Browse Prior Art Database

Analog Conversion Error Management Procedure - Digital Control Radex Procedure

IP.com Disclosure Number: IPCOM000117145D
Original Publication Date: 1995-Jun-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 44K

Publishing Venue

IBM

Related People

Woodman Jr, GR: AUTHOR

Abstract

Digital to Analog Converter (DAC) or Analog to Digital Converter (ADC) can suffer from defects of bit size due to variability of the manufacture of the parts. As the desire for more and more sophisticated functions in smaller and smaller packages continues, placement of ADC or DAC on otherwise VLSI logic chips makes the possible variability much greater.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 68% of the total text.

Analog Conversion Error Management Procedure - Digital Control Radex
Procedure

      Digital to Analog Converter (DAC) or Analog to Digital
Converter (ADC) can suffer from defects of bit size due to
variability of the manufacture of the parts.  As the desire for more
and more sophisticated functions in smaller and smaller packages
continues, placement of ADC or DAC on otherwise VLSI logic chips
makes the possible variability much greater.

      Errors in bit size in an ADC will appear as a gap or a shadow,
where there are numbers that will never be produced, no matter how
small an analog step is made by increasing or decreasing the analog
value.  Errors in bit size of a DAC will appear as a step in output
that is larger (or smaller) than the number change imposed.

      For instance, if an 8-bit binary DAC has an error in bit four
equal to four unit bit values, when the number increments from
00001111
to 00010000, the output will jump four units instead of one unit.

      In the special cases of ADCs (sequential or two step flash
ADCs, etc.) or DACs that control servo mechanisms, it is possible to
tolerate such errors by including a dual set of low order conversion
bits.  With two sets of the low order four bits of the 8-bit DAC, if
bit four is too big, it is possible to count from:
       0000 1111
         1111
  to:
       0001 0000
         1111.

Then when the servo calls for a reduction in the value, it is
possible to...