WO1995023956A1

WO1995023956A1 - Improvements in measuring apparatus

Info

Publication number: WO1995023956A1
Application number: PCT/NZ1995/000024
Authority: WO
Inventors: Steven Milsom Edmonds; Kenneth John Mardle; John Ross Mcmillan
Original assignee: Precision Transducers Limited
Priority date: 1994-03-04
Filing date: 1995-03-03
Publication date: 1995-09-08
Also published as: AU1864095A

Abstract

Correction apparatus for obtaining a corrected signal, whether in analogue or digital form, derived from or produced by measuring or transducer apparatus. The correction apparatus (10) includes a monitoring section (14) for determining the level of the signal from the measuring or transducer apparatus (13) and at least one factor (such as temperature) having an influence on the correlation between the signal and the affect or parameter being measured by the measuring or transducer apparatus (13). An adjustment section (15) is responsive to the monitoring section (14) to cause changes to the configuration of the signal derived from the circuitry of the measuring or transducer apparatus (13) to thereby result in an enhanced accuracy of the correlation between the signal and the affect or parameter being measured.

Description

IMPROVEMENTS IN MEASURING APPARATUS

The present invention relates to improvements in measuring apparatus and more particularly to apparatus for obtaining a corrected signal, whether in analogue or digital form, derived from measuring or transducer apparatus so that account is taken of inherent performance deficiencies and a compensation is applied to counter such deficiencies.

It has been long known to provide apparatus which is frequently ratiometric in nature for the measurement of various affects or parameters. For example, weighing apparatus often utilises various bridge measuring techniques. These are frequently referred to as full bridge, half bridge, quarter bridge or even multiple bridge arrangements and each produces a signal which in ideal conditions remains substantially constant in relation to an excitation or input-*, signal. This signal produced by the measuring apparatus may be of the same nature as the excitation signal (eg. volts, output for volts input) or the excitation arid output signals may be of different forms (eg. volts output for amperes input or a numerical digital representation output for volts input) .

In practice, however, such measuring apparatus generally does not provide an ideal correlation between the affect or parameter to be measured and the output signal produced. This can be due to material properties, circuit configuration, inherent design faults, or external influences such as temperature and/or pressure to mention but a few.

Weighing apparatus, which is only one of many types of measuring apparatus and utilises ratiometric measurement, commonly employs a strain gauge bridge arrangement to provide an output signal representative of the applied load or weight. Due to the exacting performance requirements for such apparatus errors of significance are often encountered. These errors may, for example, occur due to actual manufacturing procedures or they^' can arise because of inherent properties of the design. For example, the inherent nature of direct compression transducer elements can be a cause of non-linearity of output signal relative to the applied load as can poor assembly of the transducer. Also most components of weighing apparatus have temperature dependant properties which may lead to difficulties when attempting to maintain accurate performance of the apparatus over a range of operating temperatures. Still further time dependant properties can lead to errors occurring in conjunction with elapsed time.

Traditionally load cells have been manufactured to incorporate fixed circuitry to apply compensation for linearity, temperature and other effects or have relied on a correction to the signal being applied in a signal processing« display device. Adjustment of circuitry fixed within the load cell to compensate for effects mentioned above is a complex, precise and time consuming procedure which cannot be automated easily resulting in a high cost to the user of performance load cells. Correction of load cell defects within the display device has meant that load cells and displays are not easily interchangeable as one must be matched to the other.

Recent developments have seen the emergence of the logical load cell. For example, Sotoudek and Brignell (in a paper given at Weightech 1983) suggested signal processing circuitry to be located physically with a load cell in order to apply corrections to the output signal to gain an improvement in performance. Such logical load cells have been commercially developed by amongst others Toledo Scale Corporation (eg US Patent 4815547).

With these so called digital load cells changes are applied directly to the signal derived from or produced by the load cell and alter the nature of the signal by digitising same. While successful the load cells do have severe drawbacks. For example, because the signal is digitised within the load cell using an in-built analogue to digital converter the final resolution of the signal produced by the load cell is limited by the digitising process. This is due to the digitising process requiring a certain period of time and having fixed resolution. Thus the update rate or throughput of signal is limited with the result that these load cells are not so applicable to in-motion weighing or batching situations. Also these logical load cells have controlling circuitry attached to the source of signal and therefore any noise or distortion is introduced directly to the signal. Still further, the nature of these logical load cells is to provide the weight signal in a digital form with the result that the load cells cannot be installed with standard weighing equipment and reliability is greatly dependent on the reliability of many components.

The correction apparatus of the present invention incorporates circuitry which can be either built in at the time of manufacture or retrofitted This invention thus can be said to have as its primary object the providing of a means of improving the performance of measuring apparatus while still maintaining the basic nature of the signal derived from that apparatus.

Accordingly with the present invention the measuring apparatus performance is improved without any combination being made with the signal produced by the apparatus but rather by applying compensation directly at the load cell output or by reconfiguration of the circuitry components. This overcomes the introduction of distortion in the signal and the reliability of operation of the measuring apparatus can still be maintained. Also the derived signal is maintained in its original form whether that be digital or analogue. Broadly therefore in one aspect of the invention there is provided correction apparatus for obtaining a corrected signal, whether in analogue or digital form, derived from or produced by measuring or transducer apparatus, said correction apparatus including . monitoring means for determining the level of the signal and at least one factor having an influence on the correlation between the signal and the affect or parameter being measured, and adjustment means responsive to said monitoring means to cause changes to the configuration of the signal derived from the circuitry of the measuring or transducer apparatus to thereby result in an enhanced accuracy of the correlation between the signal and the affect or parameter being measured.

According to a second broad aspect the present invention provides a method of obtaining a corrected signal, whether in analogue or digital form, derived from or produced by measuring or transducer apparatus, said method including the steps of monitoring the signal and at least one factor having an influence on the correlation between the signal and the affect or parameter being measured and adjusting the configuration of the circuitry of the measuring or transducer apparatus to compensate for non-linearity and/or inaccuracy resulting from the influencing factor such that the measuring or transducer apparatus produces a corrected signal thereby ensuring correct correlation of the signal to the effect or parameter being measured.

In one form of the invention the correction apparatus is digital in nature whereby an analogue to digital converter provides a digital representation of the monitored signal and influencing factor(s) for processing by a microprocessor prior to an appropriate variation to the state of the apparatus is made under control of the microprocessor. In another form of the invention the correction apparatus is analogue in nature. According to this form the quantising effects occurring when an analogue value is given a digital representation are not present.

Consequently the following more detailed description of the invention will make reference to the invention in conjunction with both digital and analogue circuitry.

In the following more detailed description reference will be made to the accompanying drawings in which:-

Figure 1 is a schematic diagram of apparatus according to the present invention incorporated with measuring apparatus, Figure 2 is a circuit diagram of circuitry of the apparatus according to the present invention utilising both digital and analogue methods,

Figure 3 is a schematic illustration of apparatus according to the present invention when incorporated with circuitry appended to a load cell of weighing apparatus,

Figure 4 is a further circuit diagram of an embodiment of the invention, and Figure 5 is a logic flow diagram.

The apparatus of the present invention can successfully provide a method of correcting performance deficiencies of measuring apparatus. Such apparatus can, for example, be used for weighing, sensing or measuring forces, capacitance, light etc. The present invention does, however, have particular application to measuring apparatus where an effect upon the balanced state of the measuring apparatus or some portion of it can result in an undesirable change to the signal derived from the measuring apparatus. The apparatus according to the present invention can be retrofitted to existing apparatus without significant modification to the apparatus yet can provide a marked improvement in performance of the apparatus.

As stated above the apparatus can employ microprocessors and microcomputers or the like to simplify design or it can be implemented with purely analogue circuitry. Still further, however, the apparatus can combine both digital and analogue circuitry as disclosed herein.

Figure 1 of the drawings illustrates apparatus of the general type to which the correcting apparatus of the present invention is particularly applicable. Thus, for example, as shown, the correction apparatus 10 according to the present invention is coupled between the excitation signal input (lines 11 and 12) and the apparatus 13 from which an output signal is derived. The correction apparatus 10 effectively includes circuitry 14 to monitor signals and generate compensations and control and adjustment circuitry 15. Environmental sensors 16 are coupled to the monitoring and compensating circuitry 14 which in turn is coupled by line 18 to the derived signal output 19.

Determining the level of the derived signal by monitoring the signal could occur at any point from the first occurrence of a signal representative of the parameter being measured by the measuring apparatus 13 to a point after processing of the signal into its final form. The apparatus according to the present invention thus operates to apply compensation on a dynamic basis so that the signal derived from the apparatus as being representative of the parameter being measured or monitored is compensated for various performance deficiencies of the measuring apparatus. The term monitoring, as used herein, is taken to mean that the signal is observed when and as necessary. This can include continuously, intermittently or at intervals. Thus, for example, in analogue circuitry the analogue correction circuits will be continuously coupled to the derived signal with the output from the correction circuits being continuously dependent on the derived signal.

The measuring apparatus can, for example, be weighing apparatus (used for the determination of weights and forces) and for the purposes of further describing the present invention reference will be made to such apparatus though it will be appreciated by those skilled in the art that the present invention is not limited to use solely in conjunction with weighing apparatus.

In simple terms electronic weighing apparatus comprises at least one force transducer (commonly referred to as a load cell) to which is attached various fittings to support the transducer and to apply the load to be measured by the transducer. Electronic signal processing circuitry is incorporated to provide a display to the user of the weight on the apparatus. As such apparatus and its configuration is well known and can be of conventional construction a detailed description herein is not required to fully disclose apparatus according to the present invention.

A typical load cell arrangement incorporating apparatus of the present invention as an integral part is shown in Figure 2. As illustrated the correction apparatus 10 is coupled to the excitation signal inputs 11 and 12 and to a bridge circuit arrangement 20. In the illustrated arrangement the environmental sensor 16 is a temperature sensor but could include other sensors, pressure, etc.. Coupled to the correction circuitry are digital communication interfaces 21a and 21b. Interface 21a is a high speed interface and 21b, utilising a signal encoded on the excitation inputs, is of lower speed.

While in the arrangement shown in Figure 2 the correction apparatus 10 is shown as deriving a signal from the representation of weight directly at the sensing bridge 20 this could in fact be performed after some analogue or digital signal processing and in some instances, such as in the case of newly emerging digital load cells, may be the preferred method.

With the arrangement shown in Figure 3 the correction apparatus according to the present invention is coupled with an existing load cell 13.

In Figure 2 of the drawings, digital circuit means 23 (such as a microprocessor, microcomputer or digital signal processor) is used to calculate the necessary factors and apply appropriate compensation to the force transducer or load cell bridge 20. To effect a saving in components this digital circuit means, although shown separate from the load cell bridge 20, may, as indicated above, be in whole or in part incorporated into the circuitry of the load cells or other apparatus which utilises the signal, eg. the display unit or a computer or programmable logic controller.

The microprocessor 23 is programmed with the various relevant characteristics of the load cell, as observed, and with reference to the parameters of time, temperature, derived signal etc. Microprocessor 23 appropriately calculates the setting of the digital to analogue converters 25a and 25b to bring about a change in the balanced state of the circuit, ie. configuration of the circuit which will result in the derived signal (lines 19a and 19b) providing a better correlation to weight. The accuracy of the weighing apparatus will thus be improved.

In operation the voltage from the derived signal lines 19a and 19b is converted to a digital signal by the analogue/digital converter 24 and made available to the microprocessor 23. Also an input from the temperature sensor 16 is applied to the microprocessor 23 as is a signal from compensating circuits 26,27 and 28,29. Thus from this data and characteristics stored in memory (either pre-programmed or calculated from observed data by the microprocessor 23 as time goes by) the microprocessor will deem various compensations to the circuit balance to be necessary. An equation and/or stored table may be used to obtain a compensation for non-linearity dependant on the signal voltage. Also equations or stored values or stored tables of values may be used to obtain a compensation dependant on the information received from the temperature sensor 16, circuitry 26-29, and time as determined by cycles of the microprocessor clock or some other timing device (not shown). Compensation for other factors, as applicable to the various types of apparatus in use, would be determined and from these compensations and the program of the microprocessor the appropriate signal will be sent to the digital to analogue converters 25a and 25b to bring about the appropriate variation to the balanced state of the bridge 20.

Also shown in Figure 2 are communications interfaces 21a and 21b to enable communication with microprocessor 23. By means of these interfaces it is possible to make the correction apparatus responsive to external control; enable the correction apparatus to control other devices; or purely transmit data in an uncontrolled manner for use by other devices. Interface 21 thus makes it possible to amend or append the program controlling the microprocessor, to transfer the characteristics of the apparatus if required for compensation of performance deficiencies and to receive information and data from the microprocessor such as values of measured parameters, constants in use, self descriptive data and historical information such as maximum loads etc.

Thus weighing apparatus is provided which has been compensated for many of the errors common to such apparatus and with the additional advantage provided by the use of digital techniques of availability of process information prepared in a format to facilitate its ease of use.

Figure 2 illustrates the use of both digital and analogue techniques. The time dependent error estimation circuitry 26 to 29 could be implemented in microprocessor software to produce an arrangement primarily digital in the nature of compensation, as alternatively the digital circuitry could be replaced with analogue arrangements to derive compensations for the effects of non-linearity and temperature and to apply these compensations to bridge 20.

A completely analogue arrangement does not have the advantages of flexibility and external communication provided by the earlier described arrangement of Figure 2. Thus in practice it is anticipated that a commercial implementation of the present invention may incorporate both digital and analogue circuitry. For example, as shown in Figure 2, the means for correcting the signal of an existing load cell may be used with digital processing circuitry as may one or more analogue circuit means for obtaining an indication of a particular compensation that may be required.

Time dependant error correction circuitry formed by amplifiers 26 to 29 is shown in Figure 2. This circuitry is novel in that it simulates the characteristics of load dependant non-linear time dependant errors of a system. The integration of the representation of load with time is continuous and with the correct selection of resistor and capacitor values (such selection being made after testing of the apparatus or representative apparatus to determine the characteristics varying with time) and subsequent scaling of the signal from the circuitry provides a true instantaneous representation of the non-linear time dependant errors of the apparatus. This can be used to apply corrections or compensation to the apparatus to enable a signal to be derived from the apparatus which has greatly reduced non¬ linear time dependant errors.

The circuitry incorporating amplifiers 26 to 29 shown in Figure 2 is fairly simple in nature and simulates exponential growth. The circuitry including amplifiers 26 and 27 provides for compensation of a first exponential component of non-linear time dependant error, and circuits 28-29 a second.

From testing of the representative apparatus to be corrected it is possible by trial and error or other means such as least squares regression curve fitting to obtain coefficients for the following equations such that the equations closely predict the error found by test or observance of in-use performance of the apparatus. S = S_to[ (l-e^"kt)*A+(l-e^"ct)*B + 1] or S_t-S_to= S_to(A(l-e"^kt)+B(l-e~^ct)) Equation 1 where S^ = signal value at time t, measured parameter constant S-to ⁼ signal value obtained at the initial instant when the external parameter to be measured is applied after a long period of stabilisation hence S^-S^o ⁼ change in signal value with time for measured value constant A and k are constants pertaining to circuits 26-27 B and C are constants pertaining to circuits 28-29.

For circuits 26-27 the equation defining output voltage for constant input voltage is:-

where Vo = Output Voltage at 27 V-L = Input Voltage at 26.

With reference to equation 1, it can be seen that 1/RC corresponds to constant K and scaling factor A is implemented within the microprocessor. RC is the time constant in Ohms x Farads of Ri and C- . A similar relation applies between R₂, C2 and constant C from equation 1. Although shown in Figure 2 in analogue circuitry form the functions of this circuitry 26-29 can be implemented digitally in the program of the microprocessor 23 shown in Figure 2. This has the advantage of savings in components. It does, however, have the disadvantage of no longer being instantaneously correct in determination of non-linear time dependant error, as well as not being continuous in nature. To obtain the advantages of using a microprocessor based system and having continuously determined representations of non-linear time dependant errors which are correct at any instant analogue circuitry as described above can be used which feeds directly into an analogue/digital converter 24 and the microprocessor. If it is desired to implement the analogue circuitry functions of amplifiers 26 and 27 in microprocessor programming it would be done as follows, duplicated for amplifiers 28 and 29.

An accumulator (ACCl) is incremented by the scaled signed difference between ACCl and the representation of load (LD). ACCl = Kt (LD-ACC1) + ACCl

The contents of ACCl multiplied by constant C provide a representation of the non-linear time dependant error to be expected from the apparatus if uncorrected. Constants K & C are determined by test of the apparatus or determined during use of the apparatus to provide the best representation of error. t is the time since last performing this equation.

The sequence of steps to determine error are

(1) Obtain a representation of weight on the scale

(2) Subtract to obtain the difference between ACCl and the weight

(3) Add Kt times this difference to ACCl Steps 1 to 3 are repeated continuously. An equally effective means of compensating for non-linea time dependant errors can be achieved by use of digital filtering techniques, such as finite impulse response, which are commonly known in signal processing methods.

At any point in time the contents- of ACCl can be used to reduce the effect of non-linear time dependant errors on the derived signal from the apparatus.

The present invention makes no digitisation of the output signal to limit resolution, has no effect on the response to load, is configured such that the effect of noise on signal of external circuitry is attenuated up to 1000:1, is such that load cells utilising this invention can be used and mixed with most standard analogue weighing equipment and is failsafe. Failure of the electronic circuitry will not prevent operation, only reduce performance to that of the apparatus prior to enhancement by the present invention.

As stated previously the correction apparatus can be implemented utilising digital electronic circuitry, analogue electronic circuitry or a combination of techniques. For purposes of further describing the invention the following additional example will include analogue signal amplification, buffering and filtering as well as digital processing of information and data and control of circuitry by means of a digital micro processor or digital signal processor. As also mentioned previously when digital means are utilised, corrections to be applied to the device could be derived from values stored in tabular form or could be derived by equation from fewer characteristic values. For purposes of the example following the use of tabular values has been used for simplicity and the apparatus to be improved is a load cell. In Figure 4 there is a load cell 13, the performance of which is to be improved and the correction apparatus 10. In this example operation is as follows.

An excitation voltage is applied to the load cell 13 resulting in a signal which is proportional to the load applied and the excitation applied with the exception that the proportionality to load may contain an element of error. Due to the resistive bridge circuit configuration load cell signals are ratiometric to the excitation voltage.

The excitation and signal voltages are fed to analogue circuitry 35 for scaling, filtering and buffering. Temperature sensor 16 is positioned to sense load cell temperature and feeds a signal into circuit 10 also for analogue conditioning. Conditioned signals are then fed to an analogue to digital converter 24 for conversion of analogue signals to digital representations for purposes of determining a correction to be applied, rather than as a primary means of accurately transmitting the weight as determined by the load cell. Converter 24 along with other processes enclosed in dashed line representing apparatus 23 are digital in nature and performed in a single chip microprocessor or digital signal processor as it is sometimes known. From converter 24 values are determined for temperature and absolute load (ie the value for load remains constant with varying excitation voltage) and from these values corrections are calculated for linearity error correction, shift in signal with time, shift in sensitivity with temperature and shift in zero signal with temperature.

Determination of linearity error correction is performed by process 36 as explained under non linearity of signal with load following. This produces a correction based on load as perceived by the circuitry and which it is deemed will reduce the linearity error of the load cell. Signal correction for shift in load cell signal with time is determined with process 37 by means of a finite impulse response filter (FIR) as explained under shift in load signal with time following. This produces a correction based on load and temperature and which it is deemed calculated will reduce the error of the load cell signal due to shift over time.

A correction for shift in load cell signal at zero load with temperature is determined by process 39 as explained under shift in zero signal with temperature following. This produces a correction based on temperature and which it is calculated will reduce the error of the load cell due to zero load signal shift with temperature.

A correction for shift in load cell sensitivity with temperature is determined at 38 and by subsequent multiplication by load at 31 as explained under shift in sensitivity with temperature following. This produces a correction based on temperature and load and which it is calculated will reduce the error of the load cell signal due to shift in sensitivity with temperature.

All corrections for individual load cell performance inadequacies determined by the stages previously explained are summed at 30 before being fed as one net correction to a pulse width modulator at 32 which is buffered at 33 before connection to the load cell signal wires by means of scaling resistors 34. The pulse width modulator switches the output lines rapidly between excitation + and excitation - with varying proportions of time spent connected to each of the two excitation lines. Consequently when the proportion of time connected to + excitation is equal to the time spent connected to the - excitation there is no average offset applied to the signal and the signal is not altered. Switching is performed at such a speed that stability of the signal as processed by other signal conditioning circuitry attached to the load cell for conditioning of load reading (in contrast to the correction apparatus) is not compromised. As the proportion of time spent connected by PWM 32 to each of the excitation lines is varied in proportion to the summed correction, so the signal lines of the load cell will be pulled on average in the positive or negative direction, in opposite directions, producing an offset at the load cell signal. By means of calibration resistors 34 this offset is scaled to represent the sum of the corrections from 30 calculated to be necessary to reduce the error in the load cell signal to a minimum.

Reference is made in explanation of the above to the following derivation of tables of figures by use of curve fitting techniques upon observed or measured data. These techniques are well known by skilled data analysts and the observed or measured data may be gained from individual apparatus to which the invention is to be applied, representative items of apparatus, batches, families or derived by other means as provide the required degree of error reduction by the correction apparatus.

Tables of values are stored in electronic memory devices constituting a part of the correction apparatus according to the present invention. These values can be programmed at the time of programming of the microprocessor of the apparatus or can be programmed at a later date by means of the bi¬ directional serial communications section 21 which also allows for communication of status, configuration, diagnostics, and historical data to and from the correction apparatus.

NON LINEARITY OF SIGNAL WITH LOAD

A correction to be applied for non-linearity between signal and load is determined from a table of values. A table is constructed for example of 256 values representing corrections required at each of 256 loading points spaced between zero and approximately 120% of nominal capacity of the load cell. The load applied to the load cell as determined by the microprocessor is used to determine the appropriate correction from the table to be applied, being the value with corresponding temperature closest to perceived temperature.

Where LN = Nominal Load Cell Capacity

Ln = Load Corresponding to the nth Correction Value LCn = nth Correction Factor at Ln

LCp = Correction at Perceived Load Lp = Perceived Load

And assuming 255 equally spaced steps between zero load and 120%.

n = Integer Value of (Lp/LN/1.2 * 255 + 0.5) LCp = LCn

Improved accuracy for the number of correction values stored can be gained by interpolating between values LCn and LCn+1 when load is between Ln and Ln+1. A balance must be chosen between computational requirements and storage available for factors.

A means by which the stored correction values for linearity are determined is by calculation of each tabular value from a formula, the formula being the equation of a best fit curve fitting measured linearity error vs load.

SHIFT IN ZERO SIGNAL WITH TEMPERATURE

A correction to be applied is determined from a table of values. A table is constructed for example of 101 values representing corrections required at each of 101 temperatures covering the temperature range over which the load cell is to be corrected for shifts in zero reading with variation in temperature. For example corrections may be tabulated corresponding to temperatures from - 40 to + 60 ° C in 1°C steps, 101 values in all.

The temperature of the load cell as determined by the temperature sensor and microprocessor is used to determine the appropriate correction from the table to be applied, being the value with correspondence temp closest to perceived temperature.

CORRECTION

PERATURE

Where Tp = Perceived Temperature

Tn = Temperature Correspondence to nth Correction

Value

ZCp = Correction to be Applied at perceived Temp

ZCn = Correction Factor at Tn

N = Integer Value of (Tp + (Tn+l-Tn)/2)- 40

ZCp = ZCn Improved accuracy for the number of correction values stored can be gained ^'by interpolating between corrections ZCn and ZCn+1 when temperature is between Tn and Tn+1. A balance must be chosen between computational requirements and storage available for correction values. A means by which the stored correction values for zero signal shift with temperature are determined is by calculation of each tabular value from a formula, the formula being the equation of a best fit curve fitting measured zero signal error vs temperature.

SHIFT IN LOAD SIGNAL WITH TIME

A correction to be applied for the shift in load signal with time is determined instantaneously from the output value from a finite impulse response (FIR) filter. An example of such a filter, which could be implemented in either software or hardware, would consist of 2048 input registers and 2048 corresponding coefficient registers. (IRo IR2047 and CRo CR2047). In this example the load perceived at an instant in time to is stored in IRo. At a period of time later, tl as determined appropriate for the filter characteristics required, and which for example may be 1 second after to , the load perceived is stored in IRo, the contents of IRo having just previously been stored in IRl and so on through to IR2047. The movement of such input values through IRO to IR 2047 can be achieved by either moving values from register to register or by incrementing or decrementing the register index as determined appropriate. At any point in time to the contents of registers IRo through to IR 2047 contain respectively the values of load perceived LPo at to through to Lp 2047 at t - 2047, the load 2048 seconds previous.

Each input register IRn has a corresponding coefficient register CRn containing an FIR co-efficient, the output of the FIR filter is defined to be the sum of all IR * CR and is the correction to be applied for errors due to shift in load signal with time.

n=2047 TCp = Sum (IRn * CRn) n=0

Where TCp = Correction to be applied at any time and perceived load. IRn = Contents of nth load register.

CRn = Contents of nth co-efficient register.

In the example above the contents of registers were updated every second. Registers need not be spaced with equal time periods for simplicity but can be spaced variably to optimise memory locations as determined by the designer to be most appropriate.

As it is common for the shift in load signal with time to also vary with temperature a table of FIR co-efficient is constructed at each of a number of temperatures. 12 tables for example may be constructed, each table being values of co-efficients to be used at each of 12 temperatures from -10 to +45°c spaced evenly at 5'c increments. In operation the factors in the table corresponding to the temperature closest to the perceived temperature would be used at the FIR co¬ efficients in the FIR filter.

Improved accuracy for the number of co-efficient values stored can be gained by interpolating between co-efficients from the two tables corresponding to the two temperatures placed either side of the perceived temperature. A balance must be chosen between computational requirements and storage available for co-efficient values.

The co-efficient values in the tables are determined by calculation from a formula or formulae which best describe a best fit curve or series or curves fitted to observed values of error due to shift in signal over time and temperature as is commonly performed by data analysts.

SHIFT IN SENSITIVITY WITH TEMPERATURE

A correction to be applied is determined from a table of values. A table is constructed for example of 101 values representing factors by which the perceived load must be multiplied to derive a correction for shift in sensitivity with temperature. For example factors may be tabulated corresponding to temperatures from -40 to +60"c in l°c steps, 101 values in all. These factors when multiplied by the load on the load cell as a fraction" of nominal load cell capacity produces the correction to be applied. The temperature of the load cell as determined by the temperature sensor and microprocessors is used to determine the appropriate factor from the table to be used, being the value with corresponding temperature closest to perceived temperature.

FACTOR

MPERATURE

Where TP = Perceived Temperature

Tn = Temperature Corresponding to nth Factor

Lp = Perceived Load

Ln = Nominal Load Cell Capacity

SCp = Sensitivity Correction to be Applied at

Temperature Tp and Load Lp n = Integer Value of (Tp + (Tn + 1 - Tn) / 2) - 40 SCp = Fn * Lp / Ln Improved accuracy for the number of correction factors stored can be gained by interpolating between values Fn and Fn+1 when temperature is between Tn and Tn+1. A balance must be chosen between computational requirements and storage available for factors.

The co-efficient values in the tables are determined by calculation from a formula which describes a best fit curve fitted to observed values of error due to shift in sensitivity with temperature.

Figure 5 is a flow diagram in simplified form showing the main flow in processing. This follows closely the operation described in relation to the correction apparatus of Figure 4. Although not showing all processes and functions in full detail it nevertheless provides sufficient information to enable the skilled person to bring the present invention into effect.

The foregoing description of embodiments of the correction apparatus according to the present invention are by way of example and the invention is not limited thereto. Further embodiments will be readily apparent to those skilled in the art and fall within the scope and spirit of the invention as defined in the following claims.

Claims

1. Correction apparatus for obtaining a corrected signal, whether in analogue or digital form, derived from or produced by measuring or transducer apparatus (13), said correction apparatus (10) including monitoring means (14) for determining the level of the signal and at least one factor having an influence on the correlation between the signal and the affect or parameter being measured by said measuring or transducer apparatus (13), and adjustment means (15) responsive to said monitoring means (14) to cause changes to the configuration of the signal derived from the circuitry of the measuring or transducer apparatus (13) to thereby result in an enhanced accuracy of the correlation between the signal and the affect or parameter being measured.

2. Correction apparatus as claimed in claim 1 wherein the monitoring means (14) is operative to monitor the signal continuously, intermittently or at intervals.

3. Correction apparatus as claimed in claim 1 or 2 wherein the monitoring means (14) is coupled to at least one environmental sensor (16).

4. Correction apparatus as claimed in claims 1, 2 or 3 wherein the monitoring means (14) is a microprocessor (23).

5. Correction apparatus as claimed in claim 4 wherein the microprocessor (23) includes an analogue to digital converter (24) to which is applied excitation and signal voltages from the measuring or transducer apparatus (13).

Correction apparatus as claimed in claims 4 or 5 wherein the microprocessor (23) includes processor means (25 to 29) for calculating signal correction as a consequence of one or more of:-

(a) non linearity of signal with load; (b) shift in signal with time;

(c) shift in zero signal with temperature; and

(d) shift in sensitivity with temperature.

7. Correction apparatus as claimed in claim 6 wherein the microprocessor includes means (30) for summing the outputs from said processor means (36 to 39) there being a pulse width modulator (32) coupled to said means for summing (30) and connected to the excitation inputs (11 and 12) .

Correction apparatus as claimed in claim 7 wherein one or more of said processor means (36 to 39) is coupled to the pulse width modulator (32) via multiplier means (31).

9. Correction apparatus as claimed in claim 7 or 8 wherein the pulse width modulator (32) is coupled to signal lines (19a and 19b) from the measuring or transducer means (13) via buffer means (33) and scaling resistors (34).

10. Correction apparatus as claimed in any one of claims 1 to 9 wherein the measuring or transducer means (13) is a load cell.

11. Correction apparatus as claimed in claim 6 wherein the correction to be applied for the shift in load signal with time is determined instantaneously from the output value of a finite impulse response filter (37).

12. A method of obtaining a corrected signal, whether in analogue or digital form, derived from or produced by measuring or transducer apparatus, said method including the steps of monitoring the signal and at least one factor having an influence on the correlation between the signal and the affect or parameter being measured and adjusting the configuration of the circuitry of the measuring or transducer apparatus to compensate for non- linearity and/or inaccuracy resulting from the influencing factor such that the measuring or transducer apparatus produces a corrected signal thereby ensuring correct correlation of the signal to the effect or parameter being measured.