A new mm-wave IC, constructed by flip-chip bonded heterojunction transistors and microstrip lines formed on Si substrate, has been proposed and demonstrated by using MBB (micro bump boding) technology. Millimeter-wave characteristics of the MBB region has been estimated by electro-magnetic field analysis. Good agreements between calculated and measured characteristics of this new IC (named MFIC: millimeter-wave flip-chip IC) have been obtained up to 60 GHz band. Several MFIC amplifiers with their designed performances have been successfully fabricated.

We present ultra low noise- and wide dynamic range performances of an AlGaN/GaN heterostructure FET (HFET). An HFET fabricated on a high quality epitaxial layers grown on a semi-insulating SiC substrate exhibited impressively low minimum noise figure (NF min ) of 0.4 dB with 16 dB associated gain at 2 GHz. The low NF (near NF min ) operation was possible in a wide drain bias voltage range, i.e. from 3 V to 15 V. At the same time, the device showed low distortion character as indicated by the high third order input intercept point (IIP3), +13 dBm. The excellent characteristics are attributed to three major factors: (1) high quality epitaxial layers that realized a high transconductance and very low buffer leakage current; (2) excellent device isolation made by selective thermal oxidation; (3) ultra low gate leakage current realized by Pd based gate. The results demonstrate that the AlGaN/GaN HFET is a strong candidate for front-end LNAs in various mobile communication systems where both the low noise and the wide dynamic range are required.

A technique for suppressing the clipping noise of an analogue-to-digital converter (ADC) is proposed to realize a cognitive radio transceiver that offers high sensitivity carrier-sensing. When a large bandwidth cognitive radio transceiver performs carrier-sensing, it must receive a radio wave that includes many primary user transmissions. The radio wave may have high peak-to-average power ratio (PAPR) and clipping noise may be generated. Clipping noise becomes an obstacle to the achievement of high-sensitivity carrier-sensing. In the proposed technique, the original values of the samples clipped by an ADC are estimated by interpolation. Polynomial spline interpolation to the clipped signal is performed in the first step, and then SINC function interpolation is applied to the spline interpolated signal. The performance was evaluated using the signals with various PAPR. It has been found that suppression performance has a dependency on the number of samples clipped at once rather than on PAPR. Although there is an upper limit for the number of samples clipped at once that can be compensated with high accuracy, about 20 dB suppression of clipping noise was achieved with the medium degree of clipping.