Process variables include pressure, temperature, flow, level, turbidity, density, concentration, chemical composition and other properties.

In typical processing plants, a communication bus, such as a 4-20 mA current loop is used to power the process variable transmitter. Examples of such current loops include a FOUNDATIONTM Fieldbus connection or a connection in accordance with the Highway Addressable Remote Transducer- (HART) communication protocol. In transmitters powered by a two-wire loop, power must be kept low to comply with intrinsic safety requirements.

A process temperature transmitter provides an output related to a sensed process substance temperature. The temperature transmitter output can be communicated over the loop to a control room, or the output can be communicated to another process device such that the process can be monitored and controlled.

In order to monitor a process temperature, the transmitter includes a sensor, such as a resistance temperature device (RTD) or a thermocouple.

An RTD changes resistance in response to a change in temperature. By measuring the resistance of the RTD, temperature can be calculated. Such

resistance measurement is generally accomplished by passing a known current through the RTD, and measuring the associated voltage developed across the RTD.

A thermocouple provides a voltage in response to a temperature change. The Seebeck Effect provides that dissimilar metal junctions create voltage due to the union of the dissimilar metals in a temperature gradient condition. Thus, the voltage measured across the thermocouple will relate to the temperature of the thermocouple.

As temperature sensors age, their accuracy tends to degrade until the sensor ultimately fails.

However, small degradations in the output from the sensor are difficult to detect and to separate from actual changes'in the measured temperature. In the past, temperature transmitters have used two temperature sensors to detect sensor degradation. If the output from the two sensors is not in agreement, the temperature transmitter can provide an error output. However, this technique is not able to detect a degradation in the sensor output if both of the two temperature sensors degrade at the same rate and in the same manner.

One technique which has been used in situations in which power is not a constraint is described in U. S. Patent Nos. 5,713,668 and 5,887,978, issued February 3,1998 and March 30,1999, respectively, to Lunghofer et al. and entitled"SELF- VERIFYING TEMPERATURE SENSOR". These references describe a temperature sensor having multiple outputs.

The multiple outputs all vary as functions of temperature. However, the relationships between the various outputs and temperature are not the same.

Further, the various elements in the temperature sensor change over time at differing rates, and in differing manners and react differently to various types of failures. A computer monitors the output from the sensor using a multiplexer. The computer places data points from the sensor into a matrix. By monitoring the various entries in the matrix and detecting changes in the various element or elements of the matrix relative to other elements, the computer provides a "confidence level"output for the measured temperature.

If the confidence level exceeds a threshold, an alarm can be provided.

However, the art of low power process variable transmitters has an ongoing need for improved temperature sensors such as those which provide improved accuracy or a diagnostic output indicative of the condition of the temperature sensor.

SUMMARY OF THE INVENTION A two-wire temperature transmitter is coupleable to a two-wire process control loop for measuring a process temperature. The transmitter includes an analog to digital converter configured to provide digital output in response to an analog input.

A two-wire loop communicator is configured to couple to the process control loop and send information on the loop. A microprocessor is coupled to the digital output and configured to send temperature related information on the process control loop with the two-wire loop communicator. A power supply is configured to completely power the two-wire temperature transmitter with power from the two-wire process control loop. A temperature sensor comprises at least two temperature sensitive elements having element outputs which degrade

in accordance with different degradation characteristics. The element outputs are provided to the analog to digital converter, such that the microprocessor calculates temperature related information as a function of at least one element output from a first temperature sensitive element and at least as a function of one degradation characteristic of a second temperature sensitive element.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram of the environment of a process temperature transmitter.

FIG. 2 is a diagrammatic view of the process temperature transmitter of FIG. 1.

FIG. 3 is a system block diagram of a process temperature transmitter.

FIG. 4 is a diagram of a neural network implemented in the transmitter of FIG. 3.

FIG. 5 is a block diagram of a method of measuring process fluid temperature with a two-wire process temperature transmitter.

DETAILED DESCRIPTION OF THE ILLUSTRATIVE EMBODIMENTS FIGS. 1 and 2 illustrate the environment of a process temperature transmitter in accordance with embodiments of the invention. FIG. 1 shows process control system 10 including process temperature transmitter 12, two-wire process control loop 16 and monitor 14. As used herein, two-wire process control loop means a communication channel including two wires that power connected process devices and provide for communication between the connected devices.

FIG. 2 illustrates process control system 10 including process temperature transmitter 12

electrically coupled to monitor 14 (modeled as a voltage source and resistance) over two-wire process control loop 16. Transmitter 12 is mounted on and coupled to a process fluid container such as pipe 18.

Transmitter 12 monitors the temperature of process fluid in process pipe 18 and transmits temperature information to monitor 14 over loop 16.

FIG. 3 is a system block diagram of process temperature transmitter 12 in accordance with an embodiment of the invention. Process temperature transmitter 12 includes an analog to digital converter 20 configured to provide a digital output 22 in response to an analog input 24. A two-wire loop communicator 26 is configured to couple to two-wire process control loop 16 and to send information on loop 16 from a microprocessor 28. At least one power supply 30 is configured to couple to loop 16 to receive power solely from loop 16 and provide a power output (Pwr) to power circuitry in transmitter 12 with power received from loop 16. A temperature sensor 34 couples to analog to digital converter 20 through multiplexer 36 which provides the analog signal 24. Temperature sensor 34 includes temperature sensitive elements such as RTD 40 and thermocouples 42,44 and 46. Temperature sensor 34 operates in accordance with the techniques described in U. S. Patent No. 5,713,668. In addition to the transmitter shown in FIG. 3, the teachings of U. S.

Patent No. 5,828,567 to Eryurek et al., entitled "DIAGNOSTICS FOR RESISTANCE BASED TRANSMITTER"can be used with sensor 34.

Microprocessor 28 can be a low power microprocessor such as a Motorola 6805HC11 available from Motorola Inc. In many microprocessor systems, a

memory 50 is included in the microprocessor which operates at a rate determined by clock 52. Memory 50 includes both programming instructions for microprocessor 28 as well as temporary storage for measurement values obtained from temperature sensor 34, for example. The frequency of clock 52 can be reduced to further reduce power consumption of microprocessor 28.

Loop communicator 26 communicates on two-wire process control loop 16 in accordance with known protocols and techniques. For example, communicator 26 can adjust the loop current I in accordance with a process variable received from microprocessor 28 such that current I is related to the process variable. For example, a 4 mA current can represent a lower value of a process variable and 20 mA current can represent an upper value for the process variable. In another embodiment, communicator 26 impresses a digital signal onto loop current I and transmits information in a digital format. Further, such digital information can be received from two-wire process control loop 16 by communicator 26 and provided to microprocessor 28 to control operation of temperature transmitter 12.

Analog to digital converter 20 operates under low power conditions. One example of analog to digital converter 20 is a sigma-delta converter. Examples of analog to digital converters used in process variable transmitters are described in U. S. Patent No.

5,803,091, entitled"CHARGE BALANCE FEEDBACK MEASUREMENT CIRCUIT"issued January 21,1992 and U. S.

Patent No. 4,878,012, entitled"CHARGE BALANCE FEEDBACK TRANSMITTER, issued October 31,1989, which are commonly assigned with the present application.

Sensor 34 includes at least two temperature sensitive elements each having element outputs that degrade in accordance with different degradation characteristics. As illustrated, sensor 34 includes conductors 60,62,64,66 and 68. In one embodiment, at least some of conductors 60-68 are dissimilar conductors which have temperature related characteristics which change in a dissimilar manner.

For example, conductors 60 and 62 can be of dissimilar metals such that they form a thermocouple at junction 42. Using multiplexer 36, various voltage and resistance measurements of sensor 34 can be made by microprocessor 28. Further, a four point Kelvin connection to RTD 40 through conductors 60,62,66 and 68 is used to obtain an accurate measurement of the resistance of RTD 40. In such a measurement, current is injected using, for example, conductors 60 and 68 into RTD 40 and conductors 62 and 66 are used to make a voltage measurement. Conductor 64 can also be used to make a voltage measurement at some midpoint in RTD 40.

Voltage measurements can also be made between any pair of conductors such as conductors 60/62 60/64,62/66, etc. Further still, various voltage or resistance measurements can be combined to obtain additional data for use by microprocessor 28.

Microprocessor 28 stores the data points in memory 50 and operates on the data in accordance with the techniques described in U. S. Patent Nos. 5,713,668 and 5,887,978. This is used to generate a process variable output related to temperature which is provided to loop communicator 26. For example, one of the elements in sensor 34 such as RTD 40 can be the primary element while the remaining temperature related

data points provide secondary data points.

Microprocessor 28 can provide the process variable output along with an indication of the confidence level, probability of accuracy or a temperature range, i. e., plus or minus a certain temperature amount or percentage based upon the secondary data points. For example, the process variable output can be output as an analog signal (i. e., between 4 and 20 mA) while the indication of confidence can be provided as a digital signal. The confidence indication can be generated by empirical measurements in which all of the data outputs are observed over a wide range of temperatures and as the elements begin to degrade with time or other failures. Microprocessor 28 can compare actual measurements with the characteristics stored in memory 50 which have been generated using the empirical tests.

Using this technique, anomalous readings from one or more of the data measurements can be detected.

Depending on the severity of the degradation, microprocessor 28 can correct the temperature output to compensate for the degraded element. For a severely degraded element, microprocessor 28 can indicate that the sensor 34 is failing and that the temperature output is inaccurate.

Microprocessor 28 can also provide a process variable output as a function of the primary sensor element and one or more secondary sensor elements. For example, the primary sensor element can be an RTD indicating a temperature of for example 98oC while a secondary sensor element, for example a type J thermocouple, may indicate a temperature of 100°C, giving each sensor an equal numeric weight would

provide a process temperature output of 99oc. Because various types of sensors and sensor families exhibit different electrical characteristics in varying temperature ranges, microprocessor 28 can be programmed to vary sensor element weighting based upon the process variable itself. Thus, as the measured temperature begins to exceed a useful range of one type of sensor, the weighting for that sensor can be reduced or eliminated such that additional sensors with higher useful temperature ranges can be relied upon.

Moreover, because various types of sensors and sensor families have different time constants, it is contemplated that the weighting factors can be changed in response to a rate of change of the measured temperature. For example, an RTD generally has more thermal mass than a thermocouple due to the sheer mass of wound sensor wire and the fact that the sensor wire is generally wound around a ceramic bobbin which provides yet additional. thermal mass. However, the thermocouple junctions may have significantly less thermal mass than the RTD and thus track rapid temperature changes more effectively than the RTD.

Thus, as microprocessor 28 begins to detect a rapid temperature change. The sensor element weights can be adjusted such that the process variable output relies more heavily upon thermocouples.

In one embodiment, software in memory 50 is used to implement a neural network in microprocessor 28 such as neural network 100 illustrated in FIG. 4. FIG.

4 illustrates a multi-layer neural network. Neural network 100 can be trained using known training algorithms such as the back propagation network (BPN)

to develop the neural network modules. The network includes input nodes 102, hidden nodes 104 and output node 106. Various data measurements Dl-DN are provided as inputs to the input nodes 102 which act as an input buffer. The input nodes 102 modify the received data by various weights in accordance with a training algorithm and the outputs are provided to the hidden nodes 104. The hidden layer 104 is used to characterize and analyze the non-linear properties of the sensor 34. The last layer, the output layer 106 provides an output 108 which is an indication of the accuracy of the temperature measurement. Similarly, an additional output can be used to provide an indication of the sensed temperature.

The neural network 100 can be trained either through modeling or empirical techniques in which actual sensors are used to provide training inputs to the neural network 100. Additionally, a more probable estimate of the process temperature can be provided as the output based upon operation of the neural network upon the various sensor element signals.

Another technique for analyzing the data obtained from sensor 34 is through the use of a rule based system in which memory 50 contains rules, expected results and sensitivity parameters.

FIG. 5 is a block diagram of a method of measuring process temperature with a two-wire process temperature transmitter. The method begins at block 120 where a primary sensor element is measured using a two-wire temperature transmitter, such as transmitter 12. At block 122, one or more secondary sensor elements are measured using the two-wire temperature transmitter. It should be noted that block 122 need

not be performed after each and every primary sensor element measurement, but that block 122 can be performed periodically or in response to an external command. At block 124, the primary sensor element and secondary sensor element signals are provided to a transmitter microprocessor, such as microprocessor 28 (shown in FIG. 3). At block 126, microprocessor 28 calculates a process variable output based upon one or more of the primary sensor element signal and secondary sensor element signals. At block 128, the microprocessor calculates a confidence of the process variable output based upon the primary element sensor signal and one or more of the secondary sensor element signals. Finally, at block 130, the process temperature output and an indication of output validation or confidence in the process temperature output are provided by the two-wire process temperature transmitter. Such indication can be in the form of a numeric value representing a tolerance, or probability of accuracy or a temperature range, i. e., plus or minus a certain temperature amount or percentage based upon one or more secondary sensor signals; or the indication can also be an alarm or other user notification representative of the acceptability of the process variable output. Additionally, the indication of confidence can be in the form of an estimation of time remaining until the two-wire process transmitter is unable to suitably relate the process variable output to the process temperature. Further, providing a validated process temperature allows validation and diagnostics of other process variables that can be affected by the process temperature.

Another analysis technique is fuzzy logic.

For example, fuzzy logic algorithms can be employed on the data measurements Dl-DN prior to their input into neural network 100 of FIG. 4. Additionally, neural network 100 can implement a fuzzy-neural algorithm in which the various neurons of the network implement fuzzy algorithms. The various analysis techniques can be used alone or in their combinations. Additionally, other analysis techniques are considered to be within the scope of the present invention so long as they reach the requirement that the system is capable of operating completely from power received from a two- wire process control loop.

Although only a single analog to digital converter 20 is shown, such an analog to digital converter can comprise multiple analog to digital converters which can thereby either reduce or eliminate the amount of multiplexing performed when coupling the sensor 34 to the analog to digital converters.

Although the invention has been described with reference to preferred embodiments, workers skilled in the art will recognize that changes can be made in form and detail without departing from the spirit and scope of the invention. For example, various function blocks of the invention have been described in terms of circuitry, however, many function blocks may be implemented in other forms such as digital and analog circuits, software and their hybrids. When implemented in software, a microprocessor performs the functions and the signals comprise digital values on which the software operates.

A general purpose processor programmed with instructions that cause the processor to perform the desired process elements, application specific hardware

components that contain circuits wired to perform the desired elements and any combination of programming a general purpose processor and hardware components can be used. Deterministic or fuzzy logic techniques can be used as needed to make decisions in the circuitry or software. Because of the nature of complex digital circuitry, circuit elements may not be partitioned into separate blocks as shown, but components used for various functional blocks can be intermingled and shared. Likewise with software, some instructions can be shared as part of several functions and be intermingled with unrelated instructions within the scope of the invention.