can be commonly used, the variation of the system error

  material during a sampling period (interval between tl and

  t 2) is negligible We therefore have% sl = 'sl and% s 2 =%' s 2.

  From equalities (8) to (11), we can then draw: (12) 2 Vr = -NDI + ND 2 + ND 3-N Il = 2 V 2-% al + a 2 +% 'al-'a 2 (13) Si = N Dl ND 2 + ND 3 + N Il = 2% 'sl + ial-% a 2 +% al + a 2 (14) 52 = -N Dl + ND 2 * ID 3 + N Il' 2% 's 2-% al +% a 2 +%' al + 'a 2 In relation (12), the systematic error is eliminated The processing device 56 calculates half of the quantity -N Dl + ND 2 + ND 3 NI 2 and provides the calculated value as a result of the measurement of the voltage v at the instant t 2 o A new pair of direct digital data is recorded at the sampling instant t 3 which follows the instant t 2 and a new measurement result is calculated after the next sampling at time t 4, this result representing the voltage v at this last time.

  The error in a calculated result Vr is deter-

  mined from a series of sr or S'r sums obtained by relations (13) or (14) The standard deviation or the extent can be calculated on a population of values Sr or S'r taken from a time interval during which the systematic error practically does not vary It is therefore possible to use this standard deviation or this extent to characterize the associated random error

to the result Vr.

  The relations (13) and (14) also make it possible to calculate the systematic errors% 'sl and%' s 2 of each measurement chain As in the case of the assembly of figure 3, these errors calculated after treatment and conversion by means of digital-analog converters 59 and 59 'allow signals to be developed

  which, fed back into the amplification circuits 52 and 54, self-

  risk a correction of a large part of the drift, the part

  remaining being eliminated by the measuring device.

  Furthermore, if necessary, the result Vlr of the

  measurement of the voltage v at the instant t I can be calculated, after the ins

  as t 2, by combining relations (8) to (11) to form (15) 2 Vlr = ND 1 + ND 2-ND 3-NI 1 = 2 V 1 + ga 1 + la 2-R'a 2- has 1

Claims (10)

CLAIMS R E V E N D I C A T I O N S   1 Method for measuring a voltage or an intensity by analog-digital conversion, characterized in that it comprises the steps which consist in: generating direct digital data by conversion into digital form of the quantity to be measured ( v) and inverse digital data by converting the inverse (-v) of the quantity & measure into digital form, calculating the difference between a digital data   direct and inverse digital data in order to develop a large   deur representing the result of the measurement, and calculate sums, each comprising the addition of direct digital data and reverse digital data, in   view of working out a quantity representing the error on the result.   2 Method according to claim 1, characterized in cecpe   direct and reverse digital data are generated alternately   tively and the result is determined by calculating the half-diff   reference between direct digital data and digital data reverse generated successively.   3 Method according to claim 2, characterized in that consecutive sums each formed by adding a datum   direct digital data and reverse digital data generated   discontinuously are calculated and the quantity representing the error on the result is developed from a set of sum values calculated.   4 o Method according to claim 1, characterized in that it comprises the steps which consist in: simultaneously generating at a first instant first and second direct digital data (ND 1, ND 2) by separate conversions of the quantity to be measured , then simultaneously generate at a second instant a third direct digital datum (ND 3) and an inverse digital datum (N II) by separate conversions of the quantity to be measured (v) and of the inverse (-v) of this quantity,   calculate the difference between the third data item   direct risk (ND 3) and the inverse digital data (N II) and add the difference between the second and the first direct digital data (EID 2, ND 1) to develop the quantity representing the result of the measurement at the second instant, and calculate algebraic sums each comprising, on the one hand, the addition of the third direct digital data (ND 3)   and the inverse digital data (N Il) and, on the other hand, the difference   between the first and second direct digital data (NDI, ND 2) to develop the quantity representing the error on the result.   5 Method according to any one of claims 1 to 4,   for the measurement of a quantity supplied by a sensor supplied with a direct voltage, characterized in that the inversion of the quantity to be measured is carried out by reversing the supply voltage of the sensor.   6 Method according to any one of claims 1 to 5,   according to which the quantity to be measured is transmitted to a converter   analog-digital via an amplifier   tion, characterized in that sums are calculated comprising the addition of direct digital data and inverse digital data in order to develop a quantity representative of the error   systematic affecting each digital data generated and we use   read this last quantity to produce a correction signal   analog which is injected into the amplification chain.   7 Voltage or current measuring device for   the implementation of the method according to any one of claims   1 to 6, comprising at least one measurement chain with analog-digital converter, characterized in that it further comprises: an inversion device (13, 33, 53) for inverting the quantity to be measured applied to the input of the measuring chain, a processing device (16, 36, 56) receiving the   digital data delivered by the or each converter into   response to the conversion, the voltage to be measured (v) and the reverse of the latter (-v), the processing device comprising means   to calculate in particular differences each between a numerical data   direct risk and an inverse digital datum in order to develop the values of the measurement results, means for calculating sums, each formed in particular by the addition of direct numerical datum and an inverse numerical datum, means for memorize a set of calculated sums and means to calculate from   memorized sums of quantities representative of the errors af-   fecting the results produced, and a control circuit (18, 38, 58) delivering control signals, in particular from the reversing device (13, 33, 53)   and or or each converter (11, 31, 51, 55).   8 Apparatus according to claim -7, for implementing the method according to claim 4, characterized in that it comprises a first measuring chain with an inverter (53) and a first analog-digital converter (51), a second   measurement channel with a second analog-to-digital converter   rique (55), the control circuit (58) delivering signals to control the converters in synchronism and the inverter to provide the processing device (56) a couple formed by first and second direct digital data (ND 1 , ND 2) issued   simultaneously by the first and the second converter then, ulte-   laughing, a couple formed by a third inverse digital data (N II) and a direct digital data (ND 3) delivered simultaneously   by the first and the second converter.   9 Apparatus according to any one of claims 7 and 8,   for the measurement of a quantity supplied by a sensor supplied by   a continuous power source, characterized in that the invert-   sor (33) is connected between the power source (30) and   sensor supply terminals (30).   Apparatus according to any one of claims 6 and 7,   including an amplification chain upstream of or each   analog-to-digital converter, characterized in that the   processing positive (46, 56) includes means for calculating a   systematic error value from a sum including the addi-   tion of direct digital data and reverse digital data, and a digital-analog converter circuit (49; 59, 59 ') whose input is connected to the processing circuit in order to develop a   analog correction signal injects into the amplifier   tion (42 a, 42 b; 52, 54) as a function of the calculated systematic error.