CORRECTION

BACKGROUND OF THE INVENTION

The present invention relates to industrial process control and monitoring systems. More specifically the present invention relates to field devices in which at least two process variables are sampled for use in monitoring or controlling an industrial process.

Industrial processes are used in the manufacture and refinement of various goods and commodities such as oil, food stuff, pharmaceuticals, paper pulp, etc. In such systems, typically a process variable of a process fluid is measured by a field device. Examples of process variables include pressure, temperature, differential pressure, level, flow rate and others. Based upon this measured process variable, if the industrial process is controlled using a feedback system, the process variable can be used to adjust or otherwise control operation of the industrial process.

Some types of process variables are measured or calculated based upon measurement of two, or more, other process variables. For example, a differential pressure can be measured by measuring two separate process fluid pressures and subtracting the two measurements. The differential pressure can be used in determining flow rate or level of process fluid in a container. However, when using two separate process variables to determine a third process variable, errors due to time skew error can be introduced into the determination.

SUMMARY

An industrial process device for monitoring or controlling an industrial process includes a first input configured to receive a first plurality of samples related to a first process variable and a second input configured to receive a second plurality of samples related to a second process variable. Compensation circuitry is configured to compensate for a time difference between the first plurality of samples and the second plurality of samples and provide a compensated output related to at least one of the first and second process variables.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a simplified block diagram showing industrial process control system in which two separate process variables are sensed.

Figure 2 is a simplified block diagram showing a more detailed view of a field device of Figure 1.

Figure 3A and Figure 3B are graphs of a common mode signal versus time and its effect on device output for a device which senses a process variable.

Figure 4 is a block diagram similar to Figure 2 in which low pass filters are introduced to provide compensation for time skew between two devices.

Figure 5 is a simplified block diagram showing compensation circuitry illustrated within measurement circuitry.

Figure 6 is a graph showing a linear extrapolation technique for a ramping common mode signal.

Figure 7 is a graph showing non-linear extrapolation.

Figure 8 is a graph showing a interpolation technique for using a quadratic polynomial.

Figure 9A is a simplified block diagram showing example steps in accordance with one configuration of the present invention and Figure 9B is a graph of device output using time.

DETAILED DESCRIPTION

As discussed in the Background section, in some instances the measurement of a process variable requires the measurement of two or more different process variables. One common example is the measurement of a differential pressure which can, in some instances, be based upon the measurement of two separate pressures, absolute or gauge, whose difference is then determined.

In some instances, this measurement may be accomplished electronically by sampling data from two separate process variable sensors. The sample data is then used to generate an output based upon a mathematical relationship between the two sampled signals. One example configuration is illustrated in U.S. Patent No. 5,870,695, entitled DIFFERENTIAL PRESSURE MEASUREMENT ARRANGEMENT UTILIZING REMOTE SENSOR UNITS, issued February 9, 1999 to Gregory C. Brown and David A. Broden.

Figure 1 shows an industrial process controller monitoring system 100 in which a process variable transmitter 102 is shown as being configured to measure the level of a process fluid 104 within a vessel 106 such as a tank. Transmitter 102 includes a main housing 110 and a remote sensor 112. Using known techniques, the main housing 110 can be configured to measure a pressure within the vessel 106 and receive an electronic signal over connection 114 from remote sensor 112 which is representative of a second pressure within the vessel. These two pressures can be used to determine the level of the process fluid 104 within the vessel 106. Information related to this measurement can be transmitted to a remote location such as a process control room 122 over a process control loop 120. In Figure 1, the process control loop 120 is illustrated as a two wire process control loop which couples to a process control room 122 or other location while the process field device 102 can be located at a remote location. In such a configuration, process control loop 120 can carry information, such as data transmitted from field device 102, as well as be used as the sole source of power for the field device 102 using power from control room 122, illustrated as battery 126 which is series in a loop resistance 124. However, any appropriate process control loop can be used including wireless process control loops in which data is transmitted using, for example, RF signaling techniques.

Although the field device 102 is illustrated as having two main components, main body 110 and remote sensor 112, other configurations can also be used. For example two remote sensors, such as remote sensor 112 can couple to the main body. In another example configuration, more than two process variables are received from remote sensors.

Figure 2 is a simplified block diagram showing a field device 130 including a first sensor 132 configured to sense a first process variable PVl and a second sensor 134 configured to sense a second process variable sensor PV2. More specifically, process variable sensors 140A and 140B are provided in sensors 132 and 134, respectively, which are configured to sense a process variable. The sensed process variables are then sampled and digitized, using, for example, an analog to digital converter 142A in sensor 132 and 142B in sensor 134. The digitized process variables PVl and PV2 are then provided to the measurement circuitry 136. The measurement circuitry 136 provides an output which is a function of the two process variables. For example, if a differential pressure is being measured, the output may be PV1-PV2.

Typically in such a device, the sampling of the two process variables does not occur exactly in synchronization. The sampling may operate at two slightly different frequencies. In the example of Figure 2, this is shown sensor 132 operating at a sampling rate of 20 Hz while the second sensor 134 operates at a sampling rate of 20.01 Hz. Although in an ideal configuration the two sensors would operate synchronously, in most practical applications that is not achieved.

As the two sensors are not precisely synchronized, some types of common mode signals can introduce errors into the measurements. Static common mode signals that do not change appreciably from one update to the next are not a problem because the error introduced by the time skew between the two sampling rates is negligible. However, for dynamic common mode signals the error can become quite large. Figure 3A is a graph showing a ramping common mode pressure signal in which the sampling of the process variables PV1 and PV2 are synchronized. In this example, the difference between PV1 and PV2 due to the common mode input signal is zero. However, Figure 3B is a similar graph in which the updates of the sampled process variables are interleveled and do not occur at the same time. This causes the common mode input signal to introduce a difference between the two process variable measurements illustrated as ε. The value of ε is a function of the rate of change of the common mode signal and the difference in time at which the two samples are taken. As a numerical example of this error, it can be shown that with a ramp rate of 10 psi per minute, the resulting error in a differential pressure measurement will be 0.22 inches of water if each device is updating at 20 samples per second. (Assuming a worst case skew error where 10 psi per minute = .0083 psi per 50 mS = 0.229 inches of water per update assuming updates are 50 mS). The time difference is 1 update period. This error will increase if the ramp rate is increased, or if the time between updates is increased.

The present invention provides a compensation circuit or method which can be used to reduce the error due to the above described sampling skew. Figure 4 is a simplified block diagram similar to Figure 2 which shows one example implementation of such compensation circuitry. In the example of Figure 4, compensation circuitry 150 A and 150B are provided. In this example, the compensation circuitry comprises circuitry which increases the damping of the signal from the process variable sensors 140 A and 140B. In this case, low pass filters are provided prior to sampling the process signal. Note that the analog to digital converter can be located in devices 132, 134 or in measurement circuitry 136 as element 142 and the damping function can be implemented before or after the analog to digital conversion. This method reduces time skew error because the low pass filters act to reduce the rate of change, or the ramp rate of the PVl and PV2 signals.

In the configuration of Figure 4, the filters in the two devices 132 and 134 should be matched (i.e., have an identical frequency response). If they are not matched, dynamic common mode signals will appear as normal mode signals to the measurement circuitry 136. For example, dynamic common mode signal would appear as a change in a differential pressure. One disadvantage of this method is that the overall response time of the system is degraded due to the increased damping.

Figure 5 is a simplified block diagram showing a configuration in which compensation circuitry 160 is implemented within the measurement circuitry 136. In such a configuration, various processing techniques can be used to compensate for the time differential between the samples of PVl and PV2. Two examples are described below, however, the invention is not limited to these configurations. Similarly, the invention is not limited to the above described configuration in which the compensation circuitry 150A, 150B is implemented within the sensing devices 132 and 134 and comprises low pass filters.

In a first example configuration, compensation circuitry 160 is used to perform an extrapolation, such as a linear extrapolation, in order to compensate for a ramping common mode input signal. Figure 6 is a graph showing the common mode output signal from devices 132 and 134 versus time. The update times for PVl and PV2 are also illustrated. In such a configuration, a linear extrapolation is used to predict the values of one or both the process variables at a particular time. The predicted value is then used by the measurement circuitry 136 to provide an output as a function of the predicted values of PVl and PV2. This method can be made to function properly if the common mode signal is continuously ramping upwards or downwards. However, if the common mode input signal varies non-linearly, and in particular if it changes direction, errors are introduced into the predicted values. (see Figure 7).

Figure 7 is a graph showing the error for linear extrapolation method when the common mode signal is nonlinear. A second order polynomial can be fit to the last three sample updates to provide a better estimate for input signals including situations in which the common mode signal has some degree of curvature.

Figure 8 is a graph showing another method for addressing sample skew. In the example of Figure 8, an interpolation technique is used to predict a value of both process variables at a past time. Figure 8 shows a nonlinear common mode signal (dashed line). In such a configuration, the simplest method is to use linear interpolation in which the two most recent updates are used to predict the device output at some point of time that is between these two updates. If two devices are involved, a time is selected such that each device history has one update that is at, or later than, the chosen time and one update that is prior to the chosen time.

A further improvement on this technique can be obtained by using a non-linear function such as a polynomial approximation, spline method or other interpolation technique such that multiple updates from the device history are used in order to arrive at a predicted value. Such a method can be very effective provided that the rate of change of the device output is slow relative to the system sample rate period. A compromise with this method is that additional "dead' time is added due to a delay that is required to ensure that there is sufficient device history to compute the approximation. In the graph of Figure 8, an example of this method is illustrated using a quadratic polynomial to approximate the device output.

In Figure 8, the time chosen to predict the output values is between PVlb and PVlc. A predicted value of PV2 is not required because the time chosen exactly coincides with the time that the second to last value of PV2 is received. Thus, the exact value of PV2 can be implemented in this example. The value of PV1 at that time is then predicted by using a second order polynomial that passes through the PV three points PVla, PVlb and PVlc. This polynomial is illustrated as a bold line in Figure 8. The approximate error in this example is illustrated by the symbol ε and is relatively small. In this example, the simple "dead" time has been increased by one update period. However, this will typically have less effect on the overall system response than by increasing the damping to, for example, one second as described in connection with the low pass filter illustrated in Figure 4.

Figure 9 A is a simplified block diagram 180 showing example steps in accordance with one embodiment of the present invention and Figure 9B is a graph of device output versus time. Block diagram 180 begins at start block 182 and at block 184 a timer counter is initiated. At block 186, the system 180 is waiting for an indication that an update is ready to be read from either P_high or P_low. P_high and P_low represent an update from a high pressure side pressure sensor and a low pressure side pressure sensor, respectively. When a low pressure update is ready, control is passed to block 188 and variable P_Low_prev is assigned the value of variable P_Low. Variable T2_prev is assigned the value of variable T2. At block 190, variable P_Low is read and at block 192 variable T2 is updated. Alternatively, if a high pressure update is ready at block 186, control is passed to block 194 where variable P_High is read, and at block 196 variable Tl is updated. At block 198 the value of P_Low at time Tl_prev is predicted and assigned to variable P_Low*. The prediction is based on linear interpolation using variables (T2, P_Low), (T2_prev, P_Low_prev), and Tl_prev. An interpolated value based upon the current P_Low and the previously read low pressure P_Low_prev values is calculated using a straight line linear interpolated technique. Based upon the interpolated value of the low pressure, a differential pressure is calculated at block 200, and provided to the control room 122 in Figure 1. At block 202, variable P_High_prev is assigned the value of P_High and variable Tl _prev is assigned the value of variable Tl . The steps shown in Figure 9 can be implemented in, for example, a microprocessor of measurement circuitry 136. Thus, in such configuration, the microprocessor functions as the compensation circuitry 160.

In the above discussion, the term "interpolation" is used.

However, there is an exception to this case in which extrapolation is used to arrive at a predicted value of P_low at block 198. More specifically, in a typical situation the two process variables are updated at an approximately the same rate and are interlooped in such a manner that the time gap between the two process variables is very discernible. However, in some instances, the time difference between the two process variables may become very small. In this case, the updates are nearly synchronized and it may be possible to receive two consecutive updates from P_low, and then two consecutive updates from P_high. This will alter the interleaving pattern and extrapolation will be required rather than an interpolation to arrive at a predicted value of P_low.

In the above description, the examples are provided for only two process variables. However, any number of process variables may be implemented. The compensation circuitry can be implemented in the device in which the process variable is sensed, in a secondary device, for example, a device in which the process variable is received, or at some other location. With these techniques, the sampling skew error is reduced from two or more asynchronously updating devices using appropriate techniques including low pass filtering, linear or higher order extrapolation, or linear or higher order interpolation. These techniques may be well suited for systems using wireless communication in which the sampled process variables are asynchronous. Further, although the above discussion relates to developing a process variable based upon at least two other process variables, the present invention is also applicable to controlling a process, such as controlling a valve actuator or other process control device, based upon two process variables. In some configurations, the sampled process variables may be time stamped. In such a configuration, the techniques described herein can be used to reduce error due to the sampling time skew between the two devices. The sampling time skew in a wireless environment may become very uncontrolled due to the variable latency in a wireless radio system. For example, a mesh system self organizes to determine how the information routes back to a host. The information from PV1 may route directly to the host and thus have relatively low latency. The information from PV2 may hop through several nodes on the way to the host and thus have relatively high latency. In this matter the self organizing mesh network adds significant uncertainty to the sampling time skew.

Although the present invention has been described with reference to preferred embodiments, workers skilled in the art will recognize that changes may be made in form and detail without departing from the spirit and scope of the invention.

In the above discussion only one of the process variables is compensated. However, in some configurations, it may be desirable to compensate two or more process variables.