Millimeter-wave small-signal modeling with optimizing sensitive-parameters for metamorphic high electron mobility transistors

Abstract

In this paper, we present a simple and reliable technique for determining the small-signal equivalent circuit model parameters of the 0.1 μm metamorphic high electron mobility transistors (MHEMTs) in a millimeter-wave frequency range. The initial eight extrinsic parameters of the MHEMT are extracted using two S-parameter (scattering parameter) sets measured under the pinched-off and zero-biased cold field-effect transistor conditions by avoiding the forward gate biasing. Furthermore, highly calibration-sensitive values of the R s , L s and C pd are optimized by using a gradient optimization method to improve the modeling accuracy. The accuracy enhancement of this procedure is successfully verified with an excellent correlation between the measured and calculated S-parameters up to 65 GHz.

A directional method of extrinsic resistance determination for millimeter‐wave small‐signal model

“…However, these methods do not offer fully reliable extractions in the entire frequency range particularly containing W‐band frequencies due to various problems associated with oscillating and nonscalable extracted parameters. Figure illustrates the measured real parts of Z 12 and Z 22 versus frequency of our 2 × 20 µm (20 µm gate width with two gate fingers) metamorphic HEMTs (MHEMTs), and appearances of these parameter oscillations were shown for both extrinsic and intrinsic parameter extractions in many research reports . Due to the oscillations and unexpected increase in the real part of Z 22 at high frequencies, Brady's method gives unreliably negative values for the calculation of channel resistance, R ch , from the real part of Z 22 ( f L )– Z 22 ( f H ) under zero‐biased condition, where f L and f H represent the lowest frequency (0.5 GHz) and the highest frequency (110 GHz), respectively.…”

Accuracy improvement of small‐signal model minimising parameter sensitivity based on de‐embedded sub‐model method at W‐band

“…Furthermore, we propose another extraction strategy modulating the sensitivity of L s based on a directional search technique to enhance the eventual model accuracy. This differentiates our proposed extraction strategy from other methods using the optimisation schemes minimising the errors between the model S-parameters and the measurements under zero-biased [11,12] or pinch-off [17] condition. Because the hot field effect transistor (FET) condition can characterise the operating devices most closely, we utilised a strategy minimising the fitting errors of hot FET model S-parameters in determining the L s parameters.…”

“…Notably, they could not offer fully reliable extractions particularly at W-band frequencies due to various problems associated with the oscillating and non-scalable extracted parameters as was discussed in [10]. Especially, appearances of the parameter oscillations were revealed for both extrinsic and intrinsic parameter extractions [5,[11][12][13][14][15][16].…”

The large world of FET small-signal equivalent circuits (invited paper)

“…The most used and powerful approach for extracting the extrinsic elements is based on using S‐parameter measurements performed on the FET under cold condition ( V DS = 0 V, i.e., passive device). Although so far a plethora of cold techniques have been proposed and successfully applied , new developments are persistently required to model FET technologies incessantly progressing. Consequently, the small‐signal equivalent circuit modeling is an evergreen research field in perpetual evolution as novel technologies become available.…”