This paper reports on a procedure based on the use of artificial neural networks (ANN) to fully model the performance of advanced high electron mobility transistors (HEMT) operating down to cryogenic temperatures. By means of this procedure, we reproduce the DC behaviour and the scattering (S-) parameters of the device under test (DUT). The I-V curves and the S-parameters of the DUT have been compared with measurements, and a good agreement has been found for assessing the capability of the ANN structure to predict the full behaviour of the DUT. Furthermore, we have analysed in detail the performance of two typical parameters of HEMT's, namely the transconductance and the output conductance. Their values have been derived from measured data and have been compared with those obtained by the ANN approach. Both the simulated DC and RF performance have shown an accuracy degree adequate to model the device properties down to cryogenic temperatures.

Abstract-

In this paper we report the development of an artificial neural network to extract a 17-element smallsignal circuit model of high electron mobility transistors (HEMTs) and one associated noise temperature value. By this procedure, we are able to reproduce the small-signal and noise performance of several device types from only one measured scattering parameter set, one frequency point and one noise figure value. The employed noise figure is measured in input matched conditions (i.e. 50 Ω source impedance), namely F50. The output noise temperature is associated to the drain-source resistance in the HEMT equivalent circuit according to the noise temperature model by Pospieszalski. The noise parameters of the device under test are then calculated by CAD simulation of the circuit and compared with measurement results. The trained network outputs were used by means of a commercial CAD tool, to simulate and fit measurements performed down to cryogenic temperatures with very good agreement. We observed that the difference that occurs between the expected value of the noise temperature and the average value calculated by the neural network leads to negligible variations in the behavior of the simulated noise parameters.

No abstract

No abstract