A simple and efficient procedure for modeling of scattering and noise parameters for a class of microwave transistors manufactured in the same technology is presented in this article. It is based on multilayer perceptron artificial neural networks (ANN), whose inputs are device gate width, biases, and frequency that produce scattering and noise parameters at their outputs. After the ANN training, the scattering and noise parameters' prediction under different operating conditions for any device from the class requires only calculation of the ANN response, without changes in the ANN structure. Numerical examples for S-and noise parameters modeling for one specific series of pHEMT devices are presented to show the validity and effectiveness of this approach. Figure 5 Scattering parameters for bias points not used for the training process: (a) S11 black symbols, ATF35143 (2 V, 15 mA); white symbols, TF34143 (3 V, 20 mA); solid lines, neural model output; (b) S12 black symbols, ATF35143 (2 V, 15 mA); white symbols, ATF34143 (3 V, 20 mA); solid lines, neural model output; (c) S21 black symbols, ATF35143 (2 V, 15 mA); white symbols, ATF34143 (3 V, 20 mA); solid lines, neural model output; (d) S22 black symbols, ATF35143 (2 V, 15 mA); white symbols, ATF34143 (3 V, 20 mA); solid lines, neural model output ABSTRACT: An on-chip implementation of a low loss 1:9 transmission line transformer (TLT) is presented. The TLT was fabricated with broadside coupled microstrip lines in a GaAs HBT process. The wide metal transmission lines allow the high-quality factor and the high-current capability. The measured insertion loss is 0.88 dB at 880 MHz. The