1. INTRODUCTION

s. 5

1.1. Introductory remarks

The principal application of measuring transducers for non-electrical quantities is in automatic control systems for industrial processes in industrial plants. Large plants may be equipped with several thousand transducers. Since a change in the temperature of the ambient and of the measured medium is the source of the most significant measurement error, each transducer has a built-in circuit compensating this error. The compensation circuit is the more difficult to realize, the wider the temperature range and the higher the transfer accuracy.

The problems of compensating temperature errors are the subject of an extensive patent literature. In the scientific and technical literature, on the other hand, they are presented sparingly and often as secondary matters. This situation arises for the following reasons:

• compensation circuits are used essentially only in transducers produced in series, and are consequently seldom developed by institutions detached from the manufacturing plant, that is, by universities and institutes,
• the cost of realizing temperature compensation may amount to as much as 50 % of the manufacturing cost of the transducer, depending on the choice of compensation method and the technology employed [1],
• the resultant temperature errors of the transducer (as an integral part of the transfer accuracy class) are a factor that significantly affects its selling price,
• the choice of compensation method is also a factor affecting the permissible operating temperature range, and hence the range of application of the transducer.

Thus the choice of temperature compensation method affects to a considerable degree the economic results of the manufacturer, who consequently keeps it as deep a secret as

s. 6

possible. Patent applications are filed as a necessary evil, owing to the need to protect the manufacturer's own proprietary rights on the market, and errors are often deliberately introduced into such applications to make the method harder to decipher.

The measure of the effect of changes in ambient temperature on the accuracy of measurement is the relative or absolute temperature error. The relative temperature error is most often expressed as a percentage of span, equal to the change of the parameter under consideration (e.g. the lower range value) referred to the span and to a specified change of temperature (e.g. 10 K). The absolute temperature error is most often defined as the difference between the values of the parameter under consideration at the temperature in question and at the reference temperature.

1.2. Aim and thesis of the work

The aim of the work is to develop a new, effective method of compensating the effect of temperature on the transfer accuracy of semiconductor pressure transducers, and to verify its applicability in practice.

By the name "compensation method" is understood in this work a method consisting in the use of a special electronic circuit, cooperating with a semiconductor pressure sensor, whose output signal is made temperature-dependent in such a way that it reduces the effect of temperature changes on the transfer characteristic of the measuring transducer containing this circuit and sensor.

The term "temperature" denotes the temperature at which the measuring sensor finds itself. Depending on the design, this may be the temperature of the measured medium, the ambient temperature, or a temperature resulting from the superposition of both of these temperatures.

"Semiconductor pressure transducer" denotes a transducer containing a sensor built from semiconductor elements, serving to convert the value of the measured gauge pressure (the difference

s. 7

between the measured pressure and atmospheric pressure) into a standard electronic signal.

As will be shown in Section 4, to compensate the effect of temperature on a linear transfer characteristic of the sensor it is sufficient to make the lower and upper range values independent of temperature. Accordingly, we may divide transducers into classes according to the degree of the polynomial describing the temperature error of the lower and upper range values of the uncompensated transducer:

• linear, in which the error is a linear function of temperature
• quadratic, in which the error is a quadratic function of temperature
• cubic, etc.

The new method that is the aim of this work should make it possible to compensate the temperature errors of a given class of transducers by carrying out the compensation procedure at a certain number of discrete temperatures. The number of these temperatures should depend on the degree of the polynomial describing the temperature error. The values of the coefficients of the polynomial, on the other hand, should be established in the course of compensation so as to ensure the nulling of the error of the particular sensor.

The new method should be characterized by ease of introduction into series production, that is, by inexpensive production tooling and the use of mass-produced components. The number of different versions of the elements serving to realize the compensation conditions should be minimized.

The thesis of the work is formulated as follows: It is possible to develop a method of complete compensation of temperature errors in semiconductor pressure transducers using one compensation-setting element for each selected compensation temperature and for each compensated parameter of the transfer characteristic of the transducer, the settings of these elements being characterized by unidirectional independence. The number of compensation temperatures is determined by the degree of the polynomial describing the temperature error of the given parameter.

s. 8

Compensation realized by this method may be incomplete in the case where the number of compensation temperatures is smaller than the degree of the polynomial describing the temperature error characteristic of the sensor.

A compensation procedure for temperature errors using this method is moreover possible such that compensating the error at one temperature has no effect on the compensation at the preceding compensation temperatures.

It is also possible to make the compensation conditions for the lower range value and for the sensitivity of the sensor independent of each other.