9 Digital computation and intelligent devices

 9.2.3 Smart transmitters

In concept, a smart transmitter is almost identical to the intelligent instruments described earlier. The change in name has occurred over a number of years as intelligent instruments have become smaller and assumed a greater range of functions. Usage of the term ‘smart transmitter’ rather than ‘intelligent instrument’ is therefore mainly one of fashion. In some instances, smart transmitters are known alternatively as intelligent transmitters. The term multivariable transmitter is also sometimes used, particularly for a device like a smart flow-measuring instrument. This measures absolute pressure, differential pressure and process temperature, and computes both the mass flow rate and volume flow rate of the measured fluid.

There has been a dramatic reduction in the price of intelligent devices over the past few years and the cost differential between smart and conventional transmitters is now very small. Indeed, in a few cases, a smart transmitter is now cheaper than its non-smart equivalent because of the greater sales volume for the smart version. Thus, smart transmitters are now routinely bought instead of non-smart versions. However, in many cases, smart transmitters are only used at present in a conventional (non-smart) fashion to give a 4–20 mA analogue measurement signal on the two output wires. Where smart features are used at all, they are often only used during the commissioning phase of measurement systems. This is largely due to the past investment in analogue measurement systems, and the time and effort necessary to convert to measurement systems that can make proper use of intelligent features.

Almost all of the smart sensors that are presently available have an analogue output, because of the continuing popularity and investment in 4–20 mA current transmission systems. Whilst a small number of devices are now available with digital output, most users have to convert this to analogue form to maintain compatibility with existing instrumentation systems.

The capabilities of smart transmitters are perhaps best emphasized by comparing the attributes of the alternative forms of transmitter available. There are three types of transmitter, analogue, programmable and smart.

(a) Analogue transmitters:

require one transmitter for every sensor type and every sensor range

require additional transmitters to correct for environmental changes

require frequent calibration.

(b) Programmable transmitters:

include a microprocessor but do not have bi-directional communication (hence are not truly intelligent)

require field calibration.

(c) Smart transmitters:

 include a microprocessor and have bi-directional communication

include secondary sensors that can measure, and so compensate for, environmental disturbances

usually incorporate signal conditioning and a–d conversion

often incorporate multiple sensors covering different measurement ranges and allow automatic selection of the required range. The range can be readily altered if initially estimated incorrectly

have a self-calibration capability that allows removal of zero drift and sensitivity drift errors

have a self-diagnostic capability that allows them to report problems or requirements for maintenance

can adjust for non-linearities to produce a linear output.

Smart transmitters are usually a little larger and heavier than non-smart equivalents. However, their advantages can be summarized as:

Improved accuracy and repeatability

Long-term stability is improved and required recalibration frequency is reduced

Reduced maintenance costs

Large range coverage, allowing interoperability and giving increased flexibility

Remote adjustment of output range, on command from a portable keyboard or from a PC. This saves on technician time compared with carrying out adjustment manually

Reduction in number of spare instruments required, since one spare transmitter can be configured to cover any range and so replace any faulty transmitter

Possibility of including redundant sensors, which can be used to replace failed sensors and so improve device reliability

Allowing remote recalibration or re-ranging by sending a digital signal to them

Ability to store last calibration date and indicate when next calibration is required

Single penetration into the measured process rather than the multiple penetration required by discrete devices, making installation easier and cheaper

Ability to store data so that plant and instrument performance can be analysed. For example, data relating to the effects of environmental variations can be stored and used to correct output measurements over a large range.

 

Summary of smart transmitter features

Many of the features of smart transmitters are common with those of smart sensors, and the comments made earlier about smart sensors therefore apply equally. However, the use of multiple primary sensors and secondary sensors to measure environmental parameters mean that additional comments are necessary in respect of their selfcalibration and self-diagnosis capabilities.

 

Self-calibration

Calibration techniques are very similar to those already described for smart sensors and the general principle is always to use simple calibration methods if these are available. Look-up tables in a smart transmitter have a particularly large memory requirement if correction for cross-sensitivity to another parameter (e.g. temperature) is required because a matrix of correction values has to be stored. Hence, interpolation calibration is even more preferable to look-up tables than it is in the case of calibrating smart sensors.

 

Self-diagnosis and fault detection

Fault diagnosis in sensors is often difficult because it is not easy to distinguish between measurement deviation due to a sensor fault and deviation due to a plant fault. The best theoretical approach to this difficulty is to apply mathematical modelling techniques to the sensor and plant in which it is working, with the aim of detecting inconsistencies in data from the sensor. However, there are very few industrial applications of this approach to fault detection in practice, firstly, because of the cost of implementation and, secondly, because of the difficulty of obtaining plant models that are robust to plant disturbances. Thus, it is usually necessary to resort to having multiple sensors and using a scheme such as two-out-of-three voting. Further advice on self-checking procedures can be found elsewhere (Brignell, 1996).

 

Effect of sensor errors

The effect of a sensor error on the quality of measurement varies according to the nature of the fault and the type of sensor. For example, a smart pressure sensor that loses temperature measurement will still give valid measurements but the uncertainty increases.