Browse Prior Art Database

BINARY AMPLITUDE, DISCRETE TIME, DIGITAL GAUSSIAN PREMODULATION FILTER

IP.com Disclosure Number: IPCOM000006262D
Original Publication Date: 1991-Dec-01
Included in the Prior Art Database: 2001-Dec-19
Document File: 3 page(s) / 164K

Publishing Venue

Motorola

Related People

James M. Keba: AUTHOR

Abstract

The counter is clocked at N (N=32 for examples in this document) times the bit rate. The counter value com- bined with an M (M=2 for bT=OS, M=3 for bT=O.25) ADVANTAGES OF A BINARY AMPLITUDE, DISCRETE TIME APPROACH There are several problems which arise in the use of an analog gaussian premodulation filter. This type of filter causes the zero crossing locations of the output waveform to be a function of the binary input pattern. This type of filter causes the peak output obtained for certain binary inputs to not reach their theoretical maximum. Analog filters will exhibit a non uniform delay, and their perfor- mance will vary over time, temperature, and external component value. Several external resistors and capaci- tors are necessary to realize an analog fdter.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 38% of the total text.

Page 1 of 3

0 M

MOTOROLA INC. Technical Developments Volume 14 December 1991

BINARY AMPLITUDE, DISCRl3E TIME, DIGITAL GAUSSIAN PREMODULATION FILTER

by James M. Keba

  The counter is clocked at N (N=32 for examples in this document) times the bit rate. The counter value com- bined with an M (M=2 for bT=OS, M=3 for bT=O.25)

ADVANTAGES OF A BINARY AMPLITUDE, DISCRETE TIME APPROACH

  There are several problems which arise in the use of an analog gaussian premodulation filter. This type of filter causes the zero crossing locations of the output waveform to be a function of the binary input pattern. This type of filter causes the peak output obtained for certain binary inputs to not reach their theoretical maximum. Analog filters will exhibit a non uniform delay, and their perfor- mance will vary over time, temperature, and external component value. Several external resistors and capaci- tors are necessary to realize an analog fdter.

  A binary amplitude, discrete time, digital gaussian premodulation tilter will exhiiit no variation in zero cross- ing locations of the output waveform with binary input pattern. The derivative of the output waveform will con- tain no discontinuities. The peak of the output waveform will reach its theoretical maximum. The delay is pre- dictable, and will not vary as a function of binary input pattern, time, temperature, or component variation. A digital gaussian premodulation filter can be integrated into a single IC with few external components.

ARCHITECTURE

  The key feature of the digital gaussian filter archi- tecture is that the input is restricted to a binary ampli- tude, discrete time signal. This allows the filter to obtain near ideal performance. A continuous amplitude, dis- crete time digital gaussian premodulation falter would exhibit similar zero crossing jitter and eye opening reduc- tion as an analog design, but would have the other inher- ent advantages of a general digital filter design.

  The digital gaussian premodulation filter consists of a shift register to hold multiple bits of the bit pattern to address the ROM, a counter to address the ROM, a ROM to hold digital values of the premodulation filter output, a D/A converter to convert the digital values of the premodulation filter output to an analog value, and an analog reconstruction filter.

Prern0- dulah Piker OU,pUt

bit shift register value forms a ROM address which pro- duces a digital premodulation falter output which is con- verted to a analog value by the L (L=ll for examples in this document) bit D/A converter and reconstructed by a simple low pass Wer. New bits are shifted into the shift register every N clock pulses.

Digital values necessary to implement Gaussian filters with bT=ll.S and bT=O.25 are shown.

bt=O.S: 32 samples/bit; 2 bit observation window

theta w(0) w(l) w(2) w(3)

Al5m-l.ooo-l.m l.ooo 1.000

Al.46875 -l.ooO -0.995 0.995 l.ooO

4.43750-l.ooO-a.981 0.981 Loo0

-0.40625 -1.000 4.957 0.957 l.CUl

-0.37500 -l.ooo -0.924 0.924 l.ooo

~I.34375...