10.9 DATA CONVERTERS-AN INTRODUCTION In this section we begin the study of another group of analog IC circuits of great importance namely, data converters Digital Processing of Signals Most physical signals, such as those obiained at transducer outputs, exist in analog form Some of the processing required on these signals is most conveniently performed in an analog fashion. For instance, in instrumentation systems itis quite common to use a high ;put-impedance, high-gain, high-CMRR differential ampfifier right at the output of the transducer. This is usually followed by a filter whose object is to eliminate interference However, further signal processing is usually required, which can range from simply ob taining a measurement of signal strength to performing some algebraic manipulations on this and related signals to obtain the value of a particular system parameter of interes, 3s is usually the case in systems intended to provide a complex control function. Another example of signal processing can be found in the common need for transmission of signals to a remote receiver. ‘All such forms of signal processing can be performed by analog means. In previous chapters we encountered circuits for implementing a number of such tasks. However, a lttractive altemative exists: It is to convert, following some initial analog processing, ie signal from analog to digital form and then use economical, accurate, and convenient digi ICs to perform digital signal processing. Such processing can in its simplest form provide us with a measure of the signal strength as an easy-to-read number (consider, for example: the digital voltmeter). In more involved cases the digital signal processor can perform * variety of arithmetic and logic operations that implement a filtering algorithm. The ® sulting digital filter docs many of the same tasks that an analog filter performs—namel), ‘eliminate interference and noise. Yet another example of digital signal processing is foun? in digital communications systems, where signals are transmitted as a sequence of bint pulses, with the obvious advantage that corruption ofthe amplitudes ofthese pulses by nS is, 0 a large extent, of no consequence. Scanned with CamScanner109 DATA CONVERTERS-AN itTROBUCTION analog utp, Such iy the casein telecoms ems where ae eg an may be andible speech. If such an analog out is deste then brio el Pa ve tudy the basic techniques and circuits employed to convert an eae ligtal signal to analog form (digital-to-analog. or sin ‘conrerdec) Digiche are studied in Chapters 13 and 14. . eee ean al signalprocesing. Rather, sampling of Analog Signals “The principle snderyng dig! sgl pocesing is hat of sping the Fur 10.9 sacs ha conceal erm he pes of ebaning sais oe ney ‘ita The switch shown ene periodically wer he contol a pee pl oped {Glock Te closure time ofthe swith, lately shor, ad the tans ened ae ee eres ere {Gite As indeed, th S/H cet ons fan aaog vith at an be le (Send by 2 MOSPET rananision gute (Section 59), orage apa, and show) ‘Between the sampling intervals—that is, during the hold intervals—the voltage level ‘on the capacitor represents the signal samples we are after. Each of these voltage levels then fed to the input of an A/D converter, which provides an N-bit binary number propor- tional to the value of signal sample. “The fact that we can do our processing on a limited number of samples of an analog signal while ignoring the analog-signl details between samples is based on the sampling theorem [see Lathi (1965)] Signal Quantization Consider an analog signal whose values range from 0 to +10. Let us assume that we wish to convert this signal to digital form and that the required output isa 4-bit’ signal We know that a 4-bit binary number can represent 16 different values, 0 10 15; it follows that the resolution of our conversion will be 10 V/15 = 3 V. Thus an analog signal of (OV will be represented by 0000, 3 V will be represented by 0001, 6 V will be represented by 1001, and 10 V will be represented by 1111 ‘Al of the above sample numbers were multiples of the basic increment $V). A ques: tion now arises regarding the conversion of numbers that fall between these sucbessi¥® incremental levels. For instance, consider the case of a6.2-V analog level. Tis falls betwet 18/3 and 20/3. However, since itis closer to 18/3 we teat it as if it were 6 V and code it * Bir stands for binary di ‘Scanned with CamScannerass ANALOG INTEGRATED CIRCUITS o ” 4 44 us frre Hote it | Hot tt iri Horo ea Hore ot Horr tt © ee Se ae Ge be Ge ee at te ie it ty $A ee ie Wu i a Wi th tere Gan yo ee i fori th ant \ ee Pole oth tht riboey a 7 ‘ Fe 1029 ress of aly smi no sin, a Sane ant cet Te he Fo er evil) Il sie wae, Sol comet siga forte swith. (Output signa to be few A/D conver it 5 NE gle “ 61030 TWD wi cee eas ‘Scanned with CamScanner,109 ONDA CONVENTERS-AN INTRODUCTION 439 Fi, 10.31 ‘The ana samples atthe opt of DUA conver ae usualy fed to a sample-and- hold cicuit 1 obiain the staircase waveform shorn. This waveform can then be fered to obtain the smooth waveform, shown in color. The time delay wsally introduced by the fer i nx shown, 1 1001. This process is called quantization. Obviously errors are inherent inthis process: uch ertors are called quantization errors. Using, more bits to represent (encode or, simply, code) an analog signal reduces quantization errors but requires more complex circuitry. ‘The A/D and D/A Converters as Functional Blocks Figure 10.30 depicts the Functional block representations of A/D and DYA converters, As Infieated, the AID eonverter (also called an ADC) accepts an analog sample v and pro- juvee an'N-bit digital word. Conversely, the D/A converter (also called a DAC) accepts ‘an nit digital word and produces an analog sample. The output samples ofthe D/A con- werter are often fed to a sample-and-hold circuit. At the output of the S/H circuit a starease Maveform, such as that in Fig. 10.31, is obtained. The staircase waveform can then be “smoothed by a lowpass filter, giving rise tothe smooth curve shown in color in Fig. 1031, In this way an analog output signal is reconstructed. Finally, note thatthe quantization eror ‘of an A/D converter is equivalent to + + least significant bit (by). Exercise nee 1033 An analog signal inthe range 0 10 +10'V is to be converted to an $-bit digital sinal- Wists Os ‘eoltion of the conversion in volts? What isthe digital representation of an input of 6 V? What is he ‘omenain of nip of 62 V? What is the eo made th uanatin of 02 Yin atts terms and 2 percent ofthe input? asa percent of full scale? What isthe largest possible quantization eror 35 a percent Sy AIRE ein a percent of allele What sth lags posible g perce 1; ~0.064%; 0.196% ‘Scanned with CamScanner