September 12, 1997

Build a smart analog transmitter
with a µC and low-power components

Albert O'Grady and Jim Ryan, Analog Devices

Analog transmitters have evolved from simple 4- to 20-mA current loops to designs that include µCs for signal conditioning at the transmitters and designs that allow bidirectional digital communication.

An analog transmitter is a field-mounted device that senses a physical parameter, such as pressure or temperature, and provides a current signal in a 4- to 20-mA range, which is proportional to the measured variable. The 4- to 20-mA standard provides a basis for the development of process instrumentation and has numerous advantages. Changes in loop resistance don't affect the measurement signal, transmitters from different manufacturers are interchangeable, and loop voltage provides the power required to energize the transmitter circuits. A conventional 4- to 20-mA transmitter (Figure 1a) basically consists of a power supply, a transmitter, and a controller. The transmitter sends a current proportional to the measured variable across the loop.

The second generation of transmitters--"smart" analog transmitters--contains a microprocessor, although the measured variable transmitted over the loop is still an analog current (Figure 1b). Memory and computing power at the transmitter allow signal conditioning before transmission on the loop back to the controller. For example, signal conditioning can correct the well-known nonlinearities of resistance-temperature detectors and thermocouples. In these types of transmitters, the ADC converts the variable, which the µP can then linearize using mathematical computation before a DAC transmits the signal back onto the line. The provision of linearized and corrected signals to the control system takes the burden of performing these computations off the main controller in the control room.

The third generation of transmitters provides for a digital-communications channel as well as the traditional 4- to 20-mA signal on the same twisted-pair line (Figure 1c). The communications channel allows digital transmission of the measured variable over the twisted pair. You can also send other diagnostic data that is relevant to the transmitter, such as calibration coefficients, device ID, and fault diagnostics, back and forth on the twisted pair from the transmitter to the controller. This enhancement in analog-transmitter design allows greater flexibility in diagnosing transmitter faults; when the transmitter sits in a hazardous environment, you can remotely diagnose the malfunction from the transmitter. The de facto standard for digital communication over 4- to 20-mA current loops in industrial applications is the Highway Addressable Remote Transducer (HART) protocol, which Rosemount Inc (Eden Prairie, MN) developed. The standard is now under the control of the HART Communication Foundation (www.ccsi.com/hart/hcp0000a.html).

Design a low-power analog transmitter

Today, you can design a smart analog transmitter, based on the block diagram in Figure 1b, that consumes less than 3.5 mA (Figure 2). In other words, the transmitter, ADCs, µP, and DAC must consume less than 3.5 mA. This level of power consumption allows you to power the transmitter from the loop voltage.

A traditional data-acquisition system for the transmitter front end consists of a multiplexer to multiplex the inputs from various sensor inputs to the ADC (Figure 3). Before sampling by the ADC, the low-level signals from the sensors go through a signal-conditioning stage that includes a high-gain instrumentation amplifier and filter. The system processor controls the programmable-gain feature of the instrumentation amplifier and the data acquisition by the ADC. Channels share the instrumentation amplifier and the filter. If each channel has known but different noise characteristics, you can make the filter's time constant programmable under the system processor's control.

In traditional systems, the multiplexer, instrumentation amplifier, and ADC are usually independent elements on a circuit board that take up much space and are also layout-sensitive, given the low-level input signals that the sensor generates. However, you can find single ICs that integrate all of the elements within the dashed box of Figure 3, thus reducing the required board size and removing layout issues from the system design, ultimately leading to a lower cost system with improved performance. The ideal ADC for this application incorporates several on-chip features to reduce the number of front-end components. The ADC should accept low-level differential signals directly from the transducer and should have an on-chip programmable-gain amplifier to condition the bridge signal to obtain the absolute maximum dynamic range from the ADC. A resolution of 16 to 24 bits provides conversion to the required system resolution and accuracy. The ADC should typically operate from one 3 to 5V supply voltage.

On-chip features that allow you to periodically perform calibration are also useful for removing gain and offset errors--not only in the part itself, but also in the system. With minimum software overhead, the system can also calibrate out the effects of temperature drift and drift over time.

Serial interfacing is now becoming a standard in smart-transmitter applications due to its flexibility, low signal-line count (typically, three wires), and the ease with which you can incorporate isolation into the design. The ideal interface allows the system to operate with three wires and allows you to configure the gain, channel selection, and calibration over these wires.

Another key element in selecting the front-end ADC is power consumption. Because the whole transmitter when powered from the 4- to 20-mA line must consume less than 3.5 mA, the ADC itself must consume little power--typically, less than 750 µA.

Select a suitable µC

The µC is the engine of the smart transmitter, controlling the transfer of information from the sensors on the front end to the 4- to 20-mA current representing the primary variable on the loop. Memory in the µC allows periodic calibration in the system and allows the transmitter to condition the measured variable before converting it to a current and transmitting it on the loop. The µC can perform much more than just mathematical calculations in the system. For example, you can use the µC to make other measurements of temperature or pressure using an auxiliary sensor channel and then use the results of these measurements to correct the primary sensor's measured value. Having this facility locally at the transmitter reduces the burden on the controller in the control room because all the linearization and other signal conditioning occur before the data transmits over the loop. This design leads to more efficient use of the central controller and, thus, enhanced system performance.

When selecting a suitable µC for a smart transmitter, you need to achieve:

• Low power consumption: The current consumption must be low enough to enable the loop to power the µC and all other components in the transmitter.

• Adequate onboard memory: Ideally, the µC should contain enough onboard ROM and RAM to enable the implementation of all software functions (boot program plus data processing) without external memory, thereby reducing system component count and power consumption.

• Serial communications interface: Serial interfacing, usually with three wires, is preferable in transmitter applications, because you can implement galvanic isolation with a minimum of optoisolators. You should provide a serial communications port to interface with the ADC on the front end and with the DAC on the transmitter's output side.

The µC's clock speed is the main factor in determining the µC's power consumption. The power consumption of CMOS devices is generally directly proportional to clock speed. Therefore, from a power standpoint, you obtain the lowest power consumption by running the processor as slowly as possible. Fortunately, a selection of µCs (Table 1) features sufficient onboard memory and low enough power consumption to enable the loop to power them and for the µC to implement the software-function requirements in a typical transmitter application.

Some ADCs enable a simple and direct interface to the serial port of a µC, such as the 68L11 (Figure 4). No external glue logic is necessary for this interface, which configures the 68L11 in master mode with its CPOL bit set to a logic one and its CPHA bit set to a logic one. In this mode, the serial clock idles high between transfers. The µC monitors the DRDY bit in the communications register of the ADC to determine when the data register of the ADC updates and when it is valid to read data from the part.

Complete the design

After you choose the ADC and µC, you must consider how this arrangement will drive and control the loop current. Most industrial applications feature remote instrumentation that receives power across the same interface that transmits the measured result, or process variable. Some applications use the 4- to 20-mA interface to send a command signal to a valve or actuator, but these applications typically use more power than you can effectively derive from a 4- to 20-mA loop signal. Hence, the interface for these applications normally comprises two wires for the command signal and two separate wires for powering the device. (This article focuses on two-wire, remote-powered transmitters.)

Because the process-variable signal that transmits over the 4- to 20-mA interface represents the result of measuring, linearizing, scaling, and, possibly, filtering the measured parameter, the signal chain must provide accurate and monotonic behavior. Therefore, the selection of current-loop control circuitry is critical to maintaining the resolution and accuracy of the ADC and front-end-conditioning circuitry. This monotonic requirement, coupled with at least 12 bits of accuracy, provides the minimum requirement for choosing a converter to provide current-loop control. Again, the converter's power consumption is important, given that the remote circuit may have to operate on 3.5 mA from the loop supply, because this current represents the low-alarm current setting.

The control of the loop current, which is proportional to the process variable, has evolved with the advances in transmitter design. The earliest transmitters used exclusively analog-signal processing from the transducer to the scaled loop current. The advent of low-power µCs and data converters introduced the concept of smart transmitters and enabled more advanced signal processing. Hence, most designs perform signal processing in the digital domain.

The basic loop-current-control configuration comprises a DAC that generates a voltage setpoint, which in turn produces a corresponding current according to the values of R₁ and R₂ (Figure 5). Low-cost techniques for implementing the DAC include customized PWM-type DACs (Figure 6a) with which you can easily implement galvanic isolation (Figure 6b) between the input and output. Isolation is often necessary for intrinsic safety in the transmitter. You can implement a PWM DAC by using a programmable counter/timer peripheral, which is often available on modern µCs. You simply lowpass-filter this output to provide a dc level, which is the desired DAC output. You can also successfully implement low-power CMOS R-2R ladder DACs with serial interfaces in transmitter designs because of their low power requirements. Also, sigma-delta DACs provide a variety of benefits, including low power, low cost, high resolution, and guaranteed monotonicity.

You also need to choose voltage regulators and references. The loop remotely receives power from a control room or possibly through intrinsically safe barriers. You typically set the loop voltage to allow a current greater than the maximum allowed (typically 24 mA for the high alarm-current level) through the load resistance. Some head room is necessary so that the circuitry in the smart transmitter can operate. The transmitter circuitry must regulate this head-room voltage to ensure a stable operating power supply, irrespective of loop-current fluctuations.

You can add a precision reference to provide accurate, low-drift operation, and an output amplifier provides a transconductance stage, which converts a voltage setpoint to a controlled current using an external transistor (Figure 5). Given the operating-temperature range of some remote transmitters, the components you choose must have suitable temperature compensation to prevent temperature drift from impacting their key specifications. Precision references are also necessary for the data-converter circuits to ensure accuracy of operation. Most applications generally use two- or three-terminal, TO92-style devices.

Transmitters modulate current

The third generation of transmitters with digital-communications capability adds another level of functionality. These "intelligent" transmitters not only can send a process-variable signal on the current loop in analog form but also can transmit and receive digital information by modulating the current in the loop. Whereas the most basic smart transmitters are remote instruments with no means of communication other than the analog process-variable signal and high and low alarm-level settings, these transmitters can interactively communicate with the control room so that the main controller can poll the transmitter for detailed status information at any time. This communication also allows the control room to trim the transducer output to measure the span from 4 to 20 mA. A modem circuit sends and receives the digital signals. The modem can translate the received modulated current into digital ones and zeros and can translate the digital levels into modulated-current levels for transmit. Such intelligent transmitters require an external HART modem circuit and require that the µC has a suitable UART-type interface.

The HART protocol uses the Bell 202 frequency-shift-keying (FSK) standard to superimpose low-level digital signals onto the standard 4- to 20-mA signal. The technique involves modulating the loop current to 1200 and 2200 Hz, representing "mark" (1) and "space" (0), respectively. With suitable filtering, these ac signals have no significant effect on the dc signal that represents the process-variable measurement. This scheme enables two-way communication and makes it possible to relay additional information beyond just the normal process variable to and from a smart field instrument. The protocol allows a host application (master) to get two or more digital updates per second from a field device, and, because the digital FSK signal is phase-continuous, there is no interference with the 4- to 20-mA signal.

The HART protocol offers simultaneous analog and digital communications and allows access to all instrument parameters and diagnostics. The protocol is an open de facto standard with a common command and data structure and a device-description language. The relative simplicity of the HART protocol makes it easy for both end users and suppliers to gain experience and benefit from the enhanced two-way-communication capability of smart field instruments.

One major advantage of the HART protocol is that you can implement it in transmitters without installing new cabling. Thus, you can upgrade many transmitters by simply replacing the transmitters with HART-compatible ones. Often, many of these transmitters can measure the "primary" and "secondary" process variables. A simple 4- to 20-mA interface can transmit only one process variable. Intelligent transmitters can send information about two process variables as well as other relevant information. You can use the HART configuration only for digital communication, in which case the loop current is simply used to power the transmitters and conveys no information.

| EDN Access | Feedback | Table of Contents |