Efficient calibration of binary-weighted networks using a mixed analogue-digital RAM

The architecture and design methodology are described for an area-efficient calibration network organized as a mixed analog-digital RAM (random-access memory) which can be easily controlled and which possesses a very compact layout. The density of the calibrating RAM, expressed in terms of the number of calibrating arrays and of calibrating elements in each array, depends not only on the resolution and element matching accuracy of the network to be calibrated, but also on the calibrating elements themselves. Such a calibration network can be used in any data conversion system based on networks of binary-weighted elements. A CMOS IC implementation is described.<<ETX>>