GB1604939A

GB1604939A - Transducer circuits and components to correct for environmental effects

Info

Publication number: GB1604939A
Application number: GB2600778A
Authority: GB
Original assignee: Acco Industries Inc
Current assignee: FKI Industries Inc
Priority date: 1978-03-09
Filing date: 1978-05-31
Publication date: 1981-12-16
Also published as: GB1604940A; FR2419510A1; DE2823875A1; FR2419510B1

Description

(54) IMPROVEMENTS IN OR RELATING TO TRANSDUCER CIRCUITS AND COMPONENTS TO CORRECT FOR ENVIRONMENTAL EFFECTS (71) We, ACCO INDUSTRIE INC. (formerly American Chain & Cable Company, Inc.), a corporation organised under the laws of New York, United States of America, of 929 Connecticut Avenue, Bridgeport, Connecticut 06602, United States of America, do hereby declare the invention for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement:- IMPROVEMENTS IN TRANSDUCER CIRCUITS AND COMPONENTS ABSTRACT OF THE DISCLOSURE Improvements are described for obtaining temperature and span compensation in transducers such as semiconductive strain gauges. Temperature compensation at three temperatures is achieved by an iterative process which determines the optimum distribution of a span compensation resistance between two resistors. Span compensation is achieved by varying the voltage applied to the strain gauges. This technique is generally applicable to any type of device made of at least two elements having a resistance, capacitance or inductance which is a function of at least two environmental variables. Circuits are described for suppressing the zero output of such temperature and span compensation circuits as well as producing non-linear outputs as claimed in Application No. 80/39967 (Serial No. 1 604 940). Apparatus for providing overrange protection is also disclosed.

BACKGROUND OF THE INVENTION This relates to improvements in strain gauge transducers and their components.

While the invention will be described in terms of its application to strain gauge transducers it will be recognized that the principles disclosed may have application to other devices.

A strain gauge is typically used as atransducer by bonding it to a flexible object and measuring the change in voltage across the gauge or the change in gauge resistance as different loads are applied to the object. It is particularly advantageous to use a Wheatstone bridge in which two strain gauges are connected in series on one side of the bridge and two resistors are connected in series on the other side. Each of these four elements is in a separate arm of the bridge with the supply voltage applied to the nodes between the two sides of the bridge and the output voltage measured between the node between the two resistors and the node between the two strain gauges. Since the function of the two resistors is to provide a reference voltage at the node between them, their side of the bridge will be referred to as the reference side. If the gauges are mounted on opposite sides of the object so that bending of the object applies a tensile loading to one gauge and a compressive loading to the other, the changes in resistance of the gauges tend to be equal in magnitude but opposite in polarity. For these conditions, the ratio of the resistances of the two strain gauges is a function of the amount of deflection in the object. Hence, the output voltage can be related to the amount of deflection in the object.

Recent improvements in the art have led to increasing use of semiconductor strain gauges. As is well known, such strain gauges offer significant advantages over prior art foil or wire strain gauges since the sensitivity of the semiconductor gauges is hundreds of times greater than that of typical metallic gauges. However, semiconductor strain gauges have both a large temperature coefficient of resistance and a large temperature coefficient of gauge factor or sensitivity. Thus, both their resistance and their rate of change of resistance with applied stress vary appreciably with temperature. Semiconductor strain gauges can be made so that these temperature coefficients in different devices are approximately the same. However, when the gauges are bonded to an object, certain uncontrollable temperature induced strains are created that modify the temperature coefficients of the gauges. As a result, the voltage output from the bridge is a function of temperature.

This variation of voltage with temperature causes two major errors in the output of a semiconductor strain gauge transducer. When the gauges are under zero stress, the ratio of the resistances of the two strain gauges at one temperature will be different from the ratio of the resistances of the strain gauges at a second temperature. Correction of this effect is called temperature compensation or constant value compensation hereinafter. In the prior art it is known to use a series/parallel network of resistance to offset the effects of the temperature coefficient of resistance enough that the ratio of the resistances of the two strain gauges at two different temperatures is identical.

The second error is attributable to the change in strain gauge sensitivity with temperature. For a circuit which is temperature compensated at two temperatures, when maximum loading is applied to the strain gauges the output voltage observed at one of the two compensation temperatures is diffierent from that observed at the other compensation temperature. Correction of this effect is called span compensation. In the prior art it is known to add a resistor in series or parallel with the bridge to make the output voltage the same at the two compensation temperatures.

While temperature and span compensation at two temperatures does improve the performance of a semiconductor strain gauge transducer as a measuring device, the resistance ratios of the strain gauges and the output voltages of the circuit are not the same at other temperatures because of the complex effects of the temperature induced strain in the gauges. In addition, the use of temperature and span compensation circuits complicates the provision of desirable circuit features such as "suppressed zero" and non-linear outputs. Finally, because semiconductive devices are easily damaged if overloaded, it is desirable to provide overload protection for semiconductor strain gauges.

SUMMARY OF THE INVENTION To provide for improved compensation in strain gauge circuits and the like, a circuit design method has been devised which compensates for temperature effects in such circuits at three different points.

Beginning with a bridge circuit comprising two series resistors in one arm and an arbitrary resistance in series with two strain gauges in the other, the voltage drops across the strain gauges and arbitrary resistance are measured at zero and maximum stresses at three temperatures. The appropriate resistance for span compensation at the two extreme temperatures is then calculated from these values.

An iterative process is then commenced in which the span compensation resistance is distributed between a first resistor connected in series with one of the strain gauges in one arm of the bridge circuit and a second resistor connected in series with the other strain gauge in a second arm of the bridge circuit. Initially, the first resistor is assigned a value of one ohm and the second resistor the remainder. With these values the series/parallel resistances required for temperature compensation at zero stress are calculated for the two extreme temperatures. The output voltages at zero stress are then calculated for both the maximum temperature and the intermediate temperature and the difference between these outputs is obtained. Next, the resistance of the first resistor is increased by one ohm and that of the second resistor is decreased by one ohm; and the series/parallel resistances required for temperature compensation are again calculated for the extreme temperatures. From these values the output volt ages at zero stress are calculated for both the maximum and intermediate temperatures. The difference between these outputs is then compared with the difference previously calculated, and whichever value is closer to zero is retained along with the necessary circuit parameters. The foregoing process is repeated for each value of the first resistor less than the span resistance, at the end of which process the retained circuit parameters define the resistance values of the bridge circuit which produce the best three-point temperature compensation.

This technique is generally applicable to any electrical circuit made of elements having a resistive, capacitive or inductive effect for which at least two of such elements have an output which is a function of two variables.

When the bridge circuit is span compensated at two temperatures, the span error between the two temperatures reaches a peak at an intermediate temperat1lre about midway between the two temperatures. At temperatures above the upper of these two temperatures and below the lower, the span error has opposite sign and increases in magnitude both as the temperature increases above the upper temperature and as it decreases below the lower temperature. In these circumstances, improved span compensation is achieved by varying the voltage applied to the bridge circuit. To this end, a resistor and thermistor are conected in series between the two input nodes of the bridge circuit and another resistor and thermistor are connected in parallel between one input node of the bridge circuit and one terminal of a voltage supply.

A second terminal of the voltage supply is connected to the other input node of the bridge circuit. The values of the two resistors are selected so that there is substantially no span error at the intermediate temperature where there otherwise would have been a peak in span error. Since the thermistors have a negative temperature coefficient of resistance, the resistance of the series combination of a resistor and a thermistor increases with decreasing temperature thereby increasing the voltage applied to the bridge circuit. The values of the thermistor and its temperature coefficient of resistance are selected so that this increase in voltage tends to offset the changes in output voltage that would otherwise occur at temperatures less than the intermediate temperature. With respect to the parallel combination of the resistor and thermistor, the resistance of this net-work decreases with increasing temperature thereby increasing the voltage applied to the bridge circuit. The value of the thermistor and its temperature coefficient of resistance are selected so that this increase in voltage tends to offset the changes in output voltage that would otherwise occur at temperatures greater than the intermediate temperature.

The foregoing temperature and span compensation techniques are not compatible with conventional techniques for providing zero suppression and/or non-linear outputs. Zero suppression, however, can be produced in accordance with the invention by usin ga constant current source to alter the reference voltage generated in the arm of the bridge circuit which does not contain the strain gauges. Non-linear outputs can be produced by feedback circuits in the output from the compensated bridge circuit.

Calibration of such outputs can be achieved by appropriate use of a constant current source.

Because semiconductive strain gauges are relatively delicate, it is also desirable to provide overload protection in the mechanical construction of the device in which they are mounted. Where the gauges are mounted on a cantilever beam, such overload protection includes stops which prevent movement of the beam much beyond that necessary for mull scale reading.

BRIEF DESCRIPTION OF THE DRAWING These and other objects, features and elements of the invention will be more readily apparent from the following detailed description of the invention in which: Fig. 1 is a block diagram of an illustrative embodiment of the invention; Fig. 2 is a graph useful in understanding certain features of the invention; Fig. 3 is a block diagram of a second illustrative embodiment of the invention; and Fig. 4 is a schematic cross-section of the mechanical construction of an illustrative embodiment of the invention.

DETAILED DESCRIPTION OF THE INVENTION In the illustrative embodiment of Fig. 1, circuit 10 comprises a first constant current source 20, a voltage regulator 25, a span compensation network 30, a temperature compensated bridge circuit 40, an amplifier 70, a transistor 75, a second constant current source 80, a feedback resistor 85 and scaling resistors 92, 94, 96.

Circuit 10 is connected at terminals E and F to a two wire line 120 which connects a power supply 130 and a resistive load 135 to circuit 10. Power supply 130 provides all the necessary power to circuit 10 via two wire line 120. Using the same two wire line the output of circuit 10 is transmitted in the form of a variable current back to load 135. There the variable signal is sensed as a voltage drop across the load.

First constant current source 20 is connected in series between terminal E and the voltage regulator which illustratively is a Zener diode. The other side of the voltage regulator is conected to node B of bridge circuit 40.

The span compensation network comprises a first resistor 32 and a first thermistor 34 connected in parallel between the voltage regulator and a node A of bridge circuit 40 and a second resistor 36 and a second thermistor 38 connected in series between node A and node B of the bridge. Circuit 40 is isolated from network 30 by buffer amplifier 39. As will be recognized, resistors 32, 36 have a positive temperature coefficient of resistance while thermistors 34, 38 have a negative temperature coefficient of resistance.

In a reference half of bridge circuit 40 are resistors 42 and 44 connected in series in one arm and a resistor 46 in a second arm. In the variable half of the bridge circuit are a resistor 52 and a variable resistance 54 connected in series in one arm and a resistor 56 and a variable resistance 58 connected in series in another arm. A resistor 62 is connected in parallel across one or the other of the variable resistances 54, 58 as will be explained below. Illustratively, the variable resistances are produced by strain gauges mounted on opposite sides of a flexible object as shown in Fig. 4.

In this position flexing of the object produces tensile loading on one gauge and a compresive loading on the other gauge, thereby producing changes in resistance of approximately equal magnitude but opposite sign. Advantageously, the strain gauges are semiconductor devices, with the result that their resistance and their sensitivity vary appreciably with temperature. In contrast, over the operating temperatures of interest, resistors 32, 36, 42, 44, 46, 52, 56, 62 have a substantially constant value and in normal operation of the circuit are not subjected to stress. By way of example, the strain gauges may be DSC cartridges available from Kistler-Morse Inc. of Bellevue, Washington, U.S.A., and the resistors may be conventional metal film or wirewound resistors.

Amplifier 70 is a high-gain differential amplifier. One input terminal to amplifier 70 is connected to node C between variable resistances 54, 58 and the other input terminal is connected to node D between resistors 44, 46. The output of amplifier 70 is applied to the base electrode of transistor 75 to regulate current flow therethrough.

As shown in the drawing, the emitter and collector terminals of transistor 75 are connectel between terminals E and B of the circuit. Thus, current flow through the two wire line 120 is regulated by transistor 75 in accordance with the output signal from amplifier 70. Feedback is provided by resistor 85 connected between node F and the non-inverting terminal of amplifier 70.

Between node F and node B, a selector switch 98 connects one of scaling resistors 92, 94, 96 in the path of current flow through two wire line 120. As their name suggests, these scaling resistors in conjunction with feedback resistor 85 permit the same output signal on line 120 to represent different voltages at the inputs to amplifier 70.

For example, a four milliamp output signal might correspond to zero pressure dif frrental in all cases; but when switch 98 connects resistor 92 in the circuit a 20 milliamp output signal might correspond to a 20 pounds per square inch pressure differential and when the switch connects resistor 94 in the circuit such an output signal might correspond to a 40 pounds pressure differential.

Constant current source 80 is connected between node B and a node between resistors 42, 44 in the bridge circuit. As will be described below, current source 80 may be used to suppress or modify the zero level in the variable being sensed by the bridge circuit. In particular by applying current from source 80 to the node between resistors 42, 44 the four milliamp output signal can be made to correspond to a nonzero value of the variable being sensed.

The resistance values of resistors 52, 56, 62 are selected by an iterative process so as to achieve temperature compensation at three different temperatures and span compensation at two different stresses at the two extreme values of said three temperatures. More particularly, arbitrary resistance AR is initially connected in series with variable resistances 54, 58 between nodes A and B of the bridge circuit. Resistance AR has a substantially constant value over the operating temperatures of interest.

For this circuit, the voltage drops across each of resistances 54, 58 and arbitrary resistance AR at zero, stress and at maximum stress are measured for three different temperatures. As is apparent from the description of the mounting of the strain gauges, the condition of maximum stress means maximum tension for one gauge and maximum compression for the other. From these values the appropriate resistance SSR is then calculated for span compensation at the two extreme temperatures of the three temperatures measured using the following equations: VBR(H,S) - VBR(C,S). (VO(H)., VO(C)) SSR = iBR(C,S) . (VO(H)fVO(C))- iBR(H,S) where VBR(H,S) = VGA(H,S)+VGB(H,S) VBR(C,S) = VGA(C,S)+VGB(C,S) VO(H) = (ARGA(H,Z)) iBR(ll,Z) - (ARtGA(H,S))iBR(H,S) VO(C) = (AR+GA(C,Z))iBR(C,Z) - (AR+GA(C,S))iBR(C,S) iBR(H,Z) = VAR(H,Z) 'AR iBR(H,S) = VAR(HS) 'AR iBR(C,Z) = VAR(C,Z)/AR iBR(C,S) = VAR(C,S) AR H = the uppermost of the three temperatures C = the lowermost of the three temperatures Z = zero stress S = maximum stress and VGA and VGB are the measured voltage drops across resistances 54, 58, respectively, at the temperatures and stresses indicated; VAR is the measured voltage drop across the arbitrary resistance AR at the temperatures and stresses indicated; and GA is the resistance of variable resistance 54 at the temperatures and stresses indicated.

As will be evident, the resistance of variable resistances 54, 58 can readily be determined from the measured voltage drop across each gauge and the current flow iBR in that arm of the bridge at the temperatures and stresses of interest.

Next, an iterative process is commenced to find the optimum distribution of the span resistance between resistors 52, 56. Initially, it is assumed that the resistance of resistor 52 is an arbitrary value such as one ohm and that the resistance of resistor 56 is the remainder of the span resistance. With these values, the values are calculated for a series/parallel resistance network that provides temperature compensation at each of the two extreme temperatures H, C. More specifically, the ratios are calculated of the resistance GA + SRA in the first arm of the bridge to the resistance GB + SRB in the second arm at the hot and cold temperatures and zero stress, where GA and GB are the resistances of variable resistances 54, 58 at the temperatures and stresses indicated, SRA is the resistance assigned to resistor 52 and SRB is the resistance assigned to resistor 56. As is well known in the art, for two point temperature compensation, the ratios should be the same.

If, however, GA(H)+SRA GA(C)tSRA GB(H)+SRB GB(C)+SRB, resistance must be added to GA(C)+SRA to compensate.

If GA(H)+SRA GA(C)+SRA GB(H)+SRB GB(C)+SRB, resistance must be added to GB(C) + SRB to compensate.

For purposes of discussion, let us assume that resistance must be added to GA(C) + SRA. Using well known two point temperature compensation procedures, the resistance RY of parallel resistor 62 connected across variable resistance 54 can be calculated to be:

and the resistance RX of an addition series resistance in series with resistor 52 can be calculated to be: RX = A - A.RY A+RY where

D = GA(C,Z) where GA and GB are the resistances of variable resistances 54, 58, respectively, at the temperatures and stresses indicated and SRB is the resistance of resistor 56. As will be apparent, these values of RX and RY cancel each other out at the uppermost temperature H so that the resistance in the arm of the bridge where the strain gauges are located remains unchanged. At the lowermost temperature C, these values provide an additional resistance sufficient to equalize the ratio of the resistance in the first arm and the resistance in the second arm for the two temperatures H and C. If, in stead, the resistance had to be added to GB(C) + SRB, parallel resistor 62 would have been connected across variable resistance 58; and the equations would be similar with the substitution of GA, GB and SRA for GB, GA and SRB of the above equations.

Next, the voltage V at the node C between the two strain gauge arms of the compensated bridge circuit is calculated at the intermediate temperature and the uppermost temperature using the relation: V = VS . (ESRA + EGA) (ESRA + EGA + GB SRB) where VS - the voltage across the bridge circuit ESRA SRA + RX EGA (GA) . (RY)/(GA+RY) and GA and GB are the resistance of variable resistances 54, 58 at zero stress for each of the temperatures for which the calculation is made.

The difference DV between the voltage V at the intermediate temperature and the voltage V at the uppermost temperature is then determined. During the first run of the iterative process, this difference is retained along with the values SRA, SRB RX and RY.

The value of SRA is then tested to determine if SRA=SSR-1.

During the first run of the process this test will not be met. Accordingly, the value of SRA is incremented by one ohm and that of SRB is decremented by one ohm and the series/parallel resistances required for temperature compensation are calculated for the extreme temperatures using the foregoing equations for RX and RY, the measured values of the gauge resistances GA, GB and the new value of SRB or SRA as circumstances require. From these values the voltage V is calculated at the intermediate temperature and the uppermost temperature using the relation set forth above and the difference DV between these voltages is taken. This difference is then compared with the difference previously calculated and whichever value is closer to zero is retained along with the circuit parameters SRA, SRB, RX and RY needed to produce said difference.

The foregoing process is repeated for each value of SRA less than SSR. Once these calculations are completed, the difference DV which is retained is the closest value to zero for each of the values of SRA in one ohm increments between zero and SSR. Hence, the circuit parameters associated with that value DV produce the best three-point temperature compensation. Using these values, the resistance values of resistors 52 ,56, 62 of the bridge network are then set.

As will be apparent, the above-identified calculations can be made by hand or machine. If desired, the calculation process can readily be implemented in any number of small computers commercially available. The creation of a suitable program for making such calculations will be evident to any programmer in light of the discussion above.

The foregoing process provides span compensation at only two temperatures.

Typically, between these two temperatures there is a positive span error which increases gradually to a point about midway between the two temperatures and decreases thereafter. Below the lower temperature there typically is a negative span error which increases in magnitude as the temperature decreases; and above the upper temperature there likewise is a negative span error which again increaeses in magnitude as temperature increases. A typical plot of span error that is observed when there is span compensation at -70C (200F) and 71"C (1600F) is shown in Fig. 2. For this plot, maximum positive span error occurs at approximately 320C (90OF).

These span errors can be minimized by using span compensation network 30 to vary the voltage applied to bridge circuit 40. The values of resistors 32, 36 and thermistors 34 38 are chosen so that the two resistors essentially determine the voltage applied to the bridge circuit at that temperature at which the span error is a maximum. The values of resistors 32, 36 are selected so that there is substantially no span error at this intermediate temperature. Since thermistor 38 has a negative temperature coefficient of resistance, the resistance of the series combination of resistor 36 and thermistor 38 increases with decreasing temperature, thereby increasing the voltage applied to bridge circuit 40. The resistance of thermistor 38 and its temperature coefficient of resistance are selected so that this increase in voltage tends to offset the changes in span voltage at temperatures less than the intermediate temperature. Since thermistor 34 has a negative temperature coefficient of resistance, the resistance of the parallel combination of resistor 32 and thermistor 34 decreases with increasing temperature thereby causing the voltage applied to bridge circuit 40 to increase. The resistance of thermistor 34 and its temperature coefficient of resistance are selected so that this increase in voltage tends to offset the changes in span voltage at temperatures greater than the intermediate temperature.

Since span errors differ between individual strain gauges, the selection of appropriate values of resistors 32, 36 must be done empirically. Typical values of these resistors are set forth in Table I for the indicated ranges in span error.

TABLE I Span Error Resistor 32 Resistor 36 1.5 to 2.25% 12.1 K ohms 165 K ohms 2.25 to 2.75 14.7 K ohms 145 K ohms 2.75 to 3.50 14.3 K ohms 137 K ohms 3.50 to 4.50 15.0 K ohms 118 K ohms For the span compensation networks whose parameters are set forth in Table I, the voltage produced by voltage regulator 25 is 6.9 volts; and thermistor 34 is a Model FP52Jl and thermistor 38 is a model KP41J2 manufactured by Fenwall Electronics of Framingham, Mass, U.S.A. Thermistor 34 has a resistance of 200 K ohms at 250C and a negative temperature coefficient of resistance of 4.9%/C". At 700C its resistance is 28 K ohms. Thermistor 38 has a resistance of 10 K ohms at 25"C and the same temperature coefficient of resistance.

In bridge circuit 40 illustrative resistance values in the reference arm of the bridge are 2 K ohms for resistor 42, 16.8 K ohms for resistor 44 and 23 K ohms for resistor 46. As indicated above, the values in the sensing arm of the bridge depend on the properties of the variable resistances 54, 58. For the case of semiconductor strain gauges, the resistance of each strain gauge is typically about 1 K ohms. The total resistance needed for span compensation at the two extreme temperature values is typically on the order of 700 ohms. The additional series resistance which is required for temperature compensation is about 10 ohms and the parallel resistor 62 is typically in excess of 100 K ohms. For convenience, the additional series resistance can be incorporated into one of resistors 52, 56; and this additional series resistance will accordingly be described as a portion of one of these resistors in the claims below.

While the incorporation of this additional series resistance in the variable arm of the bridge does affect the series resistance in that arm between nodes A and B, the effect is small compared with the total resistance of resistors 52, 56. As a result, there is substantially no effect (usually less than 0.1% change) on span compensation. For the illustrative circuit values enumerated above, typical values of scaling resistors 92, 94, 96 are approximately 20, 40, and 80 ohms; and a typical value of feedback resistor 85 is 73.3 K ohms.

Constant current sources 20, 80, amplifier 70 and transistor 75 can be implemented using many different alternatives. Amplifier 70 preferably is implemented in the form of a high gain differential amplifier followed by an amplifier stage which drives transistor 75.

The reference voltage at node D is that voltage which will produce a minimum current output signal w 105 and an output resistor 107. Circuit 210 is connected at terminals E and F to a two wire line 120 which interconnects a power supply 130 and a resistive load 135.

For the most part, circuit 210 comprises the same elements found in circuit 10 and such elements have been identified by the same numbers and will not be discussed further.

Function generator 100 is connected between output node F and feedback resistor 105 which is connected to the non-inverting input terminal of amplifier 70. Switch 102 permits feedback resistor 105 to be connected to the output of generator 100 or to node B of the bridge circuit. When the function generator is connected to the feedback resistor the output of amplifier 70 is proportional to that function of the input signal to the amplifier which is the inverse of the function produced by the function generator. Preferably, function generator 100 is a squaring circuit so that the output signal from amplifier 70 is proportional to the square root of the signal at its input.

This is useful in calculating values such as fluid flow which is proportional to the square root of a pressure differential which could be measured by strain gauges.

Constant current source 80 of circuit 210 is used to calibrate squaring circuit 100. For the circuit shown in Fig. 3, function generator 100 has a non-zero output at the output signal level on line 120 which corresponds to the zero level in the variable sensed by the bridge circuit. This non-zero output from generator 100 is offset by using constant current source 80 to modify the reference voltage at node D in the same fashion as source 80 is used to suppress the zero level in the circuit of Fig. 1.

Fig. 4 illustrates the mechanical construction of a differential pressure sensor comprising a pressure-differential-to-displacement converter 310, a cantilever beam strain gauge arrangement 311, and resilient coupling means such as spring 312. The transducer and the cantilever beam are provided with separate overrange protective means as will be described in greater detail hereinbelow.

Converter 310 comprises bellows 313 made up of flexible, corrugated metal diaphragms 3 14A and 3 14B capable of nesting against one another when collapsed, a rigid rod member 315, and a piston member 316. The exterior of bellows 313 is coupled to one source of pressure to be differentially measured (preferably the low pressure source) via a chamber defined by walls 317 filled with incompressible di electric fluid 318, and an isolation membrane 319. The interior of the bellows is coupled to the other source of pressure (preferably the high pressure source) via a chamber 320 also filled with fluid 321, and an isolation membrane 322.

Metal diaphragms 3 14A and 3 14B are made of a material having a substantially constant temperature coefficient of elasticity such as the alloy Ni-Span-C (Registered Trade Mark) marketed by the International Nickel Company. Typical thicknesses of this material can range from about 0.006 inch to 0.0010 inch, depending on the range of pressure differential to be measured. The dielectric fluid can be a silicone dielectric fluid such as Dow-Corning 200 Dielectric Fluid (Registered Trade Mark), and both the chamber walls and isolating diaphragms can be fabricated from type 316 stain less steel, the diaphragm having a typical thickness on the order of 0.003 inch. Spring 312 is also preferably made of the aforesaid Ni-Span-C alloy.

In operation of the converter, a higher differential pressure in chamber 320 expands the bellows 313 and displaces rod 315 and piston 316 toward chamber 317 in a manner which is substantially linear with increasing pressure. The converter is provided with overrange protection in the form of an O-ring seal 323 on piston 316 which seals against shoulders 324 in chamber 320 when the differential pressure in chamber 320 exceeds a predetermined level. The converter is protected against over range differential pressures in chamber 317 by the fact that diaphragms 3 14A and 314B will nest against one another upon collapse and prevent further displacement of rod 315.

Cantilever beam strain gauge arrangement 311 preferably comprises a resilient cantilever beam 325 and a pair of semiconductor strain gauges 326 disposed on either side of the beam in the direction of displacement. In a preferred specific embodiment, the strain gauge arrangement is a commercially available unit marketed by Kistler Morse Co. under the trade designation, Deflector Sensor Cartridge 4ftDF6BB4- 1 lOAB. This unit is modified to include a dielectric projection 327 for mechanically coupling to spring 312.

In operation, within the range of measurable pressure differentials, displacement of piston 316 produces, through coupling spring 312, a corresponding displacement of the beam, with the consequence that one of gauges 326 experiences a compressive loading and the other a tensile loading. The gauges are connected by wires 329 to signal processing circuitry 330 such as that shown in Figs. 1 or 3.

The cantilever beam is provided with separate overrange protection preferably in the form of stops 328A and 328B. In practice, these stops constrain the displacement of the beam to a narrower range than the permissible range of displacement for the rod and piston. The resilient coupling spring absorbs the displacement differential.

The invention may be more clearly understood by reference to the following description of the main parameters of a specific device for sensing pressure differen tials in the range between 0 and 120 inches of water. In this device, the bellows is made of approximately 0.006 inch thick Ni-Span-C alloy and has a pressure effective area of about 0.45 square inch. It has a spring rate of approximately 140 Ibs./inch.

The isolating diaphragms have an area of about 3 square inches and a spring rate of aboout 40 lbs./inch. The coupling spring has a spring rate of about 180 Ibs./inch. In operation, the bellows has a full scale deflection displacement of about 0.011 inch and the cantilever beam displaces about 0.009 inch. The overrange protection stops are designed to constrain the converter displacement within about 0.030 inches and the cantilever beam within the narrower range of about 0.011 inches.

The result is a device which can provide high sensitivity readings in a range of low level pressure differentials and still provide overrange protection against high level pressure differentials greatly in excess of that range. The aforesaid exemplary device for sensing pressure differentials in the range from 0 to 120 inches of water provides overrange protection against pressures up to 6000 p.s.i.

As will be apparent, the present invention is not limited to the specific embodiments described above and many of the individual components of the invention can be implemented independently of others with which they were described. The use of a bridge circuit with the iterative process for achieving three-point temperature compensation is only illustrative because the function of resistors 42, 44, 46 is merely to provide a reference voltage. Thus, the invention may be practiced in any voltage divider circuit in which there are two elements having an appreciable temperature coefficient of resistance and an output which varies with at least one environmental variable such as stress. Even more generally, the invention may be practiced using any divider circuit made of elements having a resistive, capacitive or inductive effect for which at least two of such elements have an output which is a function of two variables. In such circumstances, the span compensation process described above for the calculation of the resistance SSR is used to calculate the value of a fixed electrical element such that at a first value of one variable the difference in the output between two values of the second variable is the same as the difference in the output for the same two values of the second variable at a second value of the first variable. The iterative process described above for temperature (or constant value) compensation is then used to minimize the differences in the output of the circuit at three values of the first variable and a constant value of the second variable. From the foregoing des cripu'on, the implementation of the invention in these circumstances will be evident to those skilled in the art. This compensation method may also be practiced using a constant current source in place of voltage regulator 25 in which case the span compensation resistance is connected in parallel with bridge circuit 40. The process for obtaining three point temperature compensation is similar. Initially, an arbitrary resistance is connected in series with the gauges; and the necessary span compensation resistance is calculated. The iterative process is then used to distribute the compensation resistance between two resistors and to calculate the resistance values of a series/parallel network as described previously.

WHAT WE CLAIM IS: 1. A method for compensating for environmental' effects on first and second electrical devices in a divider circuit which devices have a resistance, capacitance or inductance which is a function of two variables (X, Y), said method comprising the steps of: a) measuring the resistance, capacitance or inductance across each of said devices at (C, Z), (C, S), (H, Z) and (H, S), where C and H are first and second values of the first variable and Z and S are first and second values of the second variable, when said devices are connected in said divider circuit with an additional resistance, capacitance or inductance; b) calculating a resistance, capacitance or inductance which provides for span compensation at C and H when said span compensation resistance, capacitance or

**WARNING** end of DESC field may overlap start of CLMS **.

Claims

**WARNING** start of CLMS field may overlap end of DESC **. loading and the other a tensile loading. The gauges are connected by wires 329 to signal processing circuitry 330 such as that shown in Figs. 1 or 3. The cantilever beam is provided with separate overrange protection preferably in the form of stops 328A and 328B. In practice, these stops constrain the displacement of the beam to a narrower range than the permissible range of displacement for the rod and piston. The resilient coupling spring absorbs the displacement differential. The invention may be more clearly understood by reference to the following description of the main parameters of a specific device for sensing pressure differen tials in the range between 0 and 120 inches of water. In this device, the bellows is made of approximately 0.006 inch thick Ni-Span-C alloy and has a pressure effective area of about 0.45 square inch. It has a spring rate of approximately 140 Ibs./inch. The isolating diaphragms have an area of about 3 square inches and a spring rate of aboout 40 lbs./inch. The coupling spring has a spring rate of about 180 Ibs./inch. In operation, the bellows has a full scale deflection displacement of about 0.011 inch and the cantilever beam displaces about 0.009 inch. The overrange protection stops are designed to constrain the converter displacement within about 0.030 inches and the cantilever beam within the narrower range of about 0.011 inches. The result is a device which can provide high sensitivity readings in a range of low level pressure differentials and still provide overrange protection against high level pressure differentials greatly in excess of that range. The aforesaid exemplary device for sensing pressure differentials in the range from 0 to 120 inches of water provides overrange protection against pressures up to 6000 p.s.i. As will be apparent, the present invention is not limited to the specific embodiments described above and many of the individual components of the invention can be implemented independently of others with which they were described. The use of a bridge circuit with the iterative process for achieving three-point temperature compensation is only illustrative because the function of resistors 42, 44, 46 is merely to provide a reference voltage. Thus, the invention may be practiced in any voltage divider circuit in which there are two elements having an appreciable temperature coefficient of resistance and an output which varies with at least one environmental variable such as stress. Even more generally, the invention may be practiced using any divider circuit made of elements having a resistive, capacitive or inductive effect for which at least two of such elements have an output which is a function of two variables. In such circumstances, the span compensation process described above for the calculation of the resistance SSR is used to calculate the value of a fixed electrical element such that at a first value of one variable the difference in the output between two values of the second variable is the same as the difference in the output for the same two values of the second variable at a second value of the first variable. The iterative process described above for temperature (or constant value) compensation is then used to minimize the differences in the output of the circuit at three values of the first variable and a constant value of the second variable. From the foregoing des cripu'on, the implementation of the invention in these circumstances will be evident to those skilled in the art. This compensation method may also be practiced using a constant current source in place of voltage regulator 25 in which case the span compensation resistance is connected in parallel with bridge circuit 40. The process for obtaining three point temperature compensation is similar. Initially, an arbitrary resistance is connected in series with the gauges; and the necessary span compensation resistance is calculated. The iterative process is then used to distribute the compensation resistance between two resistors and to calculate the resistance values of a series/parallel network as described previously. WHAT WE CLAIM IS:

1. A method for compensating for environmental' effects on first and second electrical devices in a divider circuit which devices have a resistance, capacitance or inductance which is a function of two variables (X, Y), said method comprising the steps of: a) measuring the resistance, capacitance or inductance across each of said devices at (C, Z), (C, S), (H, Z) and (H, S), where C and H are first and second values of the first variable and Z and S are first and second values of the second variable, when said devices are connected in said divider circuit with an additional resistance, capacitance or inductance; b) calculating a resistance, capacitance or inductance which provides for span compensation at C and H when said span compensation resistance, capacitance or

inductance is connected in said circuit with said first and second electrical devices; c) calculating the resistances, capacitances, or inductances for a series/parallel network for one of said electrical devices which provides for constant value compensation at C and H, assuming that a small portion of the span compensation resistance, capacitance or inductance is connected with the first electrical device in one portion of the divider circuit and the remaining portion is connected with the second electrical device in a second portion of the divider circuit; d) calculating the outputs from said divider circuit at (H, Z) and (M, Z), where M is a third value of the first variable between said first and second values, using the distribution of span compensation resistance, capacitance, or inductance assumed in step (c) and the series/parallel resistances, capacitances or inductances calculated in step (c); e) repeating steps (c) and (d) for at least one other distribution of the span compensation resistance, capacitance or inductance in said first and second portions of the divider circuit; f) comparing the difference in the outputs of the divider circuit at (H, Z) and (M, Z) during one execution of step (d) with the difference in the outputs of said divider circuit at (H, Z) and (M, Z) during a second execution of step (d); and g) using in the design of the divider circuit the distribution of span compensation resistance, capacitance, or inductance and the series/parallel resistances, capacitances, or inductances for which the difference in the outputs of the divider circuit at (H, Z) and (M, Z) is closer to zero.

2. The method of claim 1 wherein: a) during the first execution of step (c) the portion of the span compensation resistance, capacitance or inductance connected with the first electrical device is relatively small; and b) step (e) comprises the steps of: (1) incrementing the resistance, capacitance or inductance connected with the first electrical device by a predetermined amount and decrementing that con nected with the second electrical device an equal amount; (2) repeating steps (c) and (d) using the new values of said resistances, capacitances or inductances determined by step 1; and (3) repeating steps 1 and 2 until the resistance, capacitance or inductance connected with said first electrical device approaches the total span compensation resistance, capacitance or inductance.

3. The method of claim 1 or claim 2 wherein said first and second electrical devices are strain gauges having a resistance which varies with temperature and stress, said first variable is temperature and said second variable is stress.

4. The method of claim 1 further comprising the step of modifying the voltage applied to said first and second electrical devices in order to reduce span error between said first and second values of the first variable.

5. The method of claim 1 further comprising the steps of: measuring the span error between said first and second values of the first variable when said span compensation resistance, capacitance or inductance is conected in said circuit with said first and second electrical devices; selecting for use in the circuit first and second resistive elements having a positive coefficient of resistance with respect to said first variable, the resistance of said first and second resistive elements being such that when connected to the divider circuit they reduce the voltage supplied to the circuit so as to minimize span error at an intermediate value between said first and second values of the first variable; and selecting for use in the circuit third and fourth resistive elements having a negative coefficient of resistance with respect to said first variable, the coefficients of resistance of said third and fourth resistive elements being such that when a series combination of the first and third resistive elements is connected in parallel with the divider circuit the voltage supplied to the divider circuit increases as the value of the first variable decreases below said intermediate value and when a parallel combination of the second and fourth resistive elements is connected in series between a voltage supply and the divider circuit the voltage supplied to the divider circuit increases as the value of the first variable increases above the intermediate value.

6. The method of claim 4 or claim 5 wherein said first and second electrical devices are strain gauges having a resistance which varies with temperature and stress, said first variable is temperature and said second variable is stress.

7. A method for compensating for temperature effects on first and second strain gauges in a voltage divider circuit, said method comprising the steps of: a) measuring the voltage across each of said strain gauges and a first series resistance at first and second stresses at first and second temperatures; b) calculating a resistance which provides for span compensation between said first and second stresses at said first and second temperatures when said span compensation resistance is connected in series with said strain gauges; c) calculating the resistances for a series/parallel network for one of said strain gauges which provides for temperature compensation at said first and second temperatures, assuming that a small portion of the span compensation resistance is in series with the first strain gauge in one portion of the divider circuit and the remaining portion of the strain compensation resistance is in series with the second strain gauge in a second portion of the divider circuit; d) calculating the voltage output at a node between said first and second portions of said divider circuit at said first stress at both said second temperature and a third temperature between said first and second temperatures using the distribution of span compensation resistance and the series/parallel resistances calculated in step (c); e) repeating steps (c) and (d) for at least one other distribution of the span compensation resistance in said first and second portions of the divider circuit; f) comparing the difference in the outputs of the divider circuit at said second and third temperatures during one execution of step (d) with the difference in the outputs of said divider circuit at said second and third temperatures during a second execution of step (d); and g) using in the design of the divider circuit the distribution of span compensation resistance and the series/parallel resistances for which the difference in the outputs of the divider circuit at said second and third temperatures is closer to zero.

8. The method of claim 7 wherein: a) during the first execution of step (c) the portion of the span compensation resistance connected in series with the first strain gauge is relatively small; and b) step (e) comprises the steps of: (1) incrementing the resistance in series with the first strain gauge by a predetermined amount and decrementing the resistance in series with the second stram gauge an equal amount; (2) repeating steps (c) and (d) using the new resistance values determined by step 1; and (3) repeating steps 1 and 2 until the resistance in series with said first strain gauge approaches the total span compensation resistance.

9. The method of claim 7 further comprising the step of modifying the voltage applied to said first and second strain gauges in order to reduce span error between said first and second temperatures.

10. The method of claim 7 further comprising the steps of: measuring the span error between said first and second temperatures when said span compensation resistance is connected in said circuit with said first and second strain gauges; selecting for use in the circuit first and second resistive elements having a positive temperature coefficient of resistance, the resistances of said first and second resistive elements being such that when connected to the divider circuit they reduce the voltage supplied to the circuit so as to minimize span error at an intermediate temperature between said first and second temperatures; and selecting for use in the circuit third and fourth resistive elements having a negative temperature coefficient of resistance, the coefficient of resistance of said third and fourth resistive elements being such that when a series combination of the first and third resistive elements is connected in parallel with the divider circuit the voltage supplied to the divider circuit increases as the temperature decreases below said intermediate temperature and when a parallel combination of the second and fourth resistive elements is connected in series between a voltage supply and the divider circuit the voltage supplied to the divider circuit increases as the temperature increases above the intermediate temperature.

11. A method for compensating for temperature effects on first and second strain gauges in a first half of a bridge circuit, said method comprising the steps of: a) measuring the voltage across each of said strain gauges and a first series resistance at first and second stresses at first and second temperatures; b) calculating a resistance which provides for span compensation between said first and second stresses at said first and second temperatures when said span compensation resistance is connected in series with said strain gauges in said first half; c) calculating the resistances for a series/parallel network for one of said strain gauges which provides for temperature compensation at said first and second temperatures, assuming that a small portion of the span compensation resistance is in series with the first strain gauge in a first arm of said first half and the remaining portion of the span compensation resistance is in series with the second strain gauge in a second arm of said first half; d) calculating the voltage output at a node between said first and second arms of said first half at said first stress at both said second temperature and a third temperature between said first and second temperatures using the distribution of span compensation resistance and the series/parallel resistances calculated in step (c); e) repeating steps (c) and (d) for at least one other distribution of the span compensation resistance in said first and second arms of the first half; f) comparing the difference in the outputs at said node at said second and third temperatures during one execution of step (d) with the difference in the outputs at said node at said second and third temperatures during a second execution of step (d) and; g) using in the design of the first half of the bridge circuit the distribution of span compensation resistance and the series/parallel resistances for which the difference in the outputs at said node at said second and third temperatures is closer to zero.

12. The method of claim 11 wherein: a) during the first execution of step (c) the portion of the span compensation resistance connected in series with the first strain gauge in said first arm of said first half is relatively small; and b) step (e) comprises the steps of: (1) incrementing the resistance in series with the first strain gauge in said first arm by a predetermined amount and decrementing the resistance in series with the second strain gauge in said second arm an equal amount; (2) repeating steps (c) and (d) using the new resistance values determined by step 1; and (3) repeating steps 1 and 2 until the resistance in series with said first strain gauge in said first arm approaches the total span compensation resistance.

13. The method of claim 11 further comprising the step of modifying the voltage applied to said bridge circuit in order to reduce span error between said first and second temperatures.

14. The method of claim 11 further comprising the steps of: measuring the span error between said first and second temperatures when said span compensation resistance is connected in said first half with said first and second strain gauges; selecting for use with the bridge circuit first and second resistive elements havmg a positive temperature coefficient of resistance, the resistances of said first and second resistive elements being such that when connected to the bridge circuit they reduce the voltage supplied to the circuit so as to minimize span error at an intermediate temperature between said first and second temperatures; and selecting for use in the circuit third and fourth resistive elements having a negative temperature coefficient of resistance, the coefficient of resistance of said third and fourth resistive elements being such that when a series combination of the first and third resistive elements is connected in parallel with the bridge circuit the voltage supplied to the bridge circuit increases as the temperature decreases below said intermediate temperature and when a parallel combination of the second and fourth resistive elements is connected in series between a voltage supply and the bridge circuit the voltage supplied to the bridge circuit increases as the temperature increases above the intermediate temperature.