Quantizing
Quantizing is the process of dividing an analog input signal into a string of
equally spaced discrete outputs, each of a constant amplitude. Binary codes
can be generated for each of these levels, and the quantized waveform can be
stored in a microcomputer memory. The sample-and-hold function performs
the equally spaced sampling of an input signal, while the A/D converter per.
forms the binary coding. Figure 5-l8 illustrates the input, quantized, and coded
waveforms.’ The quantized waveform differs slightly from the input waveform,
because it is constructed as a series of discrete steps. The difference in the
waveforms is the quantizing error. The degree of the error is dependent on th¢
sampling resolution and the irregularity of the waveform. The duration of one
sampling period is defined as Q, and the quantizing error is defined as the time
between samples, as measured from the center of each sampling period. Qua“.-
tizing error can be reduced by reducing Q through faster sampling.
The rate of sampling is limited to the rate at which samples can be con.
vetted to output codes by the A/D converter. The aperture time is the time it
takes an A/D converter to perform one conversion. The coding enor intro-
duced by the A/D is called the amplitude uncertainty.
The questions asked most often in regard to choosing a data acquisition
system are these: What amplitude resolution should the sampling be? How
many bits should the A/D converter generate’? How fast should the sampling
rate be? One guide often used in determining sampling rate is the theorem that
states that an input signal with a highest frequency component off can be
recovered without distortion using a sampling frequency of 2f. A sampling rate
Of 2f or greater will therefore minimize the likelihood that analog information is
being lost in the quantizing process.