Multichannel Analog-to-Digital Conversion
Abstract Generated abstract
This paper examines signal switching in multichannel analog-to-digital converters and compares established methods based on switching analog quantities, digital quantities, and control signals. It argues that although analog switching remains justified where accuracy requirements are moderate and equipment simplicity is important, improved accuracy motivates alternative approaches. Two additional methods are proposed: selecting channels by switching sensor supply voltage, which can reduce comparison circuitry relative to control-signal switching, and performing analog functional processing before conversion when the required channel operations permit it. The discussion outlines the equipment, frequency, conversion-time, and error implications of these approaches, while noting practical limits from analog functional converter bandwidth and the difficulty of implementing analog delay lines.
Full Text
Multichannel Analog-to-Digital Conversion
A. V. TRIKHANOV, A. N. OSOKIN
(Presented by the scientific seminar of the Department of Computer Engineering)
In multichannel analog-to-digital converters (ADCs), one of the important problems is signal switching, three methods of which:
1) switching of analog quantities,
2) switching of digital quantities, and
3) switching of control signals
are described in [1, 3].
The last two methods, unlike the first, are characterized, on the one hand, by the absence of quantization error (the error of a multichannel converter is equal to the error of one ADC), and, on the other hand, by an increased amount of equipment. In the case of \(n\) channels, \(n\) ADCs are required under the second approach, or \(n\) circuits for comparing the analog input quantity with the feedback voltage of the ADC under the third approach.
The operating frequency of the ADC under the second method is \(n\) times lower than under the first and third methods.
As for the conversion time, it may be the same for all switching methods. In this case the total error of the multichannel system will be greater under the first method because of the error of the analog switch. The total error can be reduced by increasing the speed of the conversion.
A comparison of the three known methods shows that, under conditions where high total accuracy is not required and complication of the system is undesirable, the broad use of the first method, which has taken place up to the present time, is fully justified [2].
However, the development of science and technology requires a reduction of the total error, as well as improvement of other characteristics of multichannel ADCs. The solution of these questions should be sought not only within the framework of existing methods. The development of new switching methods is very important; two of them are described below.
In addition to the three signals indicated above, the switching of which leads to the switching of analog signals, one may add the supply voltage of the sensors from which the analog signals are taken. Switching the supply voltage to a specified sensor ensures the selection of the required channel of the entire system. In this system
includes an ADC, a switch for discrete quantities (sensor supply), and a summing circuit (Fig. 1).
The switch applies the supply voltage to only one sensor. The output voltage will be only at the selected sensor. After passing through the summing circuit, it arrives at the ADC input. In a certain sequence all sensors are interrogated, and the ADC performs conversion in each channel. In its basic characteristics, this method coincides with the control-signal method and differs from the latter in requiring less equipment: instead of \(n\), only one comparison circuit is needed.
Figure labels: \(A1\), \(Ai\), \(An\); ADC; control; supply voltage.
Fig. 1. System with switching of the supply voltage
From our point of view, the four methods considered are the principal ones and constitute the complete group of switching methods with scanning of channels in time.
A new approach to solving the switching problem appears when this problem is considered jointly with the algorithm for processing the channel signals. In this case the problem of switching analog signals is carried out before conversion. It is clear that the system (Fig. 2) must include an analog multi-input functional converter.
For example, if the processing reduces to summation, then the functional converter is a summing unit. The signals from the sensors must be fed to the inputs of the functional converter with a certain delay. At the output of the functional converter there will be a continuous sequence of the results of the corresponding processing, which, after a time equal to the delay of the longest delay line (DL), will be repeated. The operating frequency of the ADC is equal to the repetition frequency of the indicated sequence.
The total error of operation of the entire system under this approach is composed of the errors of the DL operation, \(\sigma_{\mathrm{d}}\), the ADC error, \(\sigma_{\mathrm{p}}\), and the error of the functional converter, \(\sigma_{\mathrm{fp}}\); it is given
larger number of components. The error \(\sigma_{\mathrm{fp}}\) is greater than the error of the corresponding processing in the arithmetic unit (AU) of an electronic digital computer. However, if one takes into account the error with which the input signals arrive—and this error is large—then it turns out that the high accuracy of processing in the AU of the electronic digital computer is not used, and that a reduction in the accuracy of this processing is entirely permissible.
Fig. 2. System with processing of analog signals before analog-to-digital conversion
However, it should be borne in mind that analog FPs currently have a small bandwidth; consequently, rapidly varying signals are difficult to process by this method. It is difficult to obtain delay lines for analog signals. The specific solution to the question of applying this method depends on the type of functional transformation, on the required accuracy, and on the characteristics of the signals.
REFERENCES
-
E. I. Gitis. Information Converters for Electronic Digital Computing Devices. Moscow, Energiya, 1970.
-
G. I. Dovarashvili. Investigation of Multichannel Voltage-to-Code Converters. Abstract of Candidate’s Dissertation. Georgian Polytechnic Institute, Tbilisi, 1970.
-
F. E. Temnikov. Theory of Sweeping Systems. Moscow, Gosenergoizdat, 1963.