Microwave power device characterization

Doctoral thesis, 2006

The first part of the thesis covers work done on device characterization methods. A statistical method for estimating small-signal model parameters in FET-models was proposed. A maximum likelihood estimator was derived and the new method was compared to a standard direct extraction technique. The comparison showed that the proposed method gave parameter estimates that were less uncertain than the direct method. A method for extracting the thermal impedance of microwave transistors was developed. The method was derived from a thorough theoretical analysis of the self-heating feedback problem. The method uses small-signal measurements at low-frequency and a temperature controlled fixture. A technique for improving dynamic range in oscilloscope based RF measurements was also presented. The technique uses repeated measurements synchronized at baseband and an extended Kalman filter for estimating the unknown RF-phase, which allowed for averaging and thus reduction of measurement noise. The technique was then used in an error-corrected source-pull setup. The error-correction takes in to account group-delay variations over the measurement bandwidth. The second part of the thesis contains experimental results on mixer circuits fabricated using wide bandgap semiconductor devices. Hybrid resistive FET mixers were fabricated for S- and C-band operation. Both SiC-MESFETs and AlGaN/GaN-HEMTs were evaluated as mixing elements. The best performance was achieved with an AlGaN/GaN-HEMT, with a minimum conversion loss of 7 dB and a maximum third-order intercept point of 36 dBm. A monolithic integrated double balanced Schottky diode ring mixer was also designed. The mixer was fabricated in Chalmers in-house SiC-MMIC process. The mixer had a minimum conversion loss of 12 dB and a maximum third order intercept of 38 dBm.