Sensorless estimation of global device-parameters based on Fmax testing

Post-fabrication performance compensation and adaptive delay testing are indispensable means for improving yield and reliability of LSIs, The global parameter estimations, such as of threshold voltages, play a key role in maximizing their effectiveness. This paper proposes a novel technique that realizes an accurate device-parameter estimation through F
_max
testing framework. In the proposed method, statistical path delay distributions of sensitized paths in F
_max
testing are utilized to calculate device-parameters, such that they most likely explain the measurements in the F
_max
testing. Two estimation procedures are proposed: one utilizes discrete Bayesian estimation and the other uses maximum likelihood estimation. Numerical experiments demonstrate that both methods achieve 2.5mV accuracy in estimating threshold voltages.