A 2.4-GHz SiGe BiCMOS MMIC active tunable band-pass filter design using artificial neural network

An MMIC active band-pass filter is designed for the implementation of a tunable RF filter in a 0.35 ¿m SiGe BiCMOS technology. This filter structure is proposed for operation in the S-band applications and BiCMOS MMICs applications for wireless personal communication. In simulations, the reasonably good performance was obtained. This filter is centered at 2.4 GHz and has 30 dB rejections @0.7-GHz offset from pass-band edges. These sharpened edges were caused by a tuned amplifier. The simulations show a frequency tuning range from 1.9 GHz to 2.5 GHz and a noise figure of 7.5-10 dB depending on center frequency. In addition, the design was optimized with artificial neural network (ANN). A feed-forward, four-layer perceptron structure (MLP4) was used.