The aim of our laboratories are the research on electronics circuits for analog signal processing and to establish a circuit design which takes the advantages of transistor characteristics. The technique of analog circuit design is a technique to improve the circuit characteristics by choosing a proper device parameter based on the theory or proper circuit structure with consideration on characteristics of the devices such as transistors. Since the design method concerning physical amounts such as voltage and current is the foundation of various electronics and integrated circuits, we think that circuit design on transistor level is important and we carry out proposal of various analog design techniques based on it.
In the field of analog circuits design, not only the given specifications but explicit physical limitations also need to be considered. Although it is possible to match up a wide variety of specification terms by smartly utilizing the design flexibility, it is necessary to consider the circuit operation from a wide point of view. In order to promote better understandings, the graduated students are free to choose their research topics, proceed their research in a bright note, present their research progress such as new circuit's implementations, theories and ideas in front of laboratories members and receive guidances and opinions. As is simulating a circuit without knowing how it works will not give a reliable result, not only accumulating knowledges but each students learn the basic stance in circuit design through their own research.
Based on the demand of new structures, theories and designs of electronic circuits and the design specialization for various specification, the topics of our research can be divided into the followed cathegory.
Improvement of Circuit Block Performance
The aim of this topic is designing a relatively complicated circuits such as filters and PLL by improving the performance of basic circuit blocks such as operational amplifier. Specifically, improving the performance of filter by designing a low-voltage, low-power OTA with high linearity and wide bandwidth. In addition, the problem in discrete time analog signal processing such as the reduction of clock feedthrough on switched-current circuits is also considered.
Development of New Signal Processing Circuits
We are working on the design techniques of variable gain amplifiers, low-noise transimpedance amplifiers for optical communication and also filters for WCDMA baseband, video signal, harddisk and IF band. We have successfuly developed a current-mode filter that works on frequency band of a few hundreds MHz. In either design, it is not only considering the performance improvement of each circuit block, but also the efforts in each design level such as design simplification on transistor level implementation are required. Furthermore, we also make an effort to improve the performance of a PLL such as accuracy and reduction of power dissipation by introducing an automatic frequency tuning with switched capacitor technique.
In rush of the trend for low-voltage mixed-signal integrated circuits, the design of low-voltage analog circuits is not as easy as the digital ones. We investigate the implementation of circuits with linear relation between its input and output using non-linear circuits which operates at low power supply voltage. Companding technique is recently known as of the implementation although there are still a lot of practical problems under investigation. As one of the low-voltage design technique, we proposed a signal decomposition technique while there is no practical and effective solution to this problem. This signal processing technique is utilizing the signal decomposition using unused nodes in a circuit, and therefore it is expected to be able to handle a large signal voltage without increasing noise significantly. As an example, with a power supply of 1.5V, this method aim for the low noise signal processing with internal signal higher than 1.5Vp-p. Recently, it is confirmed that the amplifier and filter implemented using this method work properly even if noise and offset are added into the input signal.