On Impedance-Pattern Selection for Noise Parameter Measurement

Several signal-source impedances (admittances) are required to fully characterize the noise behavior of a linear device. This paper expands the theory of choosing admittances required for noise parameter extraction by finding sets that are guaranteed to form systems of linearly independent equations. A minimal set of four admittances is chosen from four linearly independent admittance regions. Most prior methods require a least-squares solution to noise parameters, using a redundant number of admittances. The proposed method employs a direct solution to the noise parameters using the four admittances. The proposed method also allows adapting the four admittances to overcome signal-source admittance-tuner frequency limitations and/or reducing uncertainty in the minimum noise factor extraction provided approximate knowledge of optimal admittance for minimum noise. The measurement and simulation results demonstrated that the adaptable selection criterion extracted noise parameters well within 3σ uncertainties of noise parameters found with patterns previously reported in literature. Measurements and theory in this work demonstrate that, in general, absolute reflection coefficients of source admittances do not need to exceed approximately 0.4 but should also be less than 0.9. The flexibility of selecting the signal-source admittances has immediate advantages in accurately determining noise parameters of conditionally stable amplifiers, low-frequency devices operating beyond the tuner specifications, devices operating at high frequencies where tuner losses are high, minimizing measurement time, and cases where available admittances do not encompass regions required by other methods.