The use of the triangle functions to model analog-to-digital converter dynamic errors

Triangular basis functions are used to model analog-to-digital converter (ADC) error functions. The procedure is found to require less calibration data than conventional methods without affecting the dynamic error correcting performance of the ADC. An example uses 64 parameters to model 4096 error states for a 6-b ADC. It is shown that 60 K data samples build suitable error tables where 3584 K data samples are customarily required. This reduction in data is on the order of 2/sup Nbits/.<
>