Browse Prior Art Database

Elevated Signal Amplitude as a Data Recovery Procedure

IP.com Disclosure Number: IPCOM000122553D
Original Publication Date: 1991-Dec-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 2 page(s) / 69K

Publishing Venue

IBM

Related People

Kerwin, GJ: AUTHOR

Abstract

For a partial response sampling detection channel, the expected sample values that correspond to binary one, minus one, and zero-coded data bits are 48, 16, and 32 least significant bits (LSBs) for a six-bit analog-to-digital converter (ADC). Based on these ideal or expected values, the boundary between the valid one and zero and minus one and zero is plus or minus eight LSBs either side of decimal 32 (the expected zero value). This means that the margin to error for a one to be misclassified as a zero is equal to the margin to error for a zero to be misclassified as a one. With random noise each data sample is corrupted in the same way so that this equal margin to error scheme produces the lowest probability of error.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 65% of the total text.

Elevated Signal Amplitude as a Data Recovery Procedure

      For a partial response sampling detection channel, the
expected sample values that correspond to binary one, minus one, and
zero-coded data bits are 48, 16, and 32 least significant bits (LSBs)
for a six-bit analog-to-digital converter (ADC).  Based on these
ideal or expected values, the boundary between the valid one and zero
and minus one and zero is plus or minus eight LSBs either side of
decimal 32 (the expected zero value).  This means that the margin to
error for a one to be misclassified as a zero is equal to the margin
to error for a zero to be misclassified as a one. With random noise
each data sample is corrupted in the same way so that this equal
margin to error scheme produces the lowest probability of error.

      If, on the other, the one and zero sample values were subjected
to dissimilar corruption, then dissimilar margins for one and zero
data samples would yield better error rates.  Defects in the
recording media can cause this unequal corruption of one and zero
samples.  Certain media defects will reduce the value of the plus or
minus one samples more than the zero samples are increased or
decreased from the baseline (decimal 32 for the six-bit ADC).  This
means that an automatic gain control loop that increased the signal
amplitude as it entered the ADC while allowing the expected values
and, therefore, the margin to error to remain at eight LSBs, would be
more immune to corruptio...