Any device capable of signal processing can be called a filter. Filters are widely used in modern telecommunication equipment and various control systems. Among all electronic components, the filter is the most used and the most complicated technology. The quality of the filter directly determines the quality of the product, so the research and production of the filter has been paid attention to by all countries.
-19 17 American and German scientists invented LC filters respectively, which led to the first multiplexing system in the United States the following year. In 1950s, passive filters became more and more mature. Since 1960s, due to the development of computer technology, integration technology and material industry, filters have developed to a new level, with low power consumption, high precision, small size, multifunction, stability and reliability, and low price, among which small size, multifunction, high precision, stability and reliability have become the main directions since 1970s. This leads to the rapid development of various filters, such as RC active filter, digital filter, switched capacitor filter and charge transfer device. By the end of 1970s, the monolithic integration of these filters had been developed and applied. In 1980s, we devoted ourselves to the research of various new filters, striving to improve their performance and gradually expanding their application scope. Since the 1990s, we have been mainly engaged in the development and research of various filters for various products. Of course, the research on the filter itself continues.
-Filters were widely used in China in the late 1950s, when they were mainly used for voice filtering and report filtering. After half a century's development, the development, production and application of filters in China have been in line with international standards. However, due to the lack of specialized research institutions, the integration technology and material industry can't keep up, which makes the research and development and application of many new filters in China lag behind the international development.
Classification of filters
-Filters can be classified in various ways, generally as follows. (1) Classification by processing signal type-Classification by processing signal type can be divided into two categories: analog filter and discrete filter. Analog filters can be divided into three categories: active, passive and heterogeneous. Discrete filters can be divided into three categories: digital, sampled analog and mixed. Of course, each classification can continue to be divided. In short, their classification can form a tree structure, as shown in the figure.
-Actually, some filters are difficult to classify. For example, switched capacitor filters can belong to sampled analog filters, hybrid filters and active filters. So we don't need to demand this "accurate" classification, just let people know the general types of filters and have a general concept. (2) Classification by selected physical quantities.
According to the selected physical quantities, filters can be divided into four types: frequency selection, amplitude selection, time selection (such as channel signals in PCM systems) and information selection (such as matched filters). (3) Classification by frequency passband range
According to the frequency band-pass range, filters can be divided into five categories: low-pass, Qualcomm, band-pass, band-stop and all-pass, while comb filters belong to band-pass and band-stop filters because of their periodic band-pass and band-stop.
-There are many kinds of filters, some of which are well known and some of which may be unfamiliar. The following focuses on several filters that have developed rapidly in recent years.
active filter
-The active filter consists of the following active components: operational amplifier, negative resistor, negative capacitor, negative inductor, frequency rheostat (FDNR), generalized impedance converter (GIC), negative impedance converter (NIC), positive impedance converter (PIC), negative impedance inverter (NII), positive impedance inverter (PII), four controlled sources, in addition.
-1965 monolithic integrated operational amplifier has opened up a broad prospect for active filters. At the beginning of 1970s, active filters developed remarkably, and 1978 monolithic RC active filter came out, which made a gratifying step for filter integration. Because the gain and phase shift of operational amplifier are functions of frequency, the frequency range of RC active filter is limited, and the general working frequency is about 20kHz. After compensation, the working frequency is also limited to 100kHz. 1974 has made an RC active filter with higher frequency, which makes the working frequency reach GB/4(GB is the product of operational amplifier gain and bandwidth). Because of the existence of R, the integration process is difficult, so the active C filter appears again: that is, the filter consists of C and operational amplifier. This is easy to integrate, and more importantly, it improves the accuracy of the filter, because the performance of the active C filter only depends on the ratio of capacitors, and has nothing to do with the absolute value of capacitors. However, it has a main problem: because all branch components are capacitors, there is no DC feedback channel in the operational amplifier, which makes stability a difficult problem. In 1982, Geiger, Allen and Ngo proposed to replace the resistor R in the active RC filter with the continuous switching resistor (SR), thus forming the SRC filter, which still belongs to the analog filter. However, due to the use of preset circuit and multi-phase clock, the development prospect of this filter is not great.
-In short, various variants of active filters based on RC are small in size, and the Q value can reach 1000, which overcomes the shortcomings of RLC passive filters. However, there are still many topics that need further study: the study of the deviation between ideal operational amplifier and actual characteristics; Due to the continuous improvement of hybrid integration technology of active filter, monolithic integration needs further study; It is necessary to continue to explore the filter with the least active components by using linear transformation method; The existence of absolute value tolerance of components affects the accuracy and performance of the filter, and other problems remain unsolved. Because of the existence of R, the integration occupies a large chip area, the resistance error is large (20% ~ 30%) and the linearity is poor, which makes large-scale integration still difficult. Although there are many problems, the theory and application of RC active filter are still developing.
Switched capacitor filter (SCF)
-One of the main tasks of technological transformation in the 1980s was to realize the integrated large-scale integration (LSI) of various electronic systems. The most used filters become "obstacles". RC active filters can't realize LSI, let alone passive filters and mechanical filters, so people can only find another way. The concept of SCF was put forward in 1950s, but the integration process at that time was not enough, which did not attract people's attention. 1972, an American scientist named Fried published an article simulating resistance r with switches and capacitors, saying that the performance of SCF only depends on the ratio of capacitors, and has nothing to do with the absolute value of capacitors, which attracted people's attention. From 65438 to 0979, single-chip computers in some developed countries have become a commodity (a highly confidential technology). Now SC technology is mature. SCF is realized by MOS technology and is recognized as a major breakthrough in network theory and integration technology in 1980s. At present, the value of MOS capacitance is generally within a few picofarads to 100pF, and the voltage coefficient is (10 ~100) ×10-6/v, and the voltage coefficient is (10 ~100). SCF has the following advantages: SCF can be integrated on a large scale; The accuracy of SCF is very high, because its performance depends on the matching of capacitors, and the matching error of MOS capacitors is less than one thousandth. Multifunctional, almost all electronic components and functions can be realized by SC technology; It is simpler than digital filter, because there is no need for A/D and D/A conversion; Small function, which can be less than 10mW.
The application of SCF is mainly in the audio frequency range, and its working frequency is within 100kHz. Applications in signal processing include: program-controlled SCF, analog signal processing, vibration analysis, adaptive filter, music synthesis, * * vibration spectrum, language synthesizer, tone selection, speech coding, audio analysis, equalizer, demodulator, phase-locked circuit, discrete Fourier transform ... In short, SCF has broad application prospects in instrument measurement, medical instruments, data or information processing and other fields.
-In China, some tutors and graduate students started this research on 1978, which really attracted people's attention after 1980. 1983, Tsinghua University made a monolithic SCF, and Chengdu University of Technology and the factory also developed a monolithic SCF. At present, the key is to realize SCF with MOS technology and popularize it. Because users don't know yet, the application of SCF in China has not been popularized.
-SCF has many topics to study:
① Due to the limitation of sampling frequency of operational amplifier and MOS switch, SCF can only be used in audio frequency range. In recent years, although there are SC circuits without operational amplifiers, due to the limitation of sampling frequency, the highest working frequency is only within 1MHz. ② The channel resistance of non-MOS switch and non-ideal operational amplifier characteristics will cause SCF errors.
③ The parasitic capacitance of the switched capacitor itself will distort the frequency response of SCF.
④ Thermal noise of MOS switch and MOS operational amplifier limits the dynamic range of SCF.
⑤ The SCF to be finally realized by MOS technology is a time-varying network, and it is impossible to accurately simulate it with discrete components. In this way, well-designed CAD technology is the only means to solve this problem. In addition, sensitivity analysis and noise analysis also have many topics to study.
Several New Digital Filters (DF)
-You are no stranger to DF. There is no system overview here, but some new DF will be introduced. (1) adaptive direction finding
Optimal control, adaptive control and self-learning control all involve multi-parameter and multi-variable complex control systems, and they all belong to the research topic of modern control theory. Adaptive DF has strong self-learning and self-tracking functions. It has been widely used in radar and sonar beamforming, suppression of slowly varying noise interference, noise signal processing, adaptive equalization of communication channels, echo cancellation of long-distance calls and other fields, which has promoted the development of modern control theory.
-Adaptive direction finding has the following simple algorithms: W-LMS algorithm, M-LMS algorithm, TDO algorithm, differential LMS algorithm and C-LMS algorithm. (2) Complex DF
-In signal processing systems where the input signal is narrowband, complex direction finding techniques are usually used. In order to reduce the sampling rate and keep all the information contained in the signal, the orthogonal bidirectional detection method can be used to extract the complex envelope of the narrowband signal, and then the complex envelope can be converted into a complex sequence through A/D conversion for processing. This signal processing system is called multiple direction finding. It has many functions: suppressing the clutter interference of Doppler frequency shift in MTI radar; Multiplexing of TDM/FDM signals between digital communication network and analog communication network ...
(3) Multidimensional DF
-Multidimensional direction finding (usually 2D direction finding) is used for image processing, seismic data processing and oil exploration data processing. In the design of multidimensional DF, the optimal design of one-dimensional DF is often directly extended to multidimensional DF. Multidimensional DF can also play a good role in processing two-dimensional images disturbed by fuzzy and random noise.
-In addition, there is wave DF, which is convenient for large-scale integration and digital simulation of passive and active filtering networks. Therefore, it is attracting people's attention and research.
-For DF, the topics to be studied include: coefficient sensitivity, rounding noise and limit cycle, stability of multidimensional regression filter, research on various hardware and software to realize DF, etc. In a word, direction finding plays an extremely important role in digital signal processing technology and is of great significance to its research, production and application.
Other new filters
-In order to meet various needs, many new filters have appeared. Here are some new filters that have been widely used. (1) Electronic control programmable CCD transverse filter (FPCCDTF)
-Charge Coupler (CCD) Fixed Weighted Transverse Filter (TF) In signal processing, its performance and cost are comparable to those of digital filters and various signal processing elements. This kind of filter is mainly used for adaptive filtering; Matching filtering of P-N sequence and Chirp waveform; Universal frequency domain filter and correlation product operation; Speech signal and phase equalization; Beam synthesis of phased array system and ghost elimination of TV signal have applications. Of course, more applications need to be further developed. In a word, FPCCDTF is the most promising development direction. (2) Crystal filter
It is developed to adapt to the single sideband technology. In 1970s, the appearance of integrated crystal filter made its development leap. In recent ten years, crystal filters have been devoted to the following research: to achieve the best design, with excellent selection and good time domain response; Seeking new materials; Expand the working frequency; Transform the process and make it develop towards integration. Widely used as carrier filter in multiplexing system, frequency selective filter in transceiver, IF filter in single sideband communication machine, IF filter in spectrum analyzer and sonar device. (3) Acoustic surface filter
-This is an ideal UHF device. Its amplitude-frequency characteristics and phase characteristics can be controlled separately to meet the requirements, and it also has the characteristics of small volume, good long-term stability and simple process. It is usually used as a residual sideband filter in TV broadcast transmitter; Surface comb filter for tuning system of color TV receiver. In addition, it is also widely used in national defense satellite communication systems. Acoustic surface filter is the product of the combination of electronics and acoustics, which can be integrated, so it is the most promising of all passive filters.
-Too many new filters. Limited to space, I won't describe them one by one.
At present, the application ratio of various filters in China
-The types and frequencies of existing filters in China have basically met all kinds of existing telecommunications equipment. Generally speaking, the development of active filters in China is slower than that of passive filters, and there is no mass production and application. From the following production and application ratio, we can see the application of various filters in China: LC filter accounts for 50%; Crystal filter accounts for 20%; Mechanical filter accounts for15%; Ceramic and acoustic surface filter account for 65438 0% respectively; Other filters * * * account for 13%. Judging from these application proportions, the integration of filters is still an important topic if China's electronic products want to achieve large-scale integration.
-With the development of electronic industry, the performance requirements of filters are getting higher and higher, and more and more functions are required, which requires the development towards integration. The development and production of filters in China are far from the above requirements. In order to shorten this gap, electronic engineers and technicians have a great historical responsibility.