Module 6 Sampling Theorem (With Solved Examples)

Uploaded by

ishan007rath

Available Formats

Download as PDF or read online on Scribd

0% found this document useful (0 votes)

38 views

Module 6 Sampling Theorem (With Solved Examples)

Uploaded by

ishan007rath

Available Formats

Download as PDF or read online on Scribd

You are on page 1/ 15

ter 6 - Discrete Time Signals and Systems 6.19 62.1 Baseband Samplin, (AU, Dec! 15 & '14, 16 Marks) (GPR, Nov' 16, 8 Marks) paseband signal: In communication engineering, the term baseband signal is used to indicate the unmodulated signal which has the original frequency components. Therefore, a signal in its original form is called a baseband signal. Baseband sampling: ‘The sampling of an unmodulated or: original signal is called baseband sampling. Sampling: It is the process of conversion of a continuous time signal into a discrete time signal. The sampling is performed by taking samples of the continuous time signal at definite intervals of time. Periodic (uniform) sampling: In sampling process, the time interval between two successive samples will be same and such type of sampling is called periodic or uniform sampling. Sampling time: The time interval between successive samples is called sampling time (or sampling period or sampling interval), and it is denoted by "I". The unit of sampling period is second(s). [The lower units are millisecond (ms) and microsecond (us).] Sampling frequency: The inverse of sampling period is called sampling frequency (or sampling rate), and it is denoted by F,, The unit of sampling frequency is hertz (Hz).The higher units are kHz and MHz. Let, x,(t) = Analog/Continuous time signal. x(n) = Discrete time signal obtained by sampling x,(t). Mathematically, the relation between x(n) and x,(t) can be expressed as, x(n) = XO) or = x,(nT) = +(E) ; for nin the range -0o 1/2] will bEidertical ggg “I we can conclude that the range of frequency ae ie fan analog signal is ~c0 to +20, We ime signal is — 1 the range of frequency of an an ie, time signal is~1/2 to oe Z rueney range continuous time signals are mapped (or convert), ~ analog signals, the infinite freq 3 frequency range discrete time signals. : : ; ‘The relation between frequency of analog and discrete time signal is, The range of frequency of discrete time signal is, + ae t vo(63) On substituting for f from equation (6.2) in equation (6.3) we get, Fst om (64) Fwill not result in an alids. But when the sampling frequency is se! such that F,/2 < E the frequency above F /2 will have alias with frequency below f, Hence, the point of reflection is F./2, and the Frequency F./2 is called folding frequen nt ie discrete time sinusoids, Asin (xf, + 2nk)n], will be alias for integer values of k tis# iC sampled ta neo inal with RequeneyF, willbe alas ofan signal with frequen lie aK, Fy cbect i's Fe M&A ithe sampling heqerey sony salle si fees yD +2, J the signal with frequency F, will be an alias of the sig™ Let, F,, be maximum frequenc time signal when sampled ata frequency fn tt1!°8 il that can be uniquely represented 584 Now, Fag = (68) © scanned with OKEN Scanner6- Discrete Time Signals and Systems med Foe 6.21 we(6.7), 18 frequency. From equation (6.7) we can say Tum frequency F,,., the sampling frequency to avoid aliasing F,>2F,,. : | (6.8) nae the sampling rate is called Nyquist rate, F. ‘the equation (6.7) gives a choice for selectin, nique representation of an analog signal greater than 2F,,.. ig sampliny pet with maxin aqoutd DE i xyquist rate When sampling frequency F.is equal to 27 N 7 iris observed that a nonshifted sinusoidal signal when sa Si eh 0 impled at Nyquist rate, will comple sequence (ie. diserete sequence with all zeros) because the sinuectoe) Geils anodes crossings (Refer example 6.5). Hence, to avoid zero sampling of sinewave, the sampling frequency g should be greater than 2F,., where F.,. isthe maximum frequency in the analog signal. A discrete signal obtained by sampling can be reconstructed to an analog signal, only when it s sampled without aliasing. The concepts discussed above are summarise as sampling theovem given below. Sampling theorem: A bandlimited continuous time signal with maximum frequency F, hertz can be fully recovered from its samples provided that the sampling frequency F.is greater than or equal to two times the maximum frequency, F., (i.e., F,= 2F.,). ; Note: The effects of aliasing in frequency spectrum are discussed in Section 4.3.1. Example 6.3 Consider the analog signals, x, (t) = 3 cos 2x(20t) and x,(t) = 3 cos 2n(70t) Find a sampling frequency so that 70Hz signal is an alias of the 20 Hz signal? Solution: Let the sampling frequency, F, = 70-20 = 50 Hz. (a) = x(0| = 3.cosan(20t)| mae 3 cosan(20x2) = 3c0s4En Neate . For integer values of n [cos (2xn + 6) = cos 0 = 3e0s2n(701)| = 3cos2a( 728") x(n) = x,(1)| 4n. 3cos En Amn) = 3c0s(2nn-+ f'n) ‘ i lias of x,(t) whe exnpelo™ the above analysis, we observe that x(n) and xf) are identical, and 0 (0) i an alias of) when ipl led ata frequency of 50Hz. Rample 6.4 sis Let an analog signal, x,(t) = 10 cos 200 at. Ifthe me frequen : an alias frequency corresponding to F,= 150H2- cy is 150 He, find the discrete time signal 2 = 10082008 x x(n) = | oe 10c05200nt| 1008: E © scanned with OKEN ScannerSi a eas nals ang Any = 10c08( Sn - “4 x(a) = 10e08 0055+" = 10°08"3 : 2h = 10¢0s(2nn ~ 2'n) = 10008 %3 iscrete time si whose frequencies are separated by integer mje __We know thatthe discrete time sinusoids 2 identical. 8 _ in is an alias F105, +. 10c08 28: Brn = 10c0s: an = 10cos(2k-+2n)n = 10c0s"3°m 4 Here, 10c0s 8fn = 10.cos 23-0 ncomparing the above equation wit standard form of cosine signal, A.cos 2nfn, the frequency of. 7 f 4 cycles/sample Welnowtha, f= => F=fh 4150 = 200H2 ». When, F, = 150Hz, F = 200 Hzis an alias frequency. Example 6.5 Consider the analog signal, x,() = 6 cos60zt + 3 sin 200xt - 3 cos100zt. Determine the minimum sampling frequency and the sampled version’ of analog signal a this ‘Sketch the waveform and show the sampling points. Comment on the result. Solution: ‘The given analog signal can be written as shown below: X= 6 cossost +3 sin 200nt ~ 3 cos1O0st = 6 cos 2n Ft +3 sin 2n Ft ~ 30s 2xFt where, 2xF,= 50, = F, = 25 Hz" 2nF,= 200n => F, = 100Hz 2nF\=100r => F, = 50H2 ‘ ‘The maximum analog frequency in the signal is 100Hz. The mini i cee that ofthis maximum analog frequency. eae: ie, F22P,, => F22x100Hz = Pp = 200 He Let, sampling frequency, F, = 200 Hz Let, x(n) = Sampled version of analog signal #10) = AO] = X00 m4 50mm, 6 cos fam +3sin200nn _ 3 455100xn ‘209 ~ 3 cos "hm = 6cos@ +3 sinxn-3.cosS2 aa Fig. 6 cos ~3 cos. ‘The component ts of analog waveform and the Sampling points are shown in Comment: Inthe sam ee e sampled version of analog sate 9 when sampled at 20042 fr any ae Ate), the component 3 sin 200nt will always give 2°, sampling atF.=2F M€ of m. This is the drawback in sampling at Nyauist © scanned with OKEN Scanner| - piscrete Time Signals and Systems aapter PEC 6.23. | seonton: Fase; = 2004s L— sanz F-00700 ae Fig 1: Sampling points of the components of the signal x,(). Example 6.6 (AU, May’ 15, 8 Marks) Acontinuous time sinusoidal signal cos (2nFt + 6) is sampled at a rate F, = 1000 Hz. Determine the resulting signal samples, if the input signal frequency F is 400 Hz, 600 Hz and 1000 Hz, respectively. Solution: Given that, x,(t) = cos (2nFt + 0) and F,= 1000 He. Let, x(n) = Sampled version of continuous time signal. r x(n) =x (0, 9 = c08(2aFt +0), Te When F = 400Hz (0) =x10 eco = 08(2838908.+8) = cos( n+ 0) 2nx600N , 9) — cos(SEn+0' Feikinone alg of ¥,(0)=X(D))p. gq = €08( 27 E008 +0) = cos(Fin-+0) Forttegeseloes When F = 1000H2 = = cos 2*X1000n , 9) = cos(2nn+0)=cos0 = Constant “Wort lam © s(t ) (Independent of frequency) Example 6.7 Consider the analog signal, x,(t) = 2 cos 2000nt + 5 sin 4000xt +12 cos12000nt . a) Determine the Nyquist sampling rate. ») Ifthe analog signal is sampled at F, = 5000 Hz, determine the discrete time signal obtained by sampling. 1 analog signs . © scanned with OKEN Scanner6.24 Signals and System Solution: a) To Find Nyquist Sampling Rate The given analog signal can be written as shown below. 2cos 2000nt +5 sin 4000nt - 12 cos12000xt =2cos 2n F,t+5 sin 2x F,{- 12cos 2x Ft where, 2xF, = 2000r => F, = 1000Hz 2nF, = 4000, => F, = 2000Hz anf, = 12000" => F, = 6000Hz ‘The maximum analog frequency in the given signal, F., is 6000 Hz, The Nyquist sampling rates ti that of this maximum analog frequency. ;. Nyquist sampling rate, F,, = 2F.,,, = 2 x 6000 = 12000 Hz In order to avoid aliasing the sampling frequency, F, should be greater than or equal to the Nyqust™® | b) To Determine the Discrete Time Signal Sampled at 5000 Hz | Given, F, = 5000 Hz Let x(n) be the discrete time signal obtained by sampling the given analog signal. “.x(a) = = 200s 20008. 5 gin 4000nn 12000nn () 4 s.x(a) = x,(t)|_,,=%al|_, = 2c08 +Ssin 40M. 12.c0s 120908 A a Foren - 2000nn jn 4000xn 12000nn cos (2x8 * 2cos 20088 + Sin Ooo t L2cos Macuern. 2nn in 4 en mn = 2.008289 +5 sin 48 + 12008 1259. = 2cos 2a +5sin 48.5 12.c05(28¢+ 1") 2m. jn 450. mn = 2cos 250.45 sin Ann + 1cos(2m0 ¥2nn) = 2cos ayn +5sin an 12.608 75" = 14.cos 22. + 5 cos 40. S 5 oe BEE 008 ‘Comment: When sampled at 5000 Hz, the component 12 cos 12000nt is an alias of the component 2 625 @ scanned with OKEN Scanner.g- Fourler Series and Fourier Transform of Discrete Time Signals elation between Continuous Time and Discrete 86 e Systems Jation between continuous time ae ystems and discrete time system is straight i ring preserve the frequency response of continuous time system. This will happen, mit conan ing FX « sampled at sufficiently high rate. When sampli i i “me signal is samp! : sampling rate is very low there is a chance of shifti nfrmation from one band to frequencies of another band, Due to this the high equensy ose ofr get the identity of low frequency component and this phenomena is called aliasing. The minimum sampling rate to prevent aliasing can be obtained from sampling theorem. gol Spectra, of Sampled Signal Let x(t) be an analog signal and X(jQ) be Fourier transform of x(t). Now by definition of continuous time inverse Fourier transform, a x)= Ef xGinyean 847 (8.55) Let x(nT) be a discrete time signal obtained by sampling x(t) with sampling period T. 2-x(nT) = x(0) tant Using equation (8.55). Expressing the integration as summation of infinite number of integrals. XGO) in the interval Qm-1)r, Qm+ix Tat is identical with XGQ) in the interval —.to+ Since mand n are integers ee", )) emt do The relation between analog and digital frequency is 2= 2. 2nm)) edo +281) cb do (8.56) By definition of inverse Fourier transform ofa discrete time signal, the x(nT) can be written as, weary = of Xe) eM do ‘On comparing equations (8.56) and (8.57) we can write, x00 = E SUCRE) © scanned with OKEN Scannerls 8.48 aay x x(os) = 4. s X(j(O+ 859) ey si et log signal, then x(j(a4 2: In equation (8.59) if XG) is the original spectrum of the analog signal en *i(a+2m) ‘ frequency shifted version of XGQ), shifted by 2nm._ fn equation (8.59) the term —- will scale the amply ofthe spectrum X( (2+ 2am )) by a factor t. ‘Therefore, from equation (8.59) we can say that X(e) is the sum of frequency shifted and amy scaled version of Xj). In general we can say that the frequency spectrum ofa discrete time signal obya bby sampling continuous time signal will be the sum of frequency shifted and the amplitude scaled speci of a continuous time signal. This concept is illustrated in Fig 8.7. [XGo)| 1 “2,0, — Fig a: Spectrum of a continuous time signal x(), with maximum frequency Q,, ev | et 2a 3n 4e Fig c: Spectrum of a sampled version of x(t), when 9/2=9,. A A : Fig d: Spectrum of a . ae ood sampled version Of%(), When 9./2<0,,0r ( 2 3) s Fig 8.7: 5; i ig 8.7: Spectrum of a “anal sampled at various sampling rates © scanned with OKEN Scanner_ Fourler Series and Fourier Trai cpopter 8 Fo : nsform of Discrete Time Signals ae Aliasing in Frequency Spectrum Due to Samplin, 8.49 tion © = OT, where T is the sampling 8 the radian sampling fr in this transformation, the radian frequency « ofthe», ieee ssanique inthe interval ~1 (0 +t, and the eyclic f {eal isumique in the interval 1/2 to +1/2, The maximum frequency in the spectrum shown saximum frequency of the sampled version of the disc sampled ata frequency of ©, /2. IFQ, is equal to Q, ‘ampled version of the discrete time si ignal requency fof the sampled version of the discrete time in Fig 8.7a is ©... Let o,, be the corresponding rete time signal when the spectrum of Fig 8.7a is (2, then the corresponding value of @, is given by, = = &: 2nF, Om = nT = Sep = 28 AT= qlan From the above equation we can say that if Q,, is less than Q, /2, then corresponding «, will be testhan = and if Q, is greater than Q, /2 then corresponding «, will be greater than x. From Figs 8.76 and 8.7¢ it is observed that as long as Q,, is less than Q, /2, then corresponding @,, is less than or equal to ,and so there is no overlapping of the components of the frequency spectrum, From Fig 8.74 it is observed, when ©, is greater than ©, /2, then corresponding «, will be greater than x, and so the components of frequency spectrum overlap. Due to overlap of frequency spectrum, high frequency components get the identity of low frequency components. This phenomenon is called aliasing. Due to aliasing, the information shifts from one band of frequency to another band of frequency. Therefore, in order to avoid aliasing, ©, /2 should be greater than or equal to ©, Since Q,=2nF, and Q,=2nF,, to avoid aliasing, 2nF,/2 >2nF, *. FL >2F, Therefore, in order to avoid aliasing the sampling frequency F, should be greater than twice the maximum frequency F,, of the continuous time signal. (AU Jun'l4, 2 Marks) an be bandlimited by passing through a filter £83 Antialaising Filter A continuous time signal with a large bandwidth d fil before sampling. When te requeney range of the output signal of the filter is chosen to prevent aliasing ‘te to sampling, the filter is called an antialiasing filter. S64 Sampling of Bandpass Signal Acontinuous time signal is called a bane “equencies, Let the lower and upper value o| the bandwidth "B= F,~F,". Let F, bea frequency Horse Nency spectrum of some bandpass signals is shown in Fig 8. dpass signal if its frequency spectrum lies in a narrow band f this narrow band of frequency be F, and F,, respectively. corresponding to the centre of bandwidth. The [xl oA RAE 8 RRR E um of continuous time bandpass signals. Fig 8.8: Sample frequency spect™ © scanned with OKEN Scanner850 cy ii dpass: the bandpass signal has to be s s fet 0 avoid high wat PH + will be very high. In order to avoid high sampyign "9h y hi ey, then sampling rate wi y f npling ts ee patecanbs shifted in frequency to an equivalent lowpass signal and the equivalent ie hy andpass signals s signal can be sampled at a lower rate, Ta ; - [Avanalpass signal can be shifted in frequency by an amount Flo matidite Signal into an hiv Jowpass signal, and when the upper cutofT frequency Fis an nce i uP mw B, then equivalent lowpass signal can be sampled at a rate of 2B samples per com When the s frequency Fis not an integer multiple of bandwidth B, then the sampling rate has to be slighty, itt and go upto 4B. In general, the bandpass signals with a bandwidth of B Hz can be sampled ata rate of 28 4, 865 Signal Reconstruction (Recovery of Continuous Time Signal) In the above discussion it is observed that if the sampling frequency F, > oF» then the X(@") of the sampled continuous time signal will have aliased components of the spectrum XG) of te criginal continuous time signal. The aliasing of spectral components prevents the recovery ofthe rg signal x(t) from the sampled signal x(n). signal is F,. A When the spectrum of the sampled signal has no aliasing, then it is possible to recover the orga! signal from the sampled signal. When there is no aliasing, the spectrum X(o) can be passed thoughs Jow pass filter with cut-off frequency, w,/x. Now the equation of spectrum X(e) (equation (8.59)) cant written as shown below: : X(e")=FX(jO) => X(ja)=Tx(e) (On taking inverse Fourier transform of X(jQ), transform of continuous time signal we get, (8.6) we get x(t). Hence, by definition of inverse Four: +n ‘yt = Px Goye%eo 1 pe 10-3 f X(MeMED=F- [X(jO)e™a0 Because X(j0) is zero ous ha ” [imenal-wTiont 1 =L [rx(emein on J xenon ‘Substituting for G2) in : “at [from equation (8 i es et "On f T Di x(nT)e™ el gq) Using the definition of Fourier i al transform of discrete time sigs!) [transform of discrete time $0) - { ; Ermer. yr Stary fo L fom a0 cae ale “ a an Ah aL eZ SBaqamelemato _gtcemtaan | a par a (tat) =D x(n) eterna) al nt | © scanned with OKEN Scannerir terpolation formula: i er ime signal x(t) from its samy formula. ‘6 Hold Circuits sing most recent samples and such methods us ; are implemented us in the reconstructed signal which can ya be easily eliminated by a Depending on the number of samples used to reconstr can be classified into zero order hold, first order hold, second order hold and so on. The n order hold cicuit uses n+ 1 sample to reconstruct a signal. In reconstruction process, the first and higher order told circuits offer no practical advantage over zero order hold circuits. The zero order hold followed by alow possfilter will give a satisfactory reconstructed signal. poise \ppropriate filters, uct the signal from its samples the hold circuits Zero Order Hold: It is a device or circuit in which the value of reconstructed signal during a sampling interval will be same as that of last received sample. The zero order hold is the simplest signal reconstruction device. Mathematically, the signal reconstructed by zero order hold is expressed as, x()=x (aT); nT sts(n+ IT, n=0,1,2,3,.. where, T= Sampling time period Whenn=0, x,()=x(0) ; O F, = 1000Hz 2nF, = 4000x => F, = 2000Hz nF, = 12000n => F, = 6000Hz rhe maximum analog frequency in the given signal, F,,, is 6000 Hz. The Nyquist sampling rate is twice that of this maximum analog frequency. + Nyquist sampling rate, F, = 2F_. = 2 x 6000 = 12000 Hz "order to avoid aliasing the sampling frequency, F, should be greater than or equal to Nyquist rate. +b) To Determine the Discrete Time Signal Sampled at 5000 Hz Let x, (nT) be the discrete time signal obtained by sampling the given analog signal, ~-XQT) = x0] = 2008 2000xn +5sin 400 +1205 12000m, = 2cos 200mm 5 in 4000nn 12000nn, 5000" * Sein we og9" + 12cos 22000nn, = 2c0s 22 +5 sin 400. 13 c99 = 2000 22 4 5 in Aan 2 0s 280.5 sin 4 *12605(228 + 24m) 2606254 sain = 14c0s 2xn_ 4xn cos 250+ 5cos 4am, Comment: Wh [ 'en sampled at 5000 Hz, the Component 12 cos 12000nt an alias of the component 2.cos 2000%. ¢]