ebook img

0 Nonlinear Analysis and Design of Phase-Locked Loops - InTech PDF

28 Pages·2012·0.38 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview 0 Nonlinear Analysis and Design of Phase-Locked Loops - InTech

(cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:4)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:4)(cid:8)(cid:5)(cid:12)(cid:13)(cid:6)(cid:13)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:15)(cid:7)(cid:13)(cid:6)(cid:16)(cid:4)(cid:10)(cid:3)(cid:17)(cid:10)(cid:18)(cid:19)(cid:8)(cid:13)(cid:7)(cid:20)(cid:21)(cid:3)(cid:22)(cid:23)(cid:7)(cid:14)(cid:10)(cid:21)(cid:3)(cid:3)(cid:24)(cid:13) (cid:25)(cid:26) (cid:2) 0 Nonlinear Analysis and Design of Phase-Locked Loops G.A.Leonov,N.V.KuznetsovandS.M.Seledzhi Saint-PetersburgStateUniversity Russia 1. Introduction Phase-lockedloops(PLLs)arewidelyusedintelecommunicationandcomputerarchitectures. They were invented in the 1930s-1940s (De Bellescize, 1932; Wendt & Fredentall, 1943) and thenintensivestudiesofthetheoryandpracticeofPLLswerecarriedout(Viterbi,1966;Lind- sey, 1972; Gardner, 1979; Lindsey and Chie, 1981; Leonov et al., 1992; Encinas, 1993; Nash, 1994;Brennan,1996;Stensby,1997). One of the first applications of phase-locked loop (PLL) is related to the problems of data transfer by radio signal. In radio engineering PLL is applied to a carrier synchronization, carrierrecovery,demodulation,andfrequencysynthesis(see,e.g.,(Stephens,2002;Ugrumov, 2000)). After the appearance of an architecture with chips, operating on different frequencies, the phase-lockedloopsareusedtogenerateinternalfrequenciesofchipsandsynchronizationof operationofdifferentdevicesanddatabuses(Youngetal., 1992; Egan, 2000; Kroupa, 2003; Razavi, 2003; Shu&Sanchez-Sinencio, 2005; Manassewitsch, 2005). Forexample, themod- erncomputermotherboardscontaindifferentdevicesanddatabusesoperatingondifferent frequencies, which are often in the need for synchronization (Wainner & Richmond, 2003; Buchanan&Wilson,2001). Anotheractual applicationof PLLis theproblem ofsaving energy. One ofthe solutionsof thisproblemforprocessorsisadecreasingofkernelfrequencywithprocessorload. Thein- dependentphase-lockedloopspermitonetodistributemoreuniformlyakernelloadtosave theenergyandtodiminishaheatgenerationonaccountofthateachkerneloperatesonits ownfrequency.Nowthephase-lockedloopsarewidelyusedforthesolutionoftheproblems of clock skew and synchronization for the sets of chips of computer architectures and chip microarchitecture. For example, a clock skew is very important characteristic of processors (see,e.g.,(Xanthopoulos,2001;Bindal,2003)). Variousmethodsforanalysisofphase-lockedloopsarewelldevelopedbyengineersandare consideredinmanypublications(see,e.g.,(Banerjee,2006;Best,2003;Kroupa,2003;Bianchi, 2005;Egan,2007)),buttheproblemsofconstructionofadequatenonlinearmodelsandnon- linearanalysisofsuchmodelsarestillfarfrombeingresolvedandrequireusingspecialmeth- odsofqualitativetheoryofdifferential,difference,integral,andintegro-differentialequations (Gelig et al., 1978; Leonov et al., 1996a; Leonov et al., 1996b; Leonov & Smirnova, 2000; Abramovitch, 2002; Suarez & Quere, 2003; Margaris, 2004; Kudrewicz & Wasowicz, 2007; Kuznetsov,2008;Leonov,2006).Wecouldnotlisthereallreferencesintheareaofdesignand www.intechopen.com (cid:26)$ (cid:11)(cid:27)(cid:28)(cid:29)(cid:30)(cid:11)(cid:28)(cid:31)(cid:29)(cid:2)(cid:10) (cid:10)!(cid:29)(cid:2)(cid:28)"(cid:29)(cid:21)(cid:10)(cid:20)(cid:10)(cid:28)(cid:19)(cid:7)(cid:3)(cid:9)(cid:12)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:18)(cid:9)(cid:8)(cid:22)#(cid:6)(cid:22)(cid:7) analysisofPLL,soreadersshouldseementionedpapersandbooksandthereferencescited therein. 2. MathematicalmodelofPLL InthisworkthreelevelsofPLLdescriptionaresuggested: 1)thelevelofelectronicrealizations, 2)thelevelofphaseandfrequencyrelationsbetweeninputsandoutputsinblockdiagrams, 3)thelevelofdifference,differentialandintegro-differentialequations. Thesecondlevel, involvingtheasymptoticalanalysisofhigh-frequencyoscillations, isnec- essary for the well-formed derivation of equations and for the passage to the third level of description. ConsideraPLLonthefirstlevel(Fig.1) Fig.1.BlockdiagramofPLLonthelevelofelectronicrealizations. HereOSCmaster isamasteroscillator,OSCslave isaslave(tunable)oscillator,whichgenerate high-frequency"almostharmonicoscillations" f (t)= A sin(ω (t)t+ψ) j=1,2, (1) j j j j whereA andψ aresomenumbers,ω (t)aredifferentiablefunctions.Block isamultiplier j j j ofoscillationsof f (t)and f (t)andthesignal f (t)f (t)isitsoutput. Therelationsbetween 1 2 1 2 (cid:2) theinputξ(t)andtheoutputσ(t)oflinearfilterhavetheform t σ(t)=α (t)+ γ(t τ)ξ(τ)dτ. (2) 0 − (cid:3) 0 Hereγ(t)isanimpulsetransientfunctionoffilter,α (t)isanexponentiallydampedfunction, 0 depending on the initial data of filter at the moment t = 0. The electronic realizations of generators, multipliers, and filters can be found in (Wolaver, 1991; Best, 2003; Chen, 2003; Giannini & Leuzzi, 2004; Goldman, 2007; Razavi, 2001; Aleksenko, 2004). In the simplest caseitisassumedthatthefilterremovesfromtheinputtheuppersidebandwithfrequency ω (t)+ω (t)butleavesthelowersidebandω (t) ω (t)withoutchange. 1 2 1 − 2 Nowwereformulatethehigh-frequencypropertyofoscillations f (t)andessentialassump- j tion that γ(t) and ω (t) are functions of "finite growth". For this purpose we consider the j great fixed time interval [0,T], which can be partitioned into small intervals of the form [τ,τ+δ],(τ [0,T])suchthatthefollowingrelations ∈ γ(t) γ(τ) Cδ, ω (t) ω (τ) Cδ, | − |≤ | j − j |≤ (3) t [τ,τ+δ], τ [0,T], ∀ ∈ ∀ ∈ www.intechopen.com (cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:4)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:4)(cid:8)(cid:5)(cid:12)(cid:13)(cid:6)(cid:13)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:15)(cid:7)(cid:13)(cid:6)(cid:16)(cid:4)(cid:10)(cid:3)(cid:17)(cid:10)(cid:18)(cid:19)(cid:8)(cid:13)(cid:7)(cid:20)(cid:21)(cid:3)(cid:22)(cid:23)(cid:7)(cid:14)(cid:10)(cid:21)(cid:3)(cid:3)(cid:24)(cid:13) (cid:26)% ω (τ) ω (τ) C , τ [0,T], (4) | 1 − 2 |≤ 1 ∀ ∈ ω (t) R, t [0,T] (5) j ≥ ∀ ∈ aresatisfied.Hereweassumethatthequantityδissufficientlysmallwithrespecttothefixed numbersT,C,C ,thenumberRissufficientlygreatwithrespecttothenumberδ. Thelatter 1 meansthatonthesmallintervals[τ,τ+δ]thefunctionsγ(t)andω (t)are"almostconstants" j andthefunctions f (t)rapidlyoscillateasharmonicfunctions. j ConsidertwoblockdiagramsshowninFig.2andFig.3. Fig.2.Multiplierandfilter. Fig.3.Phasedetectorandfilter. Hereθ (t) = ω (t)t+ψ arephasesoftheoscillations f (t),PDisanonlinearblockwiththe j j j j characteristic ϕ(θ)(beingcalledaphasedetectorordiscriminator). Thephasesθ (t)arethe j inputs of PD block and the output is the function ϕ(θ (t) θ (t)). The shape of the phase 1 − 2 detectorcharacteristicisbasedontheshapeofinputsignals. Thesignals f (t)f (t)andϕ(θ (t) θ (t))areinputsofthesamefilterswiththesameimpulse 1 2 1 − 2 transientfunctionγ(t).Thefilteroutputsarethefunctionsg(t)andG(t),respectively. AclassicalPLLsynthesisisbasedonthefollowingresult: Theorem1. (Viterbi,1966)Ifconditions(3)–(5)aresatisfiedandwehave 1 ϕ(θ)= A A cosθ, 2 1 2 thenforthesameinitialdataoffilter,thefollowingrelation G(t) g(t) C δ, t [0,T] | − |≤ 2 ∀ ∈ issatisfied.HereC isacertainnumberbeingindependentofδ. 2 www.intechopen.com (cid:26)& (cid:11)(cid:27)(cid:28)(cid:29)(cid:30)(cid:11)(cid:28)(cid:31)(cid:29)(cid:2)(cid:10) (cid:10)!(cid:29)(cid:2)(cid:28)"(cid:29)(cid:21)(cid:10)(cid:20)(cid:10)(cid:28)(cid:19)(cid:7)(cid:3)(cid:9)(cid:12)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:18)(cid:9)(cid:8)(cid:22)#(cid:6)(cid:22)(cid:7) ProofofTheorem1(Leonov,2006) Fort [0,T]weobviouslyhave ∈ g(t) G(t)= − t = γ(t s) A A sin ω (s)s+ψ sin ω (s)s+ψ − 1 2 1 1 2 2 − (cid:3) (cid:4) (cid:5) (cid:8) 0 (cid:6) (cid:7) (cid:6) (cid:7) ϕ ω (s)s ω (s)s+ψ ψ ds= − 1 − 2 1− 2 (cid:5) (cid:8)(cid:9) t A A = 1 2 γ(t s) cos ω (s)+ω (s) s+ψ +ψ ds. − 2 − 1 2 1 2 (cid:3) (cid:4) (cid:5) (cid:8)(cid:9) 0 (cid:6) (cid:7) Considertheintervals [kδ,(k+1)δ], where k = 0,...,m andthenumber m issuchthat t ∈ [mδ,(m+1)δ].Fromconditions(3)–(5)itfollowsthatforanys [kδ,(k+1)δ]therelations ∈ γ(t s)=γ(t kδ)+O(δ) (6) − − ω (s)+ω (s)=ω (kδ)+ω (kδ)+O(δ) (7) 1 2 1 2 arevalidoneachinterval[kδ,(k+1)δ].Thenby(7)foranys [kδ,(k+1)δ]theestimate ∈ cos ω (s)+ω (s) s+ψ +ψ =cos ω (kδ)+ω (kδ) s+ψ +ψ +O(δ) (8) 1 2 1 2 1 2 1 2 (cid:5) (cid:8) (cid:5) (cid:8) (cid:6) (cid:7) (cid:6) (cid:7) isvalid.Relations(6)and(8)implythat t γ(t s) cos ω (s)+ω (s) s+ψ +ψ ds= − 1 2 1 2 (cid:3) (cid:4) (cid:5) (cid:8)(cid:9) 0 (cid:6) (cid:7) (9) (k+1)δ m = ∑ γ(t kδ) cos ω (kδ)+ω (kδ) s+ψ +ψ ds+O(δ). − 1 2 1 2 k=0 (cid:3) (cid:4) (cid:5) (cid:8)(cid:9) kδ (cid:6) (cid:7) From(5)wehavetheestimate (k+1)δ cos ω (kδ)+ω (kδ) s+ψ +ψ ds=O(δ2) 1 2 1 2 (cid:3) (cid:4) (cid:5) (cid:8)(cid:9) kδ (cid:6) (cid:7) andthefactthatRissufficientlygreatascomparedwithδ.Then t γ(t s) cos ω (s)+ω (s) s+ψ +ψ ds=O(δ). − 1 2 1 2 (cid:3) (cid:4) (cid:5) (cid:8)(cid:9) 0 (cid:6) (cid:7) Theorem1iscompletelyproved.(cid:2) www.intechopen.com (cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:4)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:4)(cid:8)(cid:5)(cid:12)(cid:13)(cid:6)(cid:13)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:15)(cid:7)(cid:13)(cid:6)(cid:16)(cid:4)(cid:10)(cid:3)(cid:17)(cid:10)(cid:18)(cid:19)(cid:8)(cid:13)(cid:7)(cid:20)(cid:21)(cid:3)(cid:22)(cid:23)(cid:7)(cid:14)(cid:10)(cid:21)(cid:3)(cid:3)(cid:24)(cid:13) (cid:26)' Fig.4.BlockdiagramofPLLonthelevelofphaserelations Thus,theoutputsg(t)andG(t)oftwoblockdiagramsinFig.2andFig.3,respectively,differ littlefromeachotherandwecanpass(fromastandpointoftheasymptoticwithrespecttoδ) tothefollowingdescriptionlevel,namelytothesecondlevelofphaserelations. InthiscaseablockdiagraminFig.1becomesthefollowingblockdiagram(Fig.4). Considernowthehigh-frequencyimpulseoscillators,connectedasindiagraminFig.1.Here f (t)= Asign(sin(ω (t)t+ψ)). (10) j j j j Weassume,asbefore,thatconditions(3)–(5)aresatisfied. Consider2π-periodicfunctionϕ(θ)oftheform A A (1+2θ/π) for θ [ π,0], ϕ(θ)= 1 2 ∈ − (11) (cid:10)A1A2(1−2θ/π) for θ∈[0,π]. andblockdiagramsinFig.2andFig.3. Theorem2. (Leonov,2006) Ifconditions(3)–(5)aresatisfiedandthecharacteristicofphasedetector ϕ(θ)hastheform(11),then forthesameinitialdataoffilterthefollowingrelation G(t) g(t) C δ, t [0,T] | − |≤ 3 ∀ ∈ issatisfied.HereC isacertainnumberbeingindependentofδ. 3 ProofofTheorem2 Inthiscasewehave g(t) G(t)= − t = γ(t s) A A sign sin ω (s)s+ψ sin ω (s)s+ψ − 1 2 1 1 2 2 − (cid:3) (cid:4) (cid:5) (cid:8) 0 (cid:6) (cid:7) (cid:6) (cid:7) ϕ ω (s)s ω (s)s+ψ ψ ds. − 1 − 2 1− 2 (cid:9) (cid:6) (cid:7) www.intechopen.com (cid:26)( (cid:11)(cid:27)(cid:28)(cid:29)(cid:30)(cid:11)(cid:28)(cid:31)(cid:29)(cid:2)(cid:10) (cid:10)!(cid:29)(cid:2)(cid:28)"(cid:29)(cid:21)(cid:10)(cid:20)(cid:10)(cid:28)(cid:19)(cid:7)(cid:3)(cid:9)(cid:12)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:18)(cid:9)(cid:8)(cid:22)#(cid:6)(cid:22)(cid:7) Partitioningtheinterval[0,t]intotheintervals[kδ,(k+1)δ]andmakinguseofassumptions (5)and(10),wereplacetheaboveintegralwiththefollowingsum (k+1)δ m ∑ γ(t kδ) A A sign cos ω (kδ)ω (kδ) kδ+ψ ψ − 1 2 1 2 1− 2 − k=0 (cid:4) (cid:3) (cid:4) (cid:5) (cid:8) kδ (cid:6) (cid:7) cos ω (kδ)+ω (kδ) s+ψ +ψ ds − 1 2 1 2 − (cid:5) (cid:8)(cid:9) (cid:6) (cid:7) ϕ ω (kδ) ω (kδ) kδ+ψ ψ δ . − 1 − 2 1− 2 (cid:5) (cid:8) (cid:9) (cid:6) (cid:7) Thenumbermischoseninsuchawaythatt [mδ,(m+1)δ].Since(ω (kδ)+ω (kδ))δ 1, ∈ 1 2 ≫ therelation (k+1)δ A A sign cos ω (kδ) ω (kδ) kδ+ψ ψ 1 2 1 − 2 1− 2 − (cid:3) (cid:4) (cid:5) (cid:8) kδ (cid:6) (cid:7) (12) cos ω (kδ)+ω (kδ) s+ψ +ψ ds − 1 2 1 2 ≈ (cid:5) (cid:8)(cid:9) (cid:6) (cid:7) ϕ ω (kδ) ω (kδ) kδ+ψ ψ δ, ≈ 1 − 2 1− 2 (cid:5) (cid:8) (cid:6) (cid:7) issatisfied.Hereweusetherelation (k+1)δ A A sign cosα cos ωs+ψ ds ϕ(α)δ 1 2 − 0 ≈ (cid:3) kδ (cid:11) (cid:6) (cid:7)(cid:12) forωδ 1,α [ π,π],ψ R1. Thus,T≫heorem∈2−iscomple0te∈lyproved.(cid:2) Theorem2isabaseforthesynthesisofPLLwithimpulseoscillators. Fortheimpulseclock oscillatorsitpermitsonetoconsidertwoblockdiagramssimultaneously:onthelevelofelec- tronicrealization(Fig.1)andonthelevelofphaserelations(Fig.4),wheregeneralprinciples ofthetheoryofphasesynchronizationcanbeused(Leonov&Seledzhi,2005b;Kuznetsovet al.,2006;Kuznetsovetal.,2007;Kuznetsovetal.,2008;Leonov,2008). 3. DifferentialequationsofPLL LetusmakearemarknecessaryforderivationofdifferentialequationsofPLL. Consideraquantity θ˙ (t)=ω (t)+ω˙ (t)t. j j j Forthewell-synthesizedPLLsuchthatitpossessesthepropertyofglobalstability,wehave exponentialdampingofthequantityω˙ (t): j |ω˙j(t)|≤Ce−αt. Here C and α are certain positive numbers being independent of t. Therefore, the quantity ω˙ (t)t is, asarule, sufficientlysmallwithrespecttothenumber R (seeconditions(3)–(5)). j From the above we can conclude that the following approximate relation θ˙ (t) ω (t) is j ≈ j www.intechopen.com (cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:4)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:4)(cid:8)(cid:5)(cid:12)(cid:13)(cid:6)(cid:13)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:15)(cid:7)(cid:13)(cid:6)(cid:16)(cid:4)(cid:10)(cid:3)(cid:17)(cid:10)(cid:18)(cid:19)(cid:8)(cid:13)(cid:7)(cid:20)(cid:21)(cid:3)(cid:22)(cid:23)(cid:7)(cid:14)(cid:10)(cid:21)(cid:3)(cid:3)(cid:24)(cid:13) (cid:26)) valid. InderivingthedifferentialequationsofthisPLL,wemakeuseofablockdiagramin Fig.4andexactequality θ˙ (t)=ω (t). (13) j j Notethat,byassumption,thecontrollawoftunableoscillatorsislinear: ω (t)=ω (0)+LG(t). (14) 2 2 Hereω (0)istheinitialfrequencyoftunableoscillator, Lisacertainnumber,andG(t)isa 2 controlsignal,whichisafilteroutput(Fig.4).Thus,theequationofPLLisasfollows t θ˙ (t)=ω (0)+L α (t)+ γ(t τ)ϕ θ (τ) θ (τ) dτ . 2 2 0 − 1 − 2 (cid:5) (cid:3) (cid:8) 0 (cid:6) (cid:7) Assumingthatthemasteroscillatorissuchthatω (t) ω (0),weobtainthefollowingrela- 1 ≡ 1 tionsforPLL t θ1(t)−θ2(t) ′+L α0(t)+ γ(t−τ)ϕ θ1(τ)−θ2(τ) dτ =ω1(0)−ω2(0). (15) (cid:5) (cid:3) (cid:8) (cid:6) (cid:7) 0 (cid:6) (cid:7) ThisisanequationofstandardPLL.Note,thatifthefilter(2)isintegratedwiththetransfer functionW(p)=(p+α)−1 σ˙ +ασ= ϕ(θ) thenforφ(θ)=cos(θ)insteadofequation(15)from(13)and(14)wehave θ¨˜+αθ˜˙+Lsinθ˜=α ω (0) ω (0) (16) 1 − 2 π (cid:6) (cid:7) withθ˜=θ θ + . So,ifherephasesoftheinputandoutputsignalsmutuallyshiftedby 1− 2 2 π/2thenthecontrolsignalG(t)equalszero. Arguingasabove,wecanconcludethatinPLLitcanbeusedthefilterswithtransferfunctions ofmoregeneralform K(p)=a+W(p), whereaisacertainnumber,W(p)isaproperfractionalrationalfunction.Inthiscaseinplace ofequation(15)wehave t θ1(t)−θ2(t) ′+L aϕ θ1(t)−θ2(t) +α0(t)+ γ(t−τ)ϕ θ1(τ)−θ2(τ) dτ = (17) (cid:4) (cid:3) (cid:9) (cid:6) (cid:7) (cid:6) (cid:7) 0 (cid:6) (cid:7) =ω (0) ω (0). 1 − 2 Inthecasewhenthetransferfunctionofthefiltera+W(p)isnon-degenerate,i.e. itsnumer- atoranddenominatordonothavecommonroots,equation(17)isequivalenttothefollowing systemofdifferentialequations z˙ = Az+bψ(σ) (18) σ˙ =c∗z+ρψ(σ). www.intechopen.com (cid:26)* (cid:11)(cid:27)(cid:28)(cid:29)(cid:30)(cid:11)(cid:28)(cid:31)(cid:29)(cid:2)(cid:10) (cid:10)!(cid:29)(cid:2)(cid:28)"(cid:29)(cid:21)(cid:10)(cid:20)(cid:10)(cid:28)(cid:19)(cid:7)(cid:3)(cid:9)(cid:12)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:18)(cid:9)(cid:8)(cid:22)#(cid:6)(cid:22)(cid:7) Hereσ=θ θ ,Aisaconstant(n n)-matrix,bandcareconstant(n)-vectors,ρisanumber, 1− 2 × andψ(σ)is2π-periodicfunction,satisfyingtherelations: ρ= aL, − W(p)=L−1c∗(A pI)−1b, − ω (0) ω (0) ψ(σ)= ϕ(σ) 1 − 2 . − L(a+W(0)) Thediscretephase-lockedloopsobeysimilarequations z(t+1)= Az(t)+bψ(σ(t)) (19) σ(t+1)=σ(t)+c∗z(t)+ρψ(σ(t)), wheret Z, Z isasetofintegers. Equations(18)and(19)describetheso-calledstandard ∈ PLLs (Shakhgil’dyan & Lyakhovkin, 1972; Leonov, 2001). Note that there exist many other modificationsofPLLsandsomeofthemareconsideredbelow. 4. MathematicalanalysismethodsofPLL Thetheoryofphasesynchronizationwasdevelopedinthesecondhalfofthelastcenturyon thebasisofthreeappliedtheories:theoryofsynchronousandinductionelectricalmotors,the- oryofauto-synchronizationoftheunbalancedrotors,theoryofphase-lockedloops. Itsmain principleisinconsiderationoftheproblemofphasesynchronizationatthreelevels:(i)atthe levelofmechanical,electromechanical,orelectronicmodels,(ii)atthelevelofphaserelations, and(iii)atthelevelofdifferential,difference,integral,andintegro-differentialequations. In thiscasethedifferenceofoscillationphasesistransformedintothecontrolaction, realizing synchronization. Thesegeneralprinciplesgaveimpetustocreationofuniversalmethodsfor studying the phase synchronization systems. Modification of the direct Lyapunov method with the construction of periodic Lyapunov-like functions, the method of positively invari- ant cone grids, and the method of nonlocal reduction turned out to be most effective. The last method, which combines theelements of thedirect Lyapunov method and thebifurca- tion theory, allows one to extend the classical results of F. Tricomi and his progenies to the multidimensionaldynamicalsystems. 4.1 MethodofperiodicLyapunovfunctions HereweformulatetheextensionoftheBarbashin–Krasovskiitheoremtodynamicalsystems withacylindricalphasespace(Barbashin&Krasovskii,1952). Consideradifferentialinclu- sion x˙ f(x), x Rn, t R1, (20) ∈ ∈ ∈ where f(x)isasemicontinuousvectorfunctionwhosevaluesaretheboundedclosedconvex set f(x) Rn. Here Rn isann-dimensionalEuclideanspace. Recallthebasicdefinitionsof ⊂ thetheoryofdifferentialinclusions. Definition1. WesaythatUε(Ω)isanε-neighbourhoodofthesetΩif Uε(Ω)= x inf x y <ε , { | y Ω| − | } ∈ where isanEuclideannorminRn. |·| www.intechopen.com (cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:4)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:4)(cid:8)(cid:5)(cid:12)(cid:13)(cid:6)(cid:13)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:15)(cid:7)(cid:13)(cid:6)(cid:16)(cid:4)(cid:10)(cid:3)(cid:17)(cid:10)(cid:18)(cid:19)(cid:8)(cid:13)(cid:7)(cid:20)(cid:21)(cid:3)(cid:22)(cid:23)(cid:7)(cid:14)(cid:10)(cid:21)(cid:3)(cid:3)(cid:24)(cid:13) (cid:26)+ Definition2. Afunction f(x)iscalledsemicontinuousatapoint x ifforanyε > 0thereexistsa numberδ(x,ε)>0suchthatthefollowingcontainmentholds: f(y)∈Uε(f(x)), ∀y∈Uδ(x). Definition 3. A vector function x(t) is called a solution of differential inclusion if it is absolutely continuousandforthevaluesoft,atwhichthederivativex˙(t)exists,theinclusion x˙(t) f(x(t)) ∈ holds. Under the above assumptions on the function f(x), the theorem on the existence and con- tinuabilityofsolutionofdifferentialinclusion(20)isvalid(Yakubovichetal.,2004). Nowwe assumethatthelinearlyindependentvectorsd1,...,dmsatisfythefollowingrelations: f(x+d )= f(x), x Rn. (21) j ∀ ∈ Usually, d∗jx is called the phase or angular coordinate of system (20). Since property (21) allows us to introduce a cylindrical phase space (Yakubovich et al., 2004), system (20) with property(21)isoftencalledasystemwithcylindricalphasespace. Thefollowingtheoremisanextensionofthewell–knownBarbashin–Krasovskiitheoremto differentialinclusionswithacylindricalphasespace. Theorem3. SupposethatthereexistsacontinuousfunctionV(x):Rn R1suchthatthefollowing → conditionshold: 1)V(x+d )=V(x), x Rn, j=1,...,m; j ∀ ∈ ∀ m 2)V(x)+ ∑(d∗jx)2 →∞as|x|→∞; j=1 3)foranysolutionx(t)ofinclusion(20)thefunctionV(x(t))isnonincreasing; 4)ifV(x(t)) V(x(0)),thenx(t)isanequilibriumstate. ≡ Thenanysolutionofinclusion(20)tendstostationarysetast +∞. → RecallthatthetendencyofsolutiontothestationarysetΛastmeansthat lim inf z x(t) =0. t +∞z Λ| − | → ∈ AproofofTheorem3canbefoundin(Yakubovichetal.,2004). 4.2 Methodofpositivelyinvariantconegrids.Ananalogofcircularcriterion Thismethodwasproposedindependentlyintheworks(Leonov,1974;Noldus,1977).Itissuf- ficientlyuniversaland"fine"inthesensethathereonlytwopropertiesofsystemareusedsuch astheavailabilityofpositivelyinvariantone-dimensionalquadraticconeandtheinvariance offieldofsystem(20)undershiftsbythevectord (see(21)). j Hereweconsiderthismethodformoregeneralnonautonomouscase x˙ =F(t,x), x Rn, t R1, ∈ ∈ wheretheidentitiesF(t,x+d ) = F(t,x)arevalid x Rn, t R1 forthelinearlyinde- pendentvectorsd Rn (j =j1,...,m). Letx(t) = x∀(t,t∈,x )is∀a∈solutionofthesystemsuch j ∈ 0 0 thatx(t ,t ,x )=x . 0 0 0 0 www.intechopen.com (cid:26)(cid:25) (cid:11)(cid:27)(cid:28)(cid:29)(cid:30)(cid:11)(cid:28)(cid:31)(cid:29)(cid:2)(cid:10) (cid:10)!(cid:29)(cid:2)(cid:28)"(cid:29)(cid:21)(cid:10)(cid:20)(cid:10)(cid:28)(cid:19)(cid:7)(cid:3)(cid:9)(cid:12)(cid:10)(cid:8)(cid:4)(cid:14)(cid:10)(cid:18)(cid:9)(cid:8)(cid:22)#(cid:6)(cid:22)(cid:7) WeassumethatsuchaconeoftheformΩ = x∗Hx 0 ,whereHisasymmetricalmatrix { ≤ } suchthatoneeigenvalueisnegativeandalltherestarepositive,ispositivelyinvariant. The lattermeansthatontheboundaryofcone∂Ω= xHx=0 therelation { } V˙(x(t))<0 issatisfiedforallx(t)suchthat x(t)=0,x(t) ∂Ω (Fig.5). { (cid:13) ∈ } Fig.5.Positivelyinvariantcone. Bythesecondproperty,namelytheinvarianceofvectorfieldundershiftbythevectorskd, j k Z,wemultiplytheconeinthefollowingway ∈ Ω = (x kd )H(x kd ) 0 . k { − j − j ≤ } Since it is evident that for the cones Ω the property of positive invariance holds true, we k obtainapositivelyinvariantconegridshowninFig.6.Ascanbeseenfromthisfigure,allthe Fig.6.Positivelyinvariantconegrid. solutionsx(t,t ,x )ofsystem,havingthesetwoproperties,areboundedon[t ,+∞). 0 0 0 IftheconeΩhasonlyonepointofintersectionwiththehyperplane{d∗jx = 0}andallsolu- tionsx(t),forwhichatthetimettheinequality x(t)∗Hx(t) 0 ≥ issatisfied,havepropertyV˙(x(t)) ε x(t)2(hereεisapositivenumber),thenfromFig.6it ≤− | | isclearthatthesystemisLagrangestable(allsolutionsareboundedontheinterval[0,+∞)). www.intechopen.com

Description:
Dec 1, 2009 Phase-locked loops (PLLs) are widely used in telecommunication Phase- locked Loop Engineering Handbook for Integrated Circuits, Artech.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.