| 摘要 |
<p>An integrated circuit chip has circuitry for converting a binary coded value to an analog value. The chip includes first and second matrices each defined by a plurality of rows and a plurality of columns. The rows and columns have sources at different positions for producing currents in response to binary signals coding for the binary value. Each row on the first matrix is connected to a row in the second matrix on a reverse-image basis. For example, if each matrix has thirty two (32) rows, rows 1 and 32 on the first matrix are respectively connected to rows 32 and 1 on the second matrix. The rows on the first and second matrices are sequentially selected in a pattern providing particular convergences and divergences of successive pairs of such rows in each of such matrices. Such sequential selection provides for cyclic deviations between the rows in the successive pairs in each of the matrices about the center of the matrix as a reference. Such cyclic deviations provide for progressive convergences and then progressive divergences of the rows in each of the successive pairs in each of the matrices, such progressive convergences and progressive divergences occurring about the center line as a reference. Such progressive convergences and divergences may occur in an least a pair of successive cycles. When the selection occurs in at least a pair of successive cycles, the selected rows in each matrix in one of the cycles are interleaved in such matrix with the selected rows in the other cycle. Upon the selection of each row in the first and second matrices, the different positions in the selected row may be progressively selected. In this way, compensation may be provided on the chip for second order errors such as result from stresses in the chip.</p> |
