Open navigation menu
Close suggestions
Search
Search
en
Change Language
Upload
Sign in
Sign in
Download free for days
0 ratings
0% found this document useful (0 votes)
89 views
ECE3793 Notes CH 04
Uploaded by
Kalisetti Anil
AI-enhanced title
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
Download now
Download
Save ECE3793NotesCh04 For Later
Download
Save
Save ECE3793NotesCh04 For Later
0%
0% found this document useful, undefined
0%
, undefined
Embed
Share
Print
Report
0 ratings
0% found this document useful (0 votes)
89 views
ECE3793 Notes CH 04
Uploaded by
Kalisetti Anil
AI-enhanced title
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
Download now
Download
Save ECE3793NotesCh04 For Later
Carousel Previous
Carousel Next
Save
Save ECE3793NotesCh04 For Later
0%
0% found this document useful, undefined
0%
, undefined
Embed
Share
Print
Report
Download now
Download
You are on page 1
/ 76
Search
Fullscreen
Chapter 4 Brie& review of Eigenvectors & Eigenvalues -IP A is a matrix, ds vector, and KET is a Complex constant, -and if Ad= dt, D> Then U is called an “eigenvector of the matrix A with associated eigenvalue d. Bo TE ea] eel) ~ 3 r ats [33 2 |[.] ; [i] = 4. 4 76 4 2 g |. eigenvector of the matrix A with associated eigeavalue de 4. > Multiplication by A doesn't change the direction of a, vt usd “stretches” the length of He vector. me 4,|— For a system UH, Here. may be certain inputs xX (4) fxonf Such tht the output is a cwplex number times the mput = x05 fd Po yltd= dete Ate) @TH > yer = d xtH9 9 Tn these cases, we cal) x(t) fxoaf aoe enannanal of the system Wath acsociated eigenvalue ), FACT + The signal coy for any Fixed wR, is an Cigentunction p,_ any a ‘serete-time LSI system, : - jwot FAC: The signal ede’ y te any fixed “EIR, is an Cigentancticn of any Continvors-tne Let syStem, Pen . Ce ie ren jwot 7 wet e* on Notes => The eigenvalue X generally depewls on the specific value pf Ww, Given on LSL system H, each difterent choice of wr will give you a distinct inpuct Signal with 4S own dichned eigenvalue. PAGE 4.2Proo® that eM ic an eigenfunctin of any continuous-time UST syctem: Let H te a continuous-time LSI system with impulse response WCt). Let the system input be xit)= er* 4, some Fixed we eR. xltaeiet af iy Py ylt)= et) # ht Then He system output is given by yl= x) eh) > Sette) he) at = LPreimlt-o) hlt)at (° est ee" kityat ry . oo .) ob = Ee" at lt) Chley, xD) = xit)d 7 + de ends Nn wy: The inpat A complex rumler tH E PAGE 4.3Proof at et is an eigenfunction of any discrete -time LSL system: Let H be a discrete-time LST system with mpule respec WEnd, Let the system inpuct be xtar = ed fur some Lier wre R- xtc eb" aH yond = xO * ht, Then the system output is given by yOad= xonde hon} = Even kIheey ye . $= ete) ack yew a - Femme \Ce) . et D hehe peo ajook = XtnI CHEMI, xO D = XtW} d N\ complex number that depewls on Wy» The input _—_ The 4.4—In Chapters Lh 2, we wrete signa ae linear combinations of shifted deltas. — Kronecker deltas in the discrete-time case. > Dirac deltas in the continuous-time Case ~ For an UST system with a knew wmpule cecponse, His made TY eacy 40 Gnd the system output fr an arbitrary inpat >The output.was given by cnvolution . Be he dele ere were easy to thnk about beuuce md Cte gM Me atl es ETT ERS in 2 Bat Convolution is somewhat Complicated to think about because it does not di / + directly give uc ¢ cloced expression fy tye System "patna ‘et form ~ Ty chapters 445, we will write signale as linear Compositions” OF bases made Lema Sinusoidal compley exponential signals. — Since each besis signal will then be an Cigentuncticn of any LST System, this will make i+ easy to \ uw NPAT linear Compositing -This wilt usually give us easy closed farm expression s for the System output in Vespemse to any arbitrary input siqnal Cuhen the input is written in terms ot etgentanctions), PAGE 4.0— We will write cwatinvors - time signals x(t) as linear yt Combinations of the basis sex fe cae This witl be done using dot products ac betere. — Given an LST system H, we will write the eigenvalues for all of the . basis signals eiet together as function Hiv) of w, a NOTE + The book writes H (jw), but this is not standard. ~ Similarly, foe a discrete-time LOL System y, we will Write Signals in terms of the basis set Fee} Recall Put discrete-time exponentia wnigue fir We Con, wy QI get the seme. sig wel) are onl Outside of this interva! wale over agains k edn h - For any LST system Hy we will weite the eigenvalues tor all the basis signals ein together 25 @ Lunctien H (ei) of w. Note: In this case, the books notation Hei») is Sted ord ~ For any LST system H, the collection of ei or H(ei”) is called the Frequency the system. genvalues Hos) Response oF PALE 46—Tabuleted as a function of wy, the frequency response gives the eigenvalues associated with all complex eyponential siasoidal wmputs el or eter. — The eigenvalues are generally complex. > So Him) is a complex number fur each w. => Heb) is « complex number for each wo. ~Thus, the frequency Tespense can be weitlen in Fectangu ler form: Huu = Re Tuy] + de TH) J Hlet)= Re [Hlet)] +4 de Li Lede) ] OF im polar form : Hus) = [Hey] ed Hw) Hlet) = liter) | edeat (e's) Question: If the system input is.a amplex exponential (eigen buncton), how is it affected when ot Passes through Hae systewe 2 => In other words, since we know the output is a complex number times the imput, what des the output look like 7 PACE 47Answer + Continuess-time case: x(th= ehFe cost + jsinwt ylty= Hed x4) Him eir* (He) | eres te) ei uv Uikusy| @2 Got Fara He ) \ [Hea] coslwt t ry How | + 5 leo] sin [att ersten => Real and imaginary parts of X(t) each get scaled SY [Hur] and shifted by ary Heo) , Discrete-time case: XEnj= ei = cosusn + jsin con ylt)= He) xewy = tle) er = |uteir) J ]eberstlet) ein - jue |e [ort arytier) | = Jace") cos(wnt acghter) J+} (Hee*|simfian varghle) | DP Againy tre real and imagery parts of the mput eath yet sealed by [Hled) | and shi teal by arg Ulet”), PAGE 48—%) if the input to an LSE system can be written as q Sum of complex exponentials, and if the eigenvalues arc. lenuwn, then it will be easy to caleulete whet the System output is. Riemann— Lebesque Lemma “oom ~The theory of distributions y alco Known as generalized functions describes signal iacterms of how they behuve in det products and in systems, ~ THs ic not the same as “Probability distributions” and the two should net be con busecl . ~—we da't have time to cover distribution theory ™ Ece 3793, -But we do need one of the importent resuts and itt most important consequence. —The Rewmann- Lebes we lemma states tet, in the sense of distributien ¢, : wt. Jn eF > 0. => This implies that dims, Coswt = don sinwt = 0 => Tt means tat in dot products and sy Stews, as wae, he signal edvt) coswt, and sinwt all behave Jost like the signal x(t) =O. PAGE 4.9- The most important consequence. of the Reman -Lebesque Vemma is *he Following: . 7 Sint FACT: din au = FU). - We will now vse these results +p see thet the basis te Tee is orthogonal, bict not ortranormal . ~Whon we use the cbt product to write signals interns of basis, We with have to “Six things up” to accamt for his, —Reeall: a basis is orthonormal if the dot product of any basis yectam with vtself is one and the dot preduct ot any two different besis vectrs is zero. —For the basis Feit? WER, > the dot product of any two diferent basis vecturs és Bera, So the basis ig or thayenal bat the det product of any basis vector with itsel# 'S 27 stead of one, so-the basis is ngt orthonormal, PAGE 4.10~ Lets shew this: 7 - Cert, e*> = (ebtte i at or) ee Ain § lH) zap J & A 2 dn : Jorma) ~ pave yloroh) Le J, \ oe ae = fr Troma) [e* adm site oa] = Am 2 dtd A crite ad Wy, At ee 2) 2 dee Din Calera) ] Or 20 hw sim Tairua) ] Thao) ANS (-r) Ono taaue tt, toy M2 This is the norm squared _ 90 Length = Vor" ity iy PAGE 4!Recalls To write a signal in terms of an orthonormal basis, dnere are two steps: ® Take the dot product of the signal with each basis vector to compute coellicients , (D Add up the cwebGicieds times their respective basis vectors to get your signal, ie je > With the basis fed” ie all of the wegéicients wit b mee THe large by a factr of [BT since the basis 'S not oythonormal » —> In step@, we have > divide by 29 fo Correct Fr this, stiwe each basis Vector aly ty long by Van’, ~ Now lets write an arbitrary continvous-time signal X(t) in teems of the hasis feirtf wer. — For con venience, from step D We will write the reguived coetticients aN together as x Lunctien Xlw) of w. Her D+ compute te coe tts cients (det products) ; . es Xl)= Cx, e*> = Meh at @) Step®@+ naa ne en ak the wetficients times their Yespective basis : a to ge+ X(t). Remember ty divide by 2a since the basis 6 aot orthonorine |; X= de 6"xmer* a dw (+t) PME 4.1LEquation Or) is called the Fourier Transform of x lt): x)= Fw] = Pres at Equation (#) is cated te Inverse. Fourier Trane fern ot YXlu): I rane Fearn xl) = F Ther} = Kiet gy > The coordinates Xlw) of the signal x(t) with respect to i wt the basis tei Seer are called the Cregueny douain Cepresentection ae x(t), Note ; At points where x(t) is discontinuous, I'L xley] converges to the midpoint of the discontinuity, EX: X(t) 4 ” — a — xw 5 a4 aoe 2» t Pace 413NOTE + There are four commen ways thet people dea} with the fact thet +the basis 4 elt. as not orthonormal, ~ This cesults in four di€ferent definitions Le the. Fourier transferm. —When you read other books, always be careful 4p note. which definition is being used. ~The way ovr book does it is pretty much standard in electrical ty computer engineering » X(w) = CPectleniot ag x)= FS K00) ett des ~The other three ways are: A) Fic up on the Sonsaed trans ee X= trem ge OH = SP xt) e* da B) Fix up eguatly on forward and reverse transforms. Xl)= he Sinett at x= As Sxts)elt gy ©) Use the basis feuney q fer, ne to H2 instead 2F radians, This besis iS orthonormal : Klee Saceyerieett yy x= Cx egyeP™t yp Pace 4./4 where “fgives epuency- But fe €CE 3793, we will always use the Furie AYeans farm definition Siete sit stat the book + Xlye See Le (ened trmelen) x)= de Oxley (reverse trans tern) — We will often write + x lt) > Xlw) 10 indicate thet X(t) and Xl) ave a Fourier Transform pair. Remember : [xu] converges to the midpoints of discontinuities in x (4), Remember ; The hook writes X (jr) intteed of X(u), >This 5 not standard. DIT will write Xo), > hu an write Xlw) or Xlpr), Prog ANSPAGES AG cates: Are BLANK POE 4lb— PNK 4.22- Generally, Xlw) is a complex wumber Lor each w. — Thus, we can write Xtw) in vectanguler form: Xlw) = Re [ xis ] + {dn O Xl) ) or w poles form: Xlw) = Ale) eit) > Aw) = (Xl) ] Is Called the Swmaguitudle spectnm of XU). —> P&) is called the “spectral phase" of x(t). — Relationship Lebween rectangular and polar fecms: Aw) = VfRel x) $7 + fetal xtay§ Sun Xo] Plw) = arctan Rel Xia] Re(Xm] = Aw) cos dtu) J SIm[ Koy] = AW) sin [6l4)] PAGE 4.23EX: a xe § xwMewt , Sx yet at n xe Uk), 70. oat 4 free at ie - Cart ay e : “t = (aryw)t |” ae w to ~1 d = (area 1 ae -aj+ = ati -To ts i fo get this in Tectanguler torn, we conjugate the denowina tor : \ ew . 4 Hy Ee a ke | Cae X@)= ayo a=\o a > Re[Klo) | = Gat a dulXh)J = se atta - For polar form, hoy 2 txenl = [tat a Guy enn) \ —_—_ (at tur)” J (at+u*) . Vareut pace 424~ EX -altl XE) = ) 270. t ete ivt vo - 2% Xlw) = Smet ta = fever e+ fe a ) fer deren of — oe a = ajo at jw = LL atiy to ari a-ju aajw atjy a-jy = LCL Za attwr true Gru ——_ PAGE 4.25= Xlw) js real-valued ww this cose, So Ze Din LXtw)] = O ~ Re[xtw) | ce ee 2 AGe)= Gara pow) = © x(t) X(w) 4 i xcy= 2 boo TCT oy, \tlaT kote: xlt)= u(t+T) - u(t-T) T ~ jot Xtu)= f xBe ar = Feat a uF T Oo < f coswt dt +] f sin(-wt) dt CU a « integrand is EZR; ae ZsinoT aa Pace 4.26: eos cde (xt) = ReLate)] = hoy = Ikon] = 2[8t| ge) = ary Head = 0 Xl) ~~~ (x00) — ar Ty eT 9 ° ) tty T Xlw) = Fae em a = Pargetat + ([u- bye" at ae i iste 4 Telety — fe eat = fe at + r{te { v le 2 femat af teat - 5 Ptew rat -T pace 4.27ig wee XC) = 5 tos (-wt) At = T " Into ZsinwT | w Lsinw T a w 7 ( simCwt)at -TZsinwt , 2 - 2emut- 3 a SNOT TS cos T vee Xw)= we Tut ot cancel ae [iH ws | = aa (25042) | _ 4 sine (2) Xl) = = ee — Kw) ene 429INVERSION - Like the Fourier transform, the inverse Fourier traustan sas jet wine fe [Xe ae generally fequires integration. — The forword and reverse (inverse) transform tategrals are almost the same. . yet > en: ert IFT: e” 3 FT + fat LET tafde - So the same types of integration Techniques are Used for computing both forward and inverse Fourier transforms. — All of the Preceeding examples can be turned into examples of the inverse Fourier trans forn by: L Putting a out front. 2. Switching St and Sey" everywhere, 3. Changing est +o ert, PAGE 4.30TABLES le ~ In practice, we doo't always tegrate to compute. a furier transform or reverse transform. — Ta fact, yor should use integration only as a las} vesort . — where possible, a) to find the transform you weed iw a table. Then yeu Can gust write Bown the auswey, Fourien TRANSFORM PROPERTIES, -We will how look at the main properties of the Fourier transform . — Often, & you need q transform that is not in q table, you can use these properties to change the transform you need into ome that is ia the table. Cor & combination of ones that are in the table). race 4-31Linearity ; FE lt) Xl) ad xt) 2S Kl) and ai and a2 are Constants, then Ay k(t) + a2 Xp lt) et, a, Xiu) + 2X, le) . Proof: 4 [a xt) a,x, (0) = Sle Xt) eau GE ‘it ea, fxte at + af Swe at = aX wv) + a. X2). Qed. Time. Shifting % t > joto ve KLE) > Xtw), then xlt-to) FP LOVE a -jwte Cpu) - to Notes Xiwy @! = Alw) elf ~J J So time shifting introduces @ linear phase shift iw the spectrum. ° oat urt-t. teurt, Proof: HL xtt-t)] = [xcete at aucat atedu ee - _ jw lutte) Lame’ a juto i = en weit, = Xweh™ Bed. pace 4.32a Frequency Shifting » FS = 4 if x4) SX, then PK) > Xlu-we) . Proof ; USW-Wo WF Utn ¥'LXuw] = Be (Xtras deed aude - "y (utwe) E =f xt erererd -7 . - sist = emt | xme™a = exit), Ged, -w~ —The Frequency shifting property is wed often in radio and TV communications te shift a “baseband” Program up +o the carrier freguency ot the station, EX: Xxlt) is a music signal with Fourier Traws frm Xe) Ad... < IF the station hes an assigned frequency of Wo) here is what the transmitter can do! Yoo). x(t) ee y lt) a eit wo ° >This is called Awplitde Modulehion ) or AM. — Pace 4.33Time Scaling : + ie x) <—> Xt), Hen x(at) ets ia X(8), ato, Proof + (i) Suppose a70- Then o a eat t= $[xue)] = [Trove at jucedt ate ae ale 2 (xen? ** a 4 a - 48 et{ xmer da = £X(Z). -e Cii) Suppose aco. Then vat tee Fray] = f Kater tae gazed ate as ete ls = if xujel?* du -£( xue Pan ee -00 = or X(2) Combining parts Ci) and ” Slew] = XR), aco. ato.Frequency 2th Sy Xtw), then hp x(t) <5 Xtau), ato. if x(t) 9 Prook + (i) Suppose a70. Then a » | usaw ye & F'[Xta]= def xte aw accede gue de \ Bt { 7 tut = BL Xu) e2* au =i ae Xwe yy t = £x(#), Ci) Suppose Q<0. Then rn uzaw w= ae F [Xm] = RI Xewde Ae seeds aye de - . o t -i(x silt = bE fxwe**y 0 -~ a = -4x(k) Combining parts (i) and Gi), FX] = ia * x(E) ato. QED. NoTe: By combining the tine sealing and Preguency shifting. men we at: juat FL y (wow if x) Xlw), then et lat) > Ta XCM) pace. 4.35Duali Duality : a if xlt) > Kew), then X(t) 9 21 x(-w), Prook: since Ct) 2 Xiu) by hypothesis, we have 60 w= F'Lxw]= + ae Xtwy eb" Aw. Be-t So 21Ux(t)= Og w= [wer as f 2 Yow > 2Ux(-3)= S “(wei tay Deeper iste Gala amxtu)= {xte™* at = FLxey] @ed. EX . On page 4:27, we showed thet (6 by Wet LF DsineT x)= Lo, lel TT <7 —S By duality, sing gt 1, lwli
(y0)" Key) Frequency Differentiation 4 Z: it eof Xo) if xt) > XM), then -} xl ia ——_ PAGE 4.37(7 xe" at = Xtu). 00 san so £Xtw) = a{y xe" tf (Fan {her S a Proof: by hypothesis, " fF xt) Gite at = $[-ite], ass. W —Applying the Frequency differentiation property n times Tepeatedly, we get f_ a" Cit)" xt) > Ze Xe) Conjugation : S if x(t) > Xo), then ° put Proof: By hypothesis, xuye Flue) = J” xe * at, 5 own . & Xu) (ames “ae | = foxtie at =o Fx x*(t) > Meo) awrite — 5 for w; X*(-§) = i xe Fat = write w fie F + Xtewye J xterm ae = FLX), QED- pace 4:38NoTE > if X(t) is REAL» then xX#(t) = Xie). — Replying the conjugation property, we then get XFowy= F Lx] = FLxta] = Xe), ~ In other words, => iF XE is realy then Xls) is conjugecte symnetric . —> The real part of Xw) is even. —> The imaginary part of Xlw) js odd - <9 This implies Het Alw)= |X] is even; \xcen] = Vv fReLxcn df? + fr keyly FY freCxusf® + fan Law) f* =U fre lxwyR* + {aLxew)JI* = \X)]. even. > This alo implies that Pw) = arg Xiw) 1s odd : ha OXC-w)) 7 ew) = tan ———_= arg X(-w) = arctan Re Exeor) aot = dn CX] = ACN REx tad d- = arch, MACHO 2 arg Xu)» 288 arctan REx] PAGE 4.39EACT : if XU) is a purely imaginary sighal, then Klw) is conjugate antisymmetyy'e » Xtw) = — X* Ww) > ReLXw)] is odd- = edin CXC) J is even. ~> Awy=|Xw)] is odd. > dPlw)= acgXtw) is even. => Recommended Exercise: Show This. aK Even Stgnals ; if x(t) is real-valued and even, then Xlw) is also veal and even. Let x(t) be real and even. Then Evans Xw)= Lxeme* at 2 ~ = J xW cos(-wt)dt +} So x(t) sin(-wt dt “+ = J xe)eoswt dt — } J" xt) sinwt dt x Be But xlt) is even, so xit)smwt is odd) and He Secon Wtegml is zero! i Wich i \ Llwys [xtaoswtat , whkeh is real - =~ — pace 4.40Furthermore, oo Xw) = Fate wos(-wt at = ( xw) esotat = Xl), soe ce which establishes that X(-w) = Kw), QED. $ 50% Teal and even <> real and even. Odd Signals ~ — Ina similar manner, one can show thet if x(t) ts real and odd, then Xlw) is pure imaginary and odd + + Teal and odd <—> imeginary and odd « er Recommended Exercise: Shos This. ek —Seommended Exercise: NOTE; Recall that any signal x(t) can be uniquely written as the sum of an even signal and an odd signal: Xt) = & fxlelf + OAL xcs, = where Gixtey = L[xat xe] Oafxayf = 4 Cur xt) ] oa —> Pace 4.4)—Taking the Fourier transform of x(t), we have [xu] = Fe ix3 + eaixins | mere Fa sxeos ] + $[oatneas J “2 Re[ Xu] + jd CX] PATE imaginary => Se, > ReLXWI] is the Fourier trans frm of & &xCE)§ 7 joel Xe} is the Fourier tramhrm of CAL KUN. —Now fr some examples of the properties we have seen So for: EX: Apelying the duality Property to the transform par we derived on pages 4.26 and 421, we get l lw]
Xlo) = {. lwlya Fad PAGE 4.42.EX; x)= oo > KET oO) WAT Applying the fregueiney shi ting property te the transform pair we derived on pages 4:26 and 4-27 gives us Xlu)= Sin (=o) T 4 SIN (wt we) T ty Wo Wt Wo xt) X(w) EX: Applying the dualety property fo the transform pair we derived at the bottom of page 429, we get stat, _ Fleet ez = €—> Xlw)= i vay Wisse Tat © , Wwl%Za x(t) he X(w) eta AN. PACE 4.43EX: Applying the frequency shifting property to the preceeding example, Wwe get - t - . a juot Zswrat oct = Sitat (4 [e**re ” J) Os) Tat? Tat? Xlw) Time Convolution * 4 t iF xe) 5 Xilw) and L(t) > XL(4), then x, (XM) > Xv) XY, lw). pace 4.44.Proof: OL aw aly] = F[ [Pace xte-e) ax | . (" [ { xe) nit-oat| ef at te Le Change ms cot ~ ipo Soe ( xed) j x, (t-te at | at order aa a Dorly Hue shiHling prperty s 5 X(t) ew X, le) dt it « -yot X,lu) § cee at te X, lw) Xlw) = Xv) Xv), Qed. Cc NoTE: This is one of the mast important properties we with learn this semester. — For an LST system u, xe) fH be yt) = et eh), (+ says thet Yo) = Xe) Hus) , where Yue $Ly ley] and Hw) = SOKA) pace 4-45fice Ym iE xf) Sxl) and welt) > X, le), then KK) 25 AL Xiu) x ) | Proof + nel) = Fixer) F [x] [aeons Jeane) = ans ai [¢ Xi) (8) ent. dp =e Sal ) W(X Ke oD Eclp 2h w= uth a= wf dwede dw - (ok §7x.(p) % (ue) eg Jas an J Nt 2f6) X ma Change Order (eal [ SSate) x o-0Dap Jel aw of integration = am) m u &| & [ xt) # x.t0)] e* ds — ar [rte # xa 0d] § aed, pace 4.46Pacseval’s Formula; TE xlt) Ly Xl) and Keo) Is square integrable, eS 2 then f Txcol? dt = af [xtol de . Proof * By hypothess, x (t)= aS Xu) er * dw. so xP) = xteyertt de. a Now, (ior at = 5 x) xt(t) dt tt) [a § tea | dt Change Order a an cts > 3z (xt [ xe ae | do Lx) |, : au J X*eu) Xen) _ = aI bel de. @en, pace 4.47DistRiputionaL Fourie Transforms ~A general stgnal x(t) can be written a5 KU) = Xa (4) + Xp ), where —> Xalt) is aperiodic —> Xe lt) is periodic. —IF x(t) is not zero, then the Fourier transform XWoy= $Lx(Y] mush contain Dirac deltas. => In other words, if x(t) contains a periodic Component, then X(w) will contain Dirac deltas. — Before looking at periodic signals, lets derive the Fourier transform of F(t). Forward Trans form F[ew)] = Lome at =
-we) Wo . = FX Low-we) - pturwe)\ + Geet — More transform pairs tht we don't have time OC to derive: tle t a2 ey fare ele ~ at om tt) ety Ge)” t 4 aw tult) <7 jTsw)- $ 2 t® <—% 2M" Te Sw) Pace 4.55-When do Dirac deltas show up in the Fourier transform 7 aif x) So@ periodic signal, then K(w) Will he made up exclusively Ce Ditee deltas in 1. In other words, Hrere will be nothing but “ella fune-cttens” in Klo), —> if xt) isa purely aperiodic signal, then Frere will not be any Dirae deltas in Xl), — For a general signal %(t) that bas 4 pevielic Part Xpert) and an aperiodic part Naper (t), So that U4) = per Cd) + Yager lt) = The Fourier transform will be Xtw) = FE xa)§ = Ff Xeper (tH) + X per (4)5 = Fi epee (113 t Fi xper (OF aa ho Direc deltes rethiag a PAGE 4,55ADuality Iwterpretation : when a functan is made up of nothing but Dirac deltas, you con think of Itasca discrete function, where the samples are. pus to the weights of the Divac deltas, EX: xlt)= 384)+28(t-1) + d (t-2) Ba a O12 rt > lov can think of this as being eguivalent to UCnd = 390 + 2H EV-I + PLn-2) drt, n Cr 2 FACT; A signal that is pevicdic in one domain is discrete in the other. PAGE 4,55BThis is an important quality property of the Fourier trans farm. —> Et says this: +> if xlt) is periedic, then X(w) is discrete Croth ing but Dirac Aelks). aa iG x(t) is discrete. Croth ing but Direc de tas), then Xl») is periodic, HA nother way to think of this; 2 In general, any signal XE) can be though F of as having a periodic part Xpart) and an APericdic part nerd) (wore: either part might be 2er> in some cases]. => The aperiodic Part has a Fouvier- transbrm that is an ordinary function... no Dirac Ae los (iste sbections ) => The pericdic. part has a purely dictributionad Fourier transform, PAGE 4.55C=> Another interpretation of this is thet: the periadic part ef Xt) dheentt have. Q Fourier transform that& an ordinary Riemann integral, — We can emsicer it to be a distributional integra, — Or we can write the periodic part OF XG) in a Fourier series instead of @ Fourier integral. => IP we do this, the Fovrie~ Coefficients in the series wil | be Numevically egual to the weiguts of the Dirac deltas in the distributional Fxurie- Integra |, Pace 4.55 DBSo, for a general xX) or Xie), we can break It into parts like this; > A part tat is a plane old square- integrable Lumtion iw me demain cory espends to a plane ad Squere-integrable Pavt im the other domain, — A part +Hret is Periodic in one domain Corvespords tog part ih the other chmain that is purely clistributionad-.- nothing but Dirac de Has, This dictributionll Part can alternatively be tought of oS a Fourier series, — A part that is nothing bat Divac deltas in me domain covresperds toa purely Periodic Part in the other clown PAGE 4.55E— Final comments on the. Discrete <— > Periodic. duality Property of the Fourier trans Qe; © In chapter 5 we will take Foor ter transforms of discrete signals XH). = we will see thet these Four'ey— Transferms ave always periodic, @ When a siqna| is both pericclic and oliscrete in one chmes'n itmust also be both pericdic ard discrete the other awoin, = This is called the discrete Fourie transform CdET). I+ is not covered formally in the beok. => The fast Fourier transform (FFT) is a Last algorithn or Computing te DFT. Paws 4.55 FFREQUENCY DOMAIN DESCRIPTION OF LST SYSTEMS 43 On page 458) we observed that the saakeaet Signals fe See are all eigenfunctions of any LST Gontinuous-time system, - Given the eigenvalues for each wy it was easy to write down the system output when the input was a complex exponential. —On page 2 » we also showed that, in @ certain sense, the set L eT eR is an “orthonormal” basis fur a very large space of signals (up +o tt bothersome Ze factor! ). we reasoned that, given the eigenvalues, we should be able to write an arbitrary signal fea a linear composition of this basis and then use the linearity of the system to easily find the output when x(t) was the input. - We wrote Ht) for the set of eigenvalues. 4.3 . Specifically, on page BD, we saw thal Hw) = § Ree w™ ac, se where h(t) was the system impulse response. pace 456SS r—— =F system Frequency Tespanse , which gives the eigenvalues, is precisely the Fourier transform of the impulse response : How) = Smee = FT net | Hus) ets k(t) —For an LSL system H with input x(t) and output y(t) xt) LH =v), the convolution Property of the Fourier trans form teils us) that Yio) = Xlw) HQ), where Yiw) i expresses the output y¥lt) as @ linear Composition of Complex sinusoids. Xlw) 3 expresses the input x(t) as a linear Composition of complex sinusords | Hs): Gives the eigenvalues thet tel] how each individual complex sinusoid gets Scaled as tt passes through the system. paoe 457~ So we ave Thinking of Xt] as being a sum of terms like X(ur)eart Xtreme, and Klos) edert e. 20 . 2 x= EY Xwedtdes. — When x(t) is inpat to an LTL system H, each me of these terms gets multiplied yan e’genvalue when i+ goes through the. system: Inpat term Output te, in x(t) => Makes — in yt) a Xba) eit X(w1)Hlon) ed rE X(w2) esvet Xue )H lxje tet X (ws) est Xun) Hew, est — And we are thinking of yt) as being q sum of all the terms of this form; js eee ylt)= df Yue? tJ = Fa] XCone Je a Le.) Yew) = Xa) Hw), PAGE 4.57AIN PICTURES: ~1i(? _ ~“ 2W)= anf xWeis pub y= 20) Yoredths Xuyesrt > Xl) Hw.) eit Xaner® —— H X (ur) His.) eh Xt) ewat — h(t) X Gs) H(s,)ednst y (t) = & ( Xl) Hlw) ed as = | => Yume XWH&) PAGE 4.57BEX*+ His an LTL system with impulse response. hi) =e*tult)) aeR, aro. The input jg xa) = eu it), beR, b>o, b#a. Find ylt). avant een aCs) eam Table: Ha) = aps = = ——___. Yt) = Xl) Hie) = (atjw)Cb+j) — We use a partial fraction expansion to simplity You) into a sum of terms that are all in Table 4,2: &) l = A Yiu) = (at w)Cbt jw) 7 ati bre — > We have to colve Br the constants A and B in C¥), —) The equality m CK) has to hold forall choices of w, including complex choices, — PAGE 4.58—> We can avoid the pitfalls of complex avrithmetc hy Choosing w=-}8 for an arbitrary OEIR. —~ EQ. (*) on PAGE 4.58 then becomes | = A_, B (ate\(b+e) are t+8 (**) A Solve (¥4%) for A and B: —3To get A, multiply both sides of Gr¥) by (a+¢) and evaluate at @=-2 : Bate) b+® [pea A+ Tre |ge-a >To get B, multiply both sides of CH¥) by (br8) and evaluate at g=-b: os : Aer) +B are lp=-b ate |o=-b PAGE 4.58A— Now plug the solutions f A and B back into Eg. GH) on PAGE 4.58: A B Yt) = ee ot oe ax jw bri ot ,i- 4 = ba atjw a-b btyw | oS sKL at ie as — By table lookup and the lenearsty property, y= Ff Yost = oa (ewe ult) EK + Same as before, except this time azb. nit)= ev Fut), xie)= ev ult), GER ) Coy) © Table « X(v)= Hle) = eae Ytwy a Xl) HW) a (atj~)* PAGE 4.58BThis time Yur) is in the table, and we have immediately Het yl) = teu) wae NoTE* In this case we can alternatively get the answer by applying the frequency ditferentiation property, >We already know Hat eu) &S ap ;: 7 we observe that _ i ee leer Yiu) = Carpe) — §du ary — Arplying the frequency di clerentiation properly directly to the pair etudt) sy ay 2 we have -jte*Lin eral (aty)* ol diy atin Multiply both sides by d? F tetuw = Yew) (atjw)* > ytt)= te*tray a PAGE 4.59FILTERING IwTeRPRE TATION a eee eee Le “In the frequency domain) the output of an LSE system ‘5 the product of the input with the system frequency response: Ne Xl) Huw), — This can greatly simplify the analysis problem of Figuring — ' out what the output is for a given inpat (as @upared to doing convolution in the time domam ), ~ Tt also simplifies the design problem. EX: On the battlefield, there is electromagnetic energy at frequencies w between w, and w2. ~All enemy aircraft use radar Processors with oscillators that emit energy at the freguency We where ’ Wy < We < we. ~ Design an LSI system to detect +he presence of enemy —aircratt. wn enemy aircraft. bettefield clutter a a >wW be Me Pace 4.60esoution. el 4 we % Hw) = S(w- ve) ~use the inverse Fourier transform to design hit). — htt)= F'[ow-we)] = eM. — This type of approach is used in modern police radar detectors. - The system output is large when the enemy is present. EX: A telephone line is being degraded by interference from a GOHE AC power supply. ~Design a filter to remove the noise. | Re) | boule. This is called a “notch Filter! Fage 4.6}“Suppose H is an LSE Sy Stem Hew) rm. — Suppose X(w) looks likes panne — Then Yous) looks likes with frequency response eel passes low treguencies, but blocks high Rrequencies — This kind of LSE system is called a * Low-Pass Bilter Pace 4.62— Frequency response of a high-pass Filter: Hw) — Frequency response of a bandpass Filter: Hw) We % Series Connection of two LST Systems 4 x(t) (a tH Wy y tt) h(t) = h(t) © hy (E) Hwy = Hy (w) UH. Cw) pace +63Parallel Connection of two LST Systems GIy | ye Gal x(t) We= ke fe) he C#) Ho) = Hye) + He Cw) Thvertible Systems ~ Suppose G is the inverse syctem of an LST system He xe) {4 Lofe He xt) Then F is the identity system, and hes Impulse Tespense f(t)= ott) ~ Then Fo) = SC aity)= 4. =o FWw)= HwjyGw)= 4. 1 ~ Ta other words, Gl) = 7 Ww) Ths, &n LSE system Ho hes an inverse 1€ Too exists . pace 4.6Feedback Connection %®) —~9—I_ 4 | eT His) oOo a 1+ GluJHl) Prot « F Zlt)= yl kg lt) > Zlu)= YnrG) Cy) ylt) = [xte)- 2] * hd) et, Yeu) = [xtw)- 2em) Hew) Plug 9 C8) © Yay 2 (xt) - You) Geo) ] Hew) Yl) = X lw) Hl) — Yom Gl) Hl) Yew) + Yow)Gld) Hw) = X Lv) Hs) Yoo) [1+GM)HW)J = Xl) HOu) Elw)= Yu) H(w) Xt) 4+ Glu) Hw) PAGE 464A,EX: The in pat XE) and output y(t) of an LIL System Hare relted by the differential eguation 4") + 4y') +3ylt)= x/A)+ 2x), Find the impulse regpense h(t), Taking Fourier trancforms on both sides, we have Ci? Yiu + 4 jus Youd) + 3Y jus) = j?Xl) + 2X) Lapse 450 +3] Yu) = [yor 2] X (ew) Heys We 2 ete jue X(w) ~ (oP 4jo +3 ~ Gers Qqet) —Vactal Fractions: O+2 a (ors\(er!) +3 B+! - +2 as eeas A= pr lpe-s [_ = * az O+2 a B= +3 3 |p--) = 2 - |p Ne His ay + jert h(t)= 2e™ UH) +RE Tul) ————————___ pao 4.65EX: For the preceeding example, Find he output yt) when the input is given by x)= e-tult), Table: X(w)= rare Ywy= XlHte) 2 Jet jE = A,B 4’ GQutt)*(yst3) quel Gury + & m3 Fartia| Fractens + O42 _ A B Fe (0+ 1)*(0+3) d+! * fon? Ord ele multiply both sides by 8+3 and evaluate at @= 3: Ora ea Ihe SI ola | 4 (9+3)B “3 69 les” Gath, C Lis ZERO ZERO - 73tZ2 Ear * G31)? = ~ z — For 8 multiply both sides by (oti)? and evaluete at O+2 6=-]5 o3 [pe = OMA] vat Conte oa b=-1 Ors . ot re | ZERO ZERO -1+2 iL B= E ZF —_—>> PAGE 4.663 Strategy for A: O multiply both sides by (ery, > The lett side with have nO Otl) in the denemin ecto — The right side will have a “B" torn, with Ww 8 let, a XH term that is Fuadratic in (641), and an 4" tera at 'S linear in &, @ Take 5 on both sides. > The left side will sti in the denominaty > On the right side, > The * — The 'Lheve no (ett) r B! term will be game. C" term will stil] have (ot!) in the Numerator. > The Au tern, will no longer have Q. @ Evaluate both sides at ®= Solution Lo- A. => THIs STRATEGY “! to get a ALWAYS WoRKS, PAGE 4.679 = Find As raultiply both sides by (Bt!)%, take $y and evaluate at Q=-|: rT _(o+2)o+1)? 7 3 _fa 2 3 eae (0+1)* (943) 7 = (3 coral, a +7, 8 ae 2 (orH¥eC i ets os | quotient rule" 4 *y Hat the » ~ To awid using tre aS th) cl w be used; [33 (@+2)(9+3)"! I _ = [3 (or) A +O = [onto] e+] x Tewrite Spreduc vule” can [(omenersy? 1(0+3)| =A +[enr (-)(0+3)% +2(6+)( lortfc - o>) “ CdCrsy*+Cay']= At [oro] rs : i A “lye ata A= 7 PAGE 4.68“Swe hve Ae, B= 4 C= -d Yu) = 4 a jy Cite) ape : = Lot 7 Table: 44) = e* ult) + ttetuw -he 4) Note + Partial fractions is an “ag woo! technigue, There is more than me way to do it, The method presented here is called the “Yeaviside coverup method. It may not he the shortest solutien, but it always works and it aluays works the same way - PACE 4.69— The general form of the input/output equation far an LTL system governec| by a content coefficients Ordinary linea di€ ferential equation iss N M at - d" Dagaye = 2 bogeant, where the A” and bi! ave constants. Tn other words, a linear combination of The derivatives of the cutput [s egual to & linear combinatian of the derivatives of the input. 7 By takeing Favier transforms on both sides, we can derive a general solution tu- the Frequency response of any system of this type, PAGE 4°704f Easy}: J fe EAs auf bad {gaat = 2 Hx Hs $10, ( io) Yu) = ou Ge)" X le) k=o0 Yow) ¥ agli = Xlx) Yb tr k=O m=d ” A ba Gye) Hue Ls a! x) Fay iv) K=o => Wwe are dme with Chapter 4. PAGE 451|
You might also like
Emd Lab PDF
PDF
No ratings yet
Emd Lab PDF
15 pages
DAA UNIT 4 (Dynamic Programming)
PDF
No ratings yet
DAA UNIT 4 (Dynamic Programming)
16 pages
D and F Block Elements
PDF
No ratings yet
D and F Block Elements
16 pages
acs part 1
PDF
No ratings yet
acs part 1
25 pages
19umav071 - Real Analysis I - 18u5mac10
PDF
No ratings yet
19umav071 - Real Analysis I - 18u5mac10
13 pages
Unit 3 FIR Filter DSP
PDF
No ratings yet
Unit 3 FIR Filter DSP
18 pages
CS 5392 Final Exam (A05294220)
PDF
No ratings yet
CS 5392 Final Exam (A05294220)
8 pages
OS Unit2-1
PDF
No ratings yet
OS Unit2-1
18 pages
System of Rotation
PDF
No ratings yet
System of Rotation
15 pages
Adobe Scan 03 Jan 2025
PDF
No ratings yet
Adobe Scan 03 Jan 2025
18 pages
JS Notes 2 Mr900
PDF
No ratings yet
JS Notes 2 Mr900
14 pages
Equilibrium
PDF
No ratings yet
Equilibrium
26 pages
Adobe Scan 5 Sep 2023 (1)
PDF
No ratings yet
Adobe Scan 5 Sep 2023 (1)
19 pages
Sem 1 Practical, Atsc (3)
PDF
No ratings yet
Sem 1 Practical, Atsc (3)
21 pages
Adobe Scan 24 Nov 2022
PDF
No ratings yet
Adobe Scan 24 Nov 2022
9 pages
DCS Assignment PDF
PDF
No ratings yet
DCS Assignment PDF
20 pages
ECO 722 Lec1.PDF
PDF
No ratings yet
ECO 722 Lec1.PDF
32 pages
Unit 1
PDF
No ratings yet
Unit 1
43 pages
Mathematics-IB LAQ Imp
PDF
No ratings yet
Mathematics-IB LAQ Imp
11 pages
Derivatives and AOD
PDF
No ratings yet
Derivatives and AOD
18 pages
Module5 DSA 23 24
PDF
No ratings yet
Module5 DSA 23 24
20 pages
EET401-M4
PDF
No ratings yet
EET401-M4
19 pages
Unit 3
PDF
No ratings yet
Unit 3
18 pages
Unit_2-M3
PDF
No ratings yet
Unit_2-M3
24 pages
ATD Mod.01 - Solved Question Bank
PDF
No ratings yet
ATD Mod.01 - Solved Question Bank
18 pages
ATC Module-1 Notes
PDF
No ratings yet
ATC Module-1 Notes
58 pages
SEM - Electrical Measurement & Measuring Ins
PDF
No ratings yet
SEM - Electrical Measurement & Measuring Ins
39 pages
Interference Notes Part1 Nishu
PDF
No ratings yet
Interference Notes Part1 Nishu
29 pages
Tài Liệu Viết Tay
PDF
No ratings yet
Tài Liệu Viết Tay
75 pages
MWOC B Sec
PDF
No ratings yet
MWOC B Sec
15 pages
Saiful Rizam
PDF
No ratings yet
Saiful Rizam
10 pages
Quantum
PDF
No ratings yet
Quantum
14 pages
Pega Notes 2015 PDF
PDF
100% (3)
Pega Notes 2015 PDF
102 pages
TOPS Unit 3
PDF
No ratings yet
TOPS Unit 3
24 pages
Ce6451 - Fluid Mechanics and Machinery Notes
PDF
94% (16)
Ce6451 - Fluid Mechanics and Machinery Notes
466 pages
PDC U3 Rossihini
PDF
No ratings yet
PDC U3 Rossihini
20 pages
CVE 313 Lecture 5
PDF
No ratings yet
CVE 313 Lecture 5
5 pages
Adobe Scan Apr 26, 2023
PDF
No ratings yet
Adobe Scan Apr 26, 2023
11 pages
Pro D: CHQP
PDF
No ratings yet
Pro D: CHQP
30 pages
PDC U1 Rossihini
PDF
No ratings yet
PDC U1 Rossihini
24 pages
GE-2 Chemistry Notes Unit-2 (3rd & 4th Sem)
PDF
No ratings yet
GE-2 Chemistry Notes Unit-2 (3rd & 4th Sem)
25 pages
LCS Unit-4
PDF
No ratings yet
LCS Unit-4
14 pages
Notes 1
PDF
No ratings yet
Notes 1
25 pages
important derivations final xi
PDF
No ratings yet
important derivations final xi
6 pages
Book Mar 08, 2025
PDF
No ratings yet
Book Mar 08, 2025
20 pages
Digital Modulation Namaste
PDF
No ratings yet
Digital Modulation Namaste
20 pages
EMT Module1 Merged
PDF
No ratings yet
EMT Module1 Merged
77 pages
Conic Sections PDF
PDF
No ratings yet
Conic Sections PDF
42 pages
Industrial Engineering
PDF
No ratings yet
Industrial Engineering
14 pages
Adobe Scan Apr 23, 2024
PDF
No ratings yet
Adobe Scan Apr 23, 2024
18 pages
Adobe Scan Apr 18, 2025 (1)
PDF
No ratings yet
Adobe Scan Apr 18, 2025 (1)
19 pages
Ed.g.0805.21 Wasike Elphas
PDF
No ratings yet
Ed.g.0805.21 Wasike Elphas
9 pages
Oscillation S
PDF
No ratings yet
Oscillation S
12 pages
Design of Machine Elements Intern 1
PDF
No ratings yet
Design of Machine Elements Intern 1
14 pages
Circuit theorems
PDF
No ratings yet
Circuit theorems
23 pages
Som 1-25
PDF
No ratings yet
Som 1-25
25 pages
E-Leetspeak: All New! the Most Challenging Puzzles Since Sudoku!
From Everand
E-Leetspeak: All New! the Most Challenging Puzzles Since Sudoku!
Thomas Ferrante
No ratings yet
Raghu Engineering College
PDF
No ratings yet
Raghu Engineering College
30 pages
Additive Manufacturing: Singuru Rajesh Blogspot
PDF
No ratings yet
Additive Manufacturing: Singuru Rajesh Blogspot
3 pages
Innovative Sharing of Notes and Material Through Online Medium
PDF
No ratings yet
Innovative Sharing of Notes and Material Through Online Medium
1 page
Green Engineering Systems
PDF
No ratings yet
Green Engineering Systems
3 pages
3RD Year Backlogs Sec D
PDF
0% (1)
3RD Year Backlogs Sec D
2 pages
Faculty PDF
PDF
No ratings yet
Faculty PDF
12 pages
List of 3-1 Subjects and Labs
PDF
No ratings yet
List of 3-1 Subjects and Labs
1 page
Curriculum Vitae: Objective
PDF
No ratings yet
Curriculum Vitae: Objective
2 pages
Faculty PDF
PDF
No ratings yet
Faculty PDF
12 pages
Stepped Bar
PDF
No ratings yet
Stepped Bar
5 pages
Att Print 1
PDF
No ratings yet
Att Print 1
2 pages
EXPERIMENT 2: Principle & Characteristics of Strain Gauge
PDF
No ratings yet
EXPERIMENT 2: Principle & Characteristics of Strain Gauge
4 pages
MCQ Ppe Unit-1
PDF
No ratings yet
MCQ Ppe Unit-1
4 pages
Power Plant Engineering
PDF
No ratings yet
Power Plant Engineering
74 pages
New Doc 2018-06-11
PDF
No ratings yet
New Doc 2018-06-11
1 page
Jntu Anan Civil 2 2 HHM Set 4
PDF
No ratings yet
Jntu Anan Civil 2 2 HHM Set 4
8 pages