It is easy to see an analogy for this in consumer electronics. One of the reasons that the CD player became so cheap was that it could compensate for low-cost, relatively imprecise mechanical components by utilising more complex electronics. In comparison, a vinyl-record turntable requires high-performance - and hence costly - mechanical parts to achieve good audio quality.

Reducing overall cost

For many analogue applications, a major portion of the system cost is the sensor itself. Generally, less expensive sensors have larger inherent errors. The amplifier and the signal conditioning thus need to contribute fewer errors in the rest of the signal chain to maintain overall system accuracy. It is also possible to compensate for sensor errors elsewhere in the signal chain: for example, offset and gain errors can be calibrated out at the cost of more complexity in the signal-chain components. The savings from a cheaper sensor outweigh the additional cost of a more advanced amplifier with calibration capabilities, such as the AD8555 (Figure 1).

Simple compensation approaches involve removing fixed errors at a single temperature such as gain and offset. More complicated approaches compensate for linearity errors and temperature-dependent errors. The latter also require the addition of a temperature-sensing element to determine the required level of compensation.

The end result of this approach is a much more complicated signal-processing network, but one that is able to deliver good performance with low-cost sensors. While the cost of the signal-processing components might increase somewhat, this is more than made up for in the reduced cost of the sensors, delivering an overall reduction in system cost.

Figure 1: The AD8555 signal-conditioning amplifier.

Components throughout the signal chain

We have looked at the example of amplifiers so far, but the same principle of improving signal conditioning to compensate for low-cost sensors applies throughout the signal chain. Depending on the system requirements, while the use of a high-performance amplifier can help minimise errors before A/D conversion, the converter should then also be as accurate as possible, to ensure a minimum level of error is introduced at this stage. Once the signal is in the digital domain, a range of calculations can be used to calibrate out temperature effects and other errors. Calibration coefficients can be stored in advance, and then be used by a DSP or microprocessor (together with readings from a temperature sensor), to perform calculations in real time to get an accurate result.

In fact, some converters now perform so well with low signal levels that an amplifier can be omitted altogether in certain applications. Traditionally, a high-performance amplifier was used to gain up small signals enough so that the converter errors were only a small fraction of the measured input. Modern high-resolution converters' errors are now so small that these devices can provide very good accuracy, even with very small input signals.

So far, we have looked into the input side of a system, but its output should also be considered. When controlling a real-world output, be it an industrial component such as a valve or the modulation of brake pressure in a car, there is a need for an accurate output. With an accurate D/A-conversion stage, and precision linear drivers, tight control of the physical output can be maintained.

Application requirements

Demands for very high accuracy in industrial controls and electronic instrumentation drove the development of high-accuracy components. With tough requirements for both noise and DC accuracy, these applications, along with medical instrumentation and automated test systems, are still a source of strong demand for high-performance analogue components.

More recently, many automotive applications in engine management, safety systems and suspension control demand increased accuracy to achieve the required system performance. These automotive systems utilise sensors in large quantities, driving sensor costs down. In turn, these low-cost sensors have become attractive to industrial and other customers, combining with high-performance signal processing to achieve the required accuracy at a lower overall system cost.