9 Digital computation and intelligent devices

 9.2.4 Communication with intelligent devices

The subject of instrumentation networks and digital communication with instruments is covered in detail in the next chapter. The aim over many years has been to use intelligent devices to their full potential by making all communications, including the measurement signal, entirely digital. A number of digital fieldbuses are now used for instrumentation systems, with protocols such as Profibus and WorldFIP being in widespread use. However, to date, despite international efforts over many years, no standard protocol for digital fieldbus communications has yet been established.

Partly because of this delay in developing an international digital fieldbus standard, and partly because of the need to maintain compatibility with the vast current investment in analogue instrumentation systems, a number of part analogue/part digital communication protocols have been developed as an interim measure. Prominent amongst these is a protocol called HART (Highway Addressable Remote Transducer). This is a manufacturer-independent protocol that provides for analogue measurement signal transmission as well as sending command/status information digitally. The normal requirement for such dual analogue/digital communication with an intelligent device is six wires, two to convey the measurement signal, two to convey command/device status information and two to provide a power supply to the device. However, in order to economize on wiring and installation costs, HART allows this requirement to be reduced to four or even two wires by using the signal wires to convey device status/command signals or the power supply or both of these. HART has now achieved widespread use, even though it is not backed by an international standard.

 

Extended 4–20 mA current interface protocol

The 4–20 mA protocol is the most-used analogue transmission mechanism because of the protection against noise that it offers to the measurement values transmitted. This protocol has been extended for communication with intelligent devices to allow for the transmission, where necessary, of command/status information and the device power supply in analogue form on the signal wires. In this extended protocol, signals in the range 3.8 mA to 20.5 mA are regarded as ‘normal’ measurement signals, thus allowing for under- and over-range from the 4–20 mA measurement signal standard. The current bands immediately outside this in the range 3.6 mA to 3.8 mA and 20.5 mA to 21.0 mA are used for the conveyance of commands to the sensor/transmitter and the receipt of status information from it. This means that, if the signal wires are also used to carry the power supply to the sensor/transmitter, the power supply current must be limited to 3.5 mA or less to avoid the possibility of it being interpreted as a measurement signal or fault indicator. Signals greater than 21 mA (and less than 3.6 mA if the signal wires are not carrying a power supply) are normally taken to indicate either a short circuit or open circuit in the signal wiring.

 

Sending commands to sensor/transmitter

Commands can either be sent from a handheld keyboard or else communicated from a remote PC. Whilst a handheld keyboard is the cheaper option in terms of equipment requirement, it cannot store calibration data because it does not usually have any memory. It is therefore time consuming for a technician to enter the necessary calibration data manually. A PC makes communication easier because it can readily store calibration data. Also, its large screen allows more information to be viewed at one time. A PC is also able to receive status data from the sensor and store it for later use (e.g. to disclose trends in sensor status).

For hazardous environments, versions of smart transmitters are available that are made intrinsically safe by using reed-relay switches to alter transmitter parameters. In such cases, an LCD programming display is usually used to give commands to the transmitter, as this is also intrinsically safe.

 

9.2.5 Computation in intelligent devices

In the past, most computation in intelligent devices has been performed by software routines executed on a general-purpose microcomputer. However, there has been a trend in the last few years towards implementing digital signal processing, data conversion and communication interface functions in specially designed hardware elements. This achieves a large improvement in processing speed compared with the execution of software routines. The first implementations of this (Brignell, 1996) used ASICs (Application Specific Integrated Circuits). An ASIC is a gate array that is programmed by designing a mask that creates connections between elements in the device. Unfortunately, the mask design is a very costly process and therefore such devices are only cost effective in high-volume applications such as automobile systems.

More recently, alternative programming devices such as FPGAs (Field Programmable Gate Arrays) and CPLDs (Complex Programmable Logic Devices) have become available that offer a means of implementing digital signal processing and other functions that are cost effective in low-volume applications. Implementation cost is reduced because these devices are user-programmable and avoid the very expensive mask design process required by ASICs. In fact, these alternative programming devices are now routinely used to build prototypes of designs before going into production of ASICs (which are still more cost effective than FPGAs and CPLDs in high-volume applications).

A typical programmable device consists of an array of configurable logic blocks, programmable input–output blocks and memory. However, FPGAs and CPLDs differ substantially in the way that elements are connected within the device, with the connections used in a CPLD allowing faster operation of the device. These differences mean that, in general, the CPLD is preferred for applications where there is a requirement for high processing speed and the FPGA is preferred where there is a need for high capacity (number of logic gates) in the device.

Some programmable devices contain both soft and hard cores, in which a hard core performing specific functions such as a PCI (Peripheral Component Interconnect) interface is embedded in a programmable soft core. In this hard-core/soft-core approach, the inclusion of hard-core elements increases computational speed and reduces size, but the increased specialization of the device reduces the number of potential applications and therefore increases unit cost.

Further information on these modern programmable devices can be found elsewhere (Amos, 1995; Brown, 1996).

 

9.2.6 Future trends in intelligent devices

The extent of application of smart transmitters is currently limited by:

Lack of sufficient varieties of transmitter due to manufacturers’ reluctance to invest in producing them ahead of agreement on an international fieldbus standard

Limitations on the power of microprocessors available

The large investment in conventional 4–20 mA signal transmission systems and cabling, thus inhibiting the use of transmitters to their full potential in fully digital transmission mode

Limitations in the speed of bi-directional communication capabilities. If a common bus is used to transmit signals for several different transmitters, data transfer speed is slow because the bus can only service one transmitter at a time. This means that the time interval between measurements from a particular transmitter being read and responded to can become excessively large. In consequence, it is possible for dangerous conditions to develop in the controlled plant, such as high pressures.

Current research and international discussions are currently directed at solving all of these problems. Hence, a rapid growth in the application of intelligent devices, and their use in fully digital mode, is expected over the next few years. Size reductions will also continue, and indeed the first smart microsensors are now available. These are covered in greater detail in Chapter 13. The establishment of an international fieldbus standard (see Chapter 10) will also encourage greater use of intelligent devices in all-digital instrumentation and control schemes for industrial plant.

The use of programmable devices to perform signal processing functions within intelligent devices is likely to expand rapidly in the future. As well as further improvements to the processing capacity and computational speed of these devices, current research (Tempesti, 1999) is directed towards developing self-repairing capabilities in such devices.

Also, now that both sensors, processing elements and microcontrollers can all be constructed on silicon wafers, the next logical step is to extend the process of integration still further and include all of these elements on a single silicon chip. Apart from the reduction in system cost due to the reduction in the number of components, the requirement for fewer connections between components will lead to substantially improved system reliability, since most system faults can be traced to connection faults. However, whether, or how soon, this further integration will happen will depend on the relevant economics of separate and combined implementation of these system components.