Digital controllers - Intelligent Buildings and Building Automation - page 17

continuously sending back an electrical signal (either a small voltage or cur-
rent) to the input section of the outstation. This is an analogue or continuous
signal which needs to be converted into bits and bytes to form a digital signal
for the CPU to be able to process. The input section therefore contains an
analogue to digital (A/D) converter for this purpose.

2.3.1 Sampling

One of the most important functions of any building automation system
is the collection of continuous measurement data, at regular time intervals
from large numbers of individual measurement sensors, and 'binary' state
data from detectors such as smoke alarms. Measured data is generated con-
tinuously by individual sensors. However, an outstation can only read the
measurements at regular intervals, even though the interval can be rather
short. A measured variable is reconstructed in the system from the meas-
urement of these samples. If a lot of measurements are to be 'read' by the
outstation, each must be sampled at intervals in rotation.
The frequency of sampling must reflect the way in which measured values
themselves change with time. A rapidly changing measured value will have to
be sampled much more frequently than a slowly changing value, in order to
reconstruct its true nature from the samples. Figure 2.5 illustrates the dangers
of sampling too infrequently.
The sampling of the input channels, based on the A/D conversion and
sampling time alone, can be very fast. However, it is important to ask: what
sampling speed is required for the average building services plant to be
adequately controlled? Shannon's sampling theorem can be referred to in
order to determine the proper sampling interval. This states that, provided

Original
data

Sampled
data

Case B

Case A

Reconstructed
data

Figure 2.5 Examples of successful and unsuccessful sampling: A. Adequate (8 samples/
cycle); B. Bad (<2 samples/cycle).