Improving the Accuracy of RF Alternate Test Using Multi-VDD Conditions: Application to Envelope-Based Test of LNAs

This work demonstrates that multi-V
_DD
conditions may be used to improve the accuracy of machine learning models, significantly decreasing the prediction error. The proposed technique has been successfully applied to a previous alternate test strategy for LNAs based on response envelope detection. A prototype has been developed to show its feasibility. The prototype consists of a low-power 2.4GHz LNA and a simple envelope detector, integrated in a 90 nm CMOS technology. Post-layout simulation results are provided to verify the functionality of the approach.