SUMMARY

Converts continuous data into its quantized equivalent.

SYNTAX

QUANTIZE [GAINS n ...],[LEVEL v],[MANTISSA n]

INPUT

GAINS n ...:	Set list of allowed gains. They must be monotonically decreasing. The maximum number of allowed gains is 8.
LEVEL v:	Set the quantization level of the lowest gain. This is the value of the least significant bit in volts.
MANTISSA n:	Set the number of bits in the mantissa.

DEFAULT VALUES

QUANTIZE GAINS 128 32 8 1 LEVEL 0.00001 MANTISSA 14

DESCRIPTION

This command exercises a quantization algorithm equivalent to the "rounding" quantization described in Oppenheim and Schafer (1975, Fig. 9.1). The number of bits used in this algorithm are partitioned into the bits used to represent the characteristic (exponent), the sign bit, and the mantissa bits. The user can specify the number of bits used for the mantissa. The quantization level (value of least significant bit or LSB) can also be specified by the user. The default quantization level is 10 microvolts. The error of the signal represented by this quantized function is numerically equal to one-half of this quantization level. In the spectral domain, this error or quantization noise is:

ERROR = 1/12 * (DELTA * LEVEL^2)

where DELTA is the sampling interval. This quantization noise is measured in units of counts*counts/Hz, as a power spectral density. The rms-squared quantization noise is:

(1/6)*LEVEL^2.

However, this is an accurate approximation to the noise due to quantization only if the rms level of the signal is much larger than the rms quantization noise. In other words, if the signal is not resolved by several hundred counts, then there is a correlation between the quantization noise and the signal being quantized. The fraction of correlation is approximately equal to the ratio of the LEVEL to the rms of the signal being quantized (see Fig. 11.13, Oppenheim and Schaffer, 1975). The gains can be specified by the user to simulate the gain steps in an automatic gain-ranging system. The default gains are those of the Regional Seismic Test Network (RSTN.) Oppenheim, Alan V., and Ronald W. Schafer; Digital Signal Processing; Prentice-Hall; 1975; 585pp.

HEADER CHANGES

DEPMIN, DEPMAX, DEPMEN

LATEST REVISION

May 15, 1987 (Version 10.2)