ENCYCLOPEDIA OF RADIO ELECTRONICS AND ELECTRICAL ENGINEERING Universal analog programmable ICs: selection of elementary functional units. Reference data Encyclopedia of radio electronics and electrical engineering / Application of microcircuits It is difficult to overestimate the importance of reprogrammable logic integrated circuits (FPGA) in the synthesis of logical systems. The integrated development of the element base and computer-aided design systems makes it possible to implement complex logical systems in an unprecedentedly short time and with minimal material costs. Therefore, the desire to achieve similar results in the design and production of analog systems is quite understandable. However, many attempts made in this direction have not yet brought the expected results, and programmable analog ICs (PAIS) and matrix analog LSIs (MABIS) have not yet become universal. Problems of designing programmable analog LSI The rapid progress in the field of designing logical systems on FPGAs was predetermined by the fact that all logical systems are based on a well-developed mathematical apparatus of Boole algebra. This theory makes it possible to prove that the construction of an arbitrary logical function is possible by the ordered composition of only one elementary operator - the logical AND-NOT (or OR-NOT). That is, any strictly logical system can be designed from elements of only one type, for example, NAND. The situation is quite different in the field of design (synthesis) and analysis (decomposition) of circuit diagrams of analog systems. In analog electronics, there is still no single universally recognized mathematical apparatus that would allow solving problems of analysis and synthesis from a unified methodological position. The reasons for this phenomenon should be sought in the history of the development of analog electronics. In the early stages, the circuitry of analog devices developed in accordance with the concepts of the functional-nodal method, the main idea of which was the division of complex circuit diagrams into nodes. A node consists of a group of elements and performs a well-defined function. When combined, the nodes form blocks, boards, cabinets, mechanisms - i.e. some unified constructions, which are called devices. The combination of devices forms a system. The functional-nodal method assumed that the elementary components of systems should be nodes, the main task of which is to perform a well-defined function. That is why functionality, that is, the fact that a node performs some function, was taken as a criterion for classifying nodes. However, as electronics developed, there were an extremely large number of isolated and isolated functions (and, consequently, nodes). Any possibility of their minimization and unification, which is necessary for the synthesis of complex systems, has disappeared. That is why the development of matrix analog LSI (MABIS) and reprogrammable analog integrated circuits (PAIS) has been slowed down and continues to be slowed down. The state of affairs in the field of programmable analog circuits can be traced by analyzing the developments of leading Russian and foreign companies. Thus, the specialists of OAO NIITT and the Angstrem plant concentrated their efforts on the development and production of analog-digital BMCs (basic matrix crystals) of the Rul type H5515KhT1, N5515KhT101, designed for data acquisition, monitoring and control systems, for medical equipment and control measuring equipment [1]. The design of these BMCs includes an analog and digital matrix. The digital matrix contains 115 digital base cells (230 2I-NOT gates), which are arranged in five rows of 23 cells in a row. The analog matrix combines 18 analog base cells arranged in two rows of 9 cells. Between the rows of analog cells are two rows of capacitors (nominal 17,8 pF) and two rows of diffusion resistors (24,8 kOhm each). Between the analog and digital parts is a row of 3,2 kΩ resistors. The BMC provides two types of analog cells (A and B). Type A cells consist of 12 npn and 38 pnp insulated collector transistors and XNUMX multi-tap diffusion resistors. In Type B cells, the four NPN transistors are replaced by two pMOS transistors. Peripheral cells of type A and B contain four powerful npn transistors (in cells of type B - with an isolated collector) and two bipolar transistors. Digital base cells are represented by three types - four n-MOS transistors, four p-MOS transistors and a complementary pair of bipolar transistors. In addition, powerful digital cells are located on the periphery of the crystal, which contain four powerful n-MOS and p-MOS transistors, as well as two npn-transistors connected according to the Darlington circuit. For BMC, libraries of standard analog and digital elements have been developed, which greatly facilitate and speed up the process of designing devices based on BMC. These and similar BMCs contain unconnected sets of electrical radio elements (ERE), from which a number of functional units specified in the library can be obtained. The main disadvantage of such microcircuits is a very narrow scope, limited by the specific values of the ratings and other characteristics of the ERE in this set. The capabilities of the functional units developed and recommended for this set are given in the library accompanying the microcircuit.
Since 2000, Lattice Semiconductor has been producing programmable analog integrated circuits (PAIS) of the ispPAC (In-System Programmable Analog Circuit) family with in-system programming, i.e. without extraction from the printed circuit board [2, 3]. By the middle of 2000, three representatives of this family were being produced: ispPAC-Yu (Fig. 1), ispPAC-20 (Fig. 2) and ispPAC-80. They integrate up to 60 active and passive elements that are configured, modeled and programmed using the PAC-Designer package. The ispPAC PAIS contains: • Serial interface circuits, registers and elements of electrically reprogrammable non-volatile memory (EEPROM) providing matrix configuration;
The architecture embedded in this series is based on basic cells containing: instrumentation amplifier (IU); output amplifier (VU) implemented according to the adder/integrator scheme; reference voltage source 2,5 V (ION); 8-bit DAC with voltage output and dual comparator (CP). Analog inputs and outputs of the cells (except ION) to increase the dynamic range of the processed signals are made according to the differential scheme. Two DUTs and one VU form a macrocell, called a PAC block, in which the outputs of the DUT are connected to the summing inputs of the VU. The ispPAC-10 includes four PACs, and the ispPAC-20 has two. The ispPAC-20 also includes DAC and comparator cells. In the cell, the gain of the DUT is programmed in the range from -10 to +10 with a step of 1, and in the feedback circuit of the VU, the value of the capacitor capacitance (128 possible values) and the on/off resistance. A number of IC manufacturers use "switched capacitor" technology to program analog functions, which involves changing the capacitance of frequency-setting circuits using an electronic switch that switches according to the condition.
Lattice's approach is based on the use of circuits with constant characteristics over time, which can be changed during the process of reconfiguring the system without turning off the power. This improvement is significant, as it eliminates the additional signal processing required in the first method. Internal wiring facilities (Analog Routing Pool) allow you to connect to each other the input contacts of the microcircuit, the inputs and outputs of macrocells, the output of the DAC and the inputs of the comparators. Combining several macrocells, it is possible to build circuits of tunable active filters in the frequency range from 10 to 100 kHz, based on the use of an integrator link. It should be noted that Lattice's ispPACs are closest to PAIS. Their only drawback is that there is no system of universal basic elements that would allow designing not only tunable active filters, but a fairly wide variety of analog systems. It is this circumstance that prevents Lattice Semiconductor's ispPAC from becoming an analogue of FPGAs from companies such as Altera and Xilinx. In general, analyzing the situation in the field of development and practical implementation of analog microcircuits, we can make a number of generalizations: • the bulk of industrially implemented analog microcircuits cannot be classified as LSIs in terms of the degree of integration;
A single basis for the design of FPGAs and MABIS However, the task of developing a unified circuit design basis for the design of analog systems still has a solution, which we will try to theoretically substantiate and show possible directions for the practical implementation of the ideas outlined. First of all, one should choose a mathematical model of a large analog electronic system that would allow one to single out a small group of basic elements. In the field of analysis and synthesis of electronic circuits, there are practically no alternatives to the mathematical apparatus of systems of linear differential equations, which was recognized back in the sixties of the last century [4, 5]. Note, however, that the idea of the practical mass use of this methodology has not yet mastered the minds of all specialists. The system of differential equations consists of elements, their connections and is characterized by a certain structure. The elemental basis of differential equations was studied in the first half of the last century within the framework of the scientific discipline "automatics". In this area, such an advantage of differential equations as unification has manifested itself: their form does not depend on the described process model. However, in the standard form of writing a differential equation, there is no visual information about the nature of the relationships in the system under study. Therefore, methods for visualizing the structure of systems of differential equations in the form of various kinds of schemes were developed throughout the development of the theory of automatic control. By the end of the 60s of the twentieth century, the modern point of view on the structural organization of models of dynamical systems was fully formed [6]. The formation of a mathematical model of the system begins with its division into links and their subsequent description - either analytically in the form of equations relating the input and output values of the link; or graphically in the form of mnemonic diagrams with characteristics. According to the equations or characteristics of individual links, equations or characteristics of the system as a whole are compiled. Links of dynamic systems identified as typical
Note that if for a functional scheme the system is divided into links based on the functions they perform, then for a mathematical description the system is fragmented based on the convenience of obtaining a description. Therefore, the links should be as simple as possible (small). On the other hand, when dividing the system into links, the mathematical description of each link must be compiled without taking into account its connections with other links. This is possible if the links have a direction of action - i.e. transmit action in only one direction, from input to output. Then a change in the state of any link does not affect the state of the previous link. If the condition for the directivity of the action of the links is satisfied, the mathematical description of the entire system can be obtained in the form of a system of independent equations of individual links, supplemented by the equations of connection between them. The most common (typical) are such links as aperiodic, oscillatory, integrating, differentiating, constant delay link [6]. The problem of elementary links in models of the form of a system of differential equations was studied by a number of authors [7-9]. The analysis shows [10] that their positions are mainly reduced to stating the fact of the existence of typical links and studying their role in the process of formation of more complex structures. The selection into the group of typical links is made arbitrarily, without any criteria. Different links are included in the lists of typical ones without explanation and justification, and the terms "simple" and "elementary" are also used equally to designate typical links (see table). Meanwhile, the study of numerous "typical" links of dynamic systems by the methods of structural matrices [10-12] shows that only three links - proportional, integrating and differentiating - do not contain matrix cycles in their structural matrices. Therefore, only they can be called elementary. All other links are built by combining elementary links. So, if a proportional link with a transfer function WB(s) = kB and differentiating link with transfer function WA(s) = kAs connect according to the negative feedback scheme (Fig. 3), then the equivalent transfer function Thus, the result, up to the values of the time constants, coincides with the transfer function of the aperiodic link of the first order. This means that this link can be obtained by connecting the proportional and differentiating links according to the scheme with negative feedback and, therefore, it cannot be considered elementary.
In the same way, you can build the rest of the links included in the table. Particular attention should be paid to the transfer function of the oscillatory link (T2p2 + 2ξTp + 1)y = ku. So, if we connect in series two aperiodic links with transfer functions that differ only in time constants, then the equivalent transfer function will take the form Thus, the result, up to the values of the time constants, coincides with the transfer function of the link under study. Therefore, oscillatory, conservative and aperiodic links of the 2nd order can be obtained by connecting the links of the first order in series. This means that they cannot be considered elementary, although in principle it is permissible to call them typical. An analysis of the results given in the last column of the table allows us to conclude that such links as aperiodic, isodromic, forcing, differentiating inertial and integrating inertial can be obtained by connecting elementary links. To prove that the transfer functions of other typical links can be obtained by connecting elementary links, it would be necessary to analyze the connections of three, four, and so on links according to typical connection schemes. The same result can be obtained if we consider the connections of elementary links with typical first-order links. Part of such a study has already been done, its results are presented in [10]. Thus, it has been proven that by connecting elementary links, it is quite simple to obtain all the transfer functions of the so-called typical dynamic links. Consequently, arbitrary dynamical systems can be synthesized using the multiplication and connection operators of only three elementary links: proportional, differentiating and integrating. This conclusion is of fundamental importance, since it determines the elemental basis necessary for the construction of linear dynamic systems of any order, including radio electronic circuits. And if dynamic systems are supposed to be built from a limited range of dynamic links, as in the case of MABIS and PAIS, then the conclusion drawn is especially important.
It becomes possible to synthesize arbitrary analog devices from only five functional units - a multiplexer, an adder, a multiplier, an integrator and a differentiator (Fig. 4)! Note that those shown in Fig. 4 circuits should not be taken as actually worked out circuit solutions, but only as a justification for the possibility of replacing elementary links on a functional circuit with basic radio-electronic elements. Replacing the elementary links of functional circuits with their hardware counterparts, it is possible to design analog devices with specified characteristics. Analog Device Synthesis Example Consider a very simple example of synthesizing a circuit diagram of an analog device according to a model given by a system of differential equations in the form of Laplace transforms of the form: x0 = g, x1 = x0 - 2x2/s,x2 = 10x1/s,x3 = x2 - 10x4/s,x4 = 500x3/ s.
Let us construct the structural matrix of this system of differential equations and highlight the matrix cycles with arrows:
From the simulation results (Fig. 6) of the synthesized circuit, it can be seen that, with the given parameters, it represents two generators connected in series. That is, a very simple device, consisting of only four integrating links, performs a relatively complex function of modulating a low-frequency oscillation with a high-frequency one. Note that when designing and manufacturing MABIS and PA-IS, it is absolutely not necessary to use hardware analogs of elementary links made on operational amplifiers, as in Fig. 4, although in this basis they are best worked out [13-16]. The most promising is the implementation of hardware analogs of elementary links on optoelectronic components, although any other options are possible.
Universal MABIS and PAIS - it's possible Thus, it is possible to single out five elementary (simplest) components of any REA, corresponding to the main operators of systems of differential equations: multiplication, differentiation, integration, addition and multiplication (multiplexing). The methodology for designing analog electronic devices assumes [10]: • use as initial data for designing a mathematical model in the form of a system of n differential equations of the first order (or a differential equation of the l-th order);
The proposed approach has a number of decisive advantages. Thus, the functional diagram of the designed device is synthesized from the original system of differential equations by standard matrix transformations, which can be ordered and converted into an algorithm for automatic calculations. The electrical circuit diagram is synthesized from the functional circuit by a simple replacement of elementary dynamic links with equivalent basic elements. Also, the modeling of the device using CAD tools can be significantly simplified. Thus, since the set of elementary links is not numerous, there is a real possibility of designing universal MABIS and PAIS. Which, in turn, greatly simplifies the design of analog and digital-analog devices and opens up tempting prospects for the further development of electronics in general. Literature 1. Alenin S., Ivanov V., Polevikov V., Trudnovskaya E. Implementation of specialized analog-to-digital devices based on NIC MOS BMK type H5515KhT1. - ChipNews, 2000, No. 2.
Author: G. Mishin; Publication: cxem.net See other articles Section Application of microcircuits. Read and write useful comments on this article. Latest news of science and technology, new electronics: Machine for thinning flowers in gardens
02.05.2024 Advanced Infrared Microscope
02.05.2024 Air trap for insects
01.05.2024
Other interesting news: ▪ Energy cases for iPhone 6 and iPhone 6s ▪ Lattice Semiconductor Embedded Vision Board ▪ CoolSiC 1200V Silicon Carbide MOSFETs in TO247-3/-4 Package News feed of science and technology, new electronics
Interesting materials of the Free Technical Library: ▪ site section Regulators of current, voltage, power. Article selection ▪ article He was a man. Popular expression ▪ article Can dogs see colors? Detailed answer ▪ Phimosis article. Health care ▪ article Energy-saving photorelay. Encyclopedia of radio electronics and electrical engineering ▪ article Spectators can't guess which hand the coin is in. Focus Secret
Leave your comment on this article: All languages of this page Home page | Library | Articles | Website map | Site Reviews www.diagram.com.ua |