Delivering mismatch data that reflect design reality is a real challenge. Indeed, from test structures to final data utilization, many steps can be the source of distortion. The first possible source of distortion is linked to the differences in terms of environment and spacing that might exist between test structure transistors and circuit transistors. The second potential source of distortion is related to the measurements and extraction that can both add extra mismatch. Finally, the data treatment and utilization can constitute other error sources. In this paper, thanks to results from various test structures and device types, the main sources of distortion are pointed out in order to help to set up a reliable chain from matching test structures to matching data utilization.
