Dynamic Constraints for Large-Signal Measurements on Arbitrary Grids

This paper presents a novel method of dynamic constraints, which allows safe large-signal measurements on arbitrary grids of samples. The method is a generalization of popular sweeps, and alike, it is based on the monotonic relationships linking some of the input variables with the setup degradation level. However, the knowledge of exact physical mechanisms is not required, which makes dynamic constraints applicable to wide range of experiments. The detailed explanation of the dynamic constraints principles and implementation is provided. Two cases are analyzed depending whether the degradation level is a monotonic function of all the experiment's input variables or only some of them. The method is assessed on 6 × 50 μm GaAs HEMT large-signal measurements with the sparse grid of samples generated with adaptive sampling methods. Thanks to dynamic constraints, none of the samples seriously violated the constraints, even though no a priori knowledge about the device behavior was available. No significant device degradation is observed in the measurements.