Kurganskiy Michael
Faculty of Computer Science
Major: “Computer Systems and Networks“
Dissertation topic:
«Research of analog-to-digital transformers of firm Altera»
Scientific adviser: Malcheva Raisa Victorovna, Ph.D.
Faculty of Computer Science
Major: “Computer Systems and Networks“
Dissertation topic:
Scientific adviser: Malcheva Raisa Victorovna, Ph.D.
The relevance of the topic. In the evolutionary schedule it is necessary to consider as the first control systems in which as a controlling link the person appeared. Disadvantages of similar systems are obvious: the person can do errors, and the quantity of committed errors grows in connection with weariness of the person. Mechanical regulators partially resolved a problem but construction of similar regulators for difficult systems is labour-intensive process and not always possible. Appearance of analogue regulators has allowed to lift managerial process on qualitatively new level. However analogue control systems are not deprived disadvantages among which main consider impossibility of construction of control systems as difficult objects.
Advantages of digital treatment above an analog. At the present stage of development of engineering of distribution have gained digital control systems (DCS). Considerable computing and logical possibilities of digital devices define their usage for handle of the automated objects. The devices of digital processing fulfilled on the basis of modern digital microelectronics have serious advantages in front of analogue devices.
The offered advantages of digital control systems have caused their rapid development and the usage extension. Especially it has appeared in connection with appearance of specialised microcontrollers which connect almost all digital DCS sites in one tank. If earlier DCS were expensive and were used for handle of difficult objects and processes microcontrolled systems, thanks to low cost, can be used for handle of simple devices, thus possibilities of devices extend without cost substantial growth.
The common structure DCS is represented on fig. 1. The Mainframe of the given system is the controlling computer (KOM). The information on a current object state of handle DCS receives from sensors (analogue and digital), makes behind certain algorithm controlling influences which arrive on executive mechanisms. Information interchange between the calculator (processor) and the handle object is carried out through intercouplers with object (IO) - the complex technical and software which provide information interchange between the processor and the handle object. Depending on type of a signal and a direction of information transfer IO are divided into four subsystems:
Construction of a subsystem of digital input-output does not present especial difficulties because the KOM by the nature is digital and the intrinsic error consists in definition of digital entry influences available a period between change of a digital signal and response of system to this change.
The subsystem analogue is intended input for introduction for the KOM of the analogue information. Thus there is a quantization of analogue signals, both behind level, and on time that leads to appearance of additional errors. Therefore at construction DCS it is necessary to adhere to two main principles:
At implementation of algorithms of the control and handle on KOM it is necessary to consider such circumstances:
KOM at calculations capable to fulfil only arithmetic and logic operations which log on commands of the given computer;
All calculations and operations to information input-output are fulfilled in a definite time, caused by speed of functional devices the KOM that does impossible incessant information interchange between the KOM and its abonents (information sensors, executive mechanisms, mapping devices). It results before that incessant time in which processes in the handle object digitization on inputs and outputs the KOM proceed;
The entry information the KOM should be presented numerically, in the form of numbers of the limited digit capacity;
Results of the arithmetic operations, which digit capacity digit capacity of registers of the arithmetic device exceeds the KOM, should be rounded off;
Constants which enter into computing algorithms and are used in the course of calculations, at their binary is finite-digit representation can differ from preset values.
Errors DCS at the analysis are expedient for subdividing depending on factors that them generate, and features of manifestation on methodological, transformed, tool and methodical [5].
Methodological errors are caused by discrepancy between expedient behaviour of system as a result of influence of external indignations and behaviour of system which is provided with the mathematical process model assumed as a basis at designing of managing directors of system. These errors characterise a degree of perfection of the theory, instead of quality DCS.
Transformed errors result from transformation of entry errors of arguments (in some sources as a transformed error understand purely an error of entry arguments). The error value depends on sort of function and errors of the entry information.
Methodical errors are errors of the numerical methods accepted in KOM for calculation of controlling influences. Errors of numerical methods do not depend on characteristics the KOM and can be in each specific case calculated precisely enough.
Tool errors are caused by a finite number of the bits intended for representation of values, and necessity of roundoff of results of some elementary operations.
Especially it is necessary to select a dynamic error. The dynamic is understood as an error which results from delay of solution of the task which is caused by a managerial process continuity. The dynamic error depends on the period of digitization of entry analogue signals and type of the object of handle. For the analysis of dynamic errors the mathematical apparatus of z-transformations [3] is often used.
Ways of an estimation of methodical and tool errors for digital computers are in details enough considered in Zhuravlyov's books J. P [5], Sorenkova E.I. [13], and also in Zhelnova J.A.'s robot [4]. However ways of an estimation of influence of a transformed error on accuracy of framing of controlling influences in the given sources are shined insufficiently and often as a source of a transformed error consider exclusively noise of quantization ADC, disregarding that the signal on input ADC arrives with a margin error and ADC is not ideal [4 13]. From these reasons consideration of a question on an estimation of a transformed error is necessary.
The analysis of sources of a transformed error the Source of a transformed error on output DCS is an error of definition of entry influences. The given error on the structure is enough difficult. Generally it includes following units:
State information of the object of handle DCS receives from sensors of physical values. the sensor is a device that, coming under influence of the physical measured value, produces an equivalent signal, mainly electrical nature which is function of measured [8] value. Accuracy of a state information of the object of handle is defined by accuracy of sensors. It speaks about importance of a correct choice of the sensor. For more parts of sensors dependence of the initial value from measured is not linear. In this case for definition of value of the measured value on value of an initial signal of the sensor following methods are used:
Last from these methods it is used often enough, thus the linearization circuit can be both external concerning the sensor, and internal. In engineering specifications parametres are specified in sensors in the form of generalising characteristics to accuracy:
In a picture 2 the schedule of dependence of the initial value of the sensor from entry influence for sensors with the internal circuit of linearization [18] is directed.
The analysis of the resulted dependence shows that error amount of the sensor has strictly natural (not casual) character and depends on the measured value and a direction of change of this value. Exact character of this dependence is not given in the documentation and parametre CNL&H (combined nonlinearity and a hysteresis) which specifies maximum deviation of the real characteristic of the sensor from the ideal is used. For this reason at modelling of the sensor the error either is not considered, or approximated by close function.
The information from sensors arrives on input DCS through communication circuits, there is a negotiation of levels, switching of analogue signals and analogue-digital transformation. At this stage the error is brought by analogue units, and also obstacles in communication circuits are added. For reduction of this making entry error circuit and program solutions are used.
Variants circuit solutions are analogue filters, differential amplifiers with the big coefficient of reduction of inphase obstacles. Efficiency of such solutions is high, their disadvantage is substantial growth of cost DCS. It is necessary to mark that the error value appreciably depends on the accepted structure of a subsystem analogue to input. The most rational from the point of view of lowering of obstacles is the structure offered in [3]. According to this structures ADC and the negotiation circuit concern in immediate proximity from the sensor and data transfer on input DCS is carried out numerically. Advantage of such system is high noise immunity and possibility of an information transfer from sensors on the big distance without twistings (numerically), disadvantage of similar system is high cost.
To program ways lowerings of obstacles it is necessary to carry digital filters and software of correction of entry influences. The given ways are cheaper in comparison with equipment rooms because writing of a fragment of the program suffices for their implementation in DCS.
Quantity of sensors with a digital output among total of sensors small (the application of digital sensors is considered in [9]). Therefore to one of mandatory units of a subsystem analogue to input remains ADC. The main characteristics ADC is digit capacity and time of transformation and also accuracy of the characteristic. Cost ADC increases with digit capacity and speed increase. If to take ADC identical digit capacity and speed their cost depends on accuracy of parametres. An example of such dependence for 8-digit ADC a general purpose (transformation time makes 100mcs) it is represented on a figure 3.
The resulted dependence speaks about necessity at choice ADC to consider not only its digit capacity, but also others of accuracy of the characteristic.
Conclusion. The analysis of sources of a transformed error shows:
Important notice! This work is still in progress. ETA is December 2009. Feel free to contact the author or scientific adviser for more information.