Applications of Cnmmuniculiuna Thuiewtrrj:F
`Series Edilur: EW. Lucky
`
`
`
`CBM2013—00009
`
`

`

`lenrlll unf- Cunurall h1llun1nu-Ifl-Puhl1tfllnh Ill-Hi
`
`
`ul'rh lannratnrll-
`
`Traflfir. Ste-LIED fl.
`Ebuun'ucatmn syn" duly-I urn-1g DEF “gum-Inn
`”pen-nan.” Hr 1h. THESEDEEEI
`.n' Eleven- A. Trauan
`u, u. -- Clap-phclllfllfil {If
`Iflllul'l'lfllilllll'll
`In“)???
`Inc1udli bln1lnqrapn1cal rl+|rrnnll lflfl 1nutt.
`ISBN 6-353-45632-1
`2. DLgIta-l
`IEthl‘l-lquE'E.
`1. 5'9"!" pruca::1ng—-Dlg11‘fll
`:n-llun|fiatibn$--EQu1pllfl-1 and :upn1IH--D-E:Ipn--Dau nfucflsmn.
`3. Tuna: 1n|1FUIln11 THsaan IIF1II nucrnnrncaslnrl.
`d. Hill-f1ll
`alt: pruclntlnq.
`I. Tllll.
`II. Elr1il.
`TKfi1fl2.fi.T?-fll
`+995
`521 .EEE' 2 ' UTE-4min
`
`EI'EI'B'EB-EU
`EIP
`
`
`"'wa diikflm ii dcffltiw. in mnul'wun: war In: mu “mind in trlmil. il will b: rcpllccd.
`II nu fihlrflt iI" FtEI-Ifl'ltd wialu'n 3D- dm ufnuipl :«n Min-Ilia; Edimr. Fltrlum Fran. :33
`5pm; Elmer. New ‘r'urlu. H“? mm 3.
`
`Thu: puhlnhcr mm: nu wlmnty Mm}.I kind. upwind ur implitd. with regard In III:
`inn-Fl wm mprndumd an IR dislitm ur Ih: mmmfinyina datumnllfiun. Th: publiihcr flull
`rum be liahle in any ever“ I'm Incident-1 nr mnnqumfinl dam-3:5 a: loss in cunnmlinn with,
`15' arising UL" UT. the filnl'l‘llflhing. perfmn‘flncc. or In: I:II' lh: mflwrt.
`
`FEBH WIDE-450314
`
`Elm!!- Filer-urn Frau. Mew Yuk
`
`A Dirilinn 11F Plenum Puhl'nhina. Carper-rim
`.23! 5mm. Sin-:1. NIL-w 'r‘mk. HT. lull]
`
`IDQSTfifl-JE
`
`A11 r-illus mutwd
`
`Ha pm: ul' II'III bwk may b: mrmmed. “mad in I fllril'l'l] ljl'llill'l'l. :H' mulmiuud L11. Iflj'
`farm or by my means. mnmnic. machininl. phmucnpyiu. microfilm“. “cording. m
`«hen-me. wrlI'l-[Iul mmm prrmu'n'un I'mm ll": Publiihrr
`
`Puntnd an [ht IJI'IiItd Eta-In of America
`
`

`

`Cements
`
`l. llrlelr I‘JI'I'erIIrle‘IIr nl.’ the led'n'lre III! Eert'IIIrere Tent: ...........................................
`
`1
`
`LI. Same DSP l|.'C.'hi|:I Hist-313' and Typical Applieatinns ...........................................
`I.1_ intrndI-Ictiu-n In the TMEHEflCfl-‘I Planting-Paint DEF .........................................
`I .1. l. The Email Preeessing UnitiCPUl .................................................
`l...11 Memeryflrganizmien..........-.-
`I...11 'I'IlelnternelEI-Ie
`l...14 TheExtemnlBusesendInteifipliI
`I...15 Peripherals .....................................................................
`l.1.I'-‘r. TheDirectHeinmjrheeessiDMMCnnlmller.......-.
`I-1-7. BriefDeecfiptin-nufmelnetruetiunSe‘t..................
`l.3. TheWS31flC‘3DEvnIuelinnMe-dule
`I..4 SeftwateTeels.
`I .4. 1. Software Enliplieii In}; TeiuIi: IhetriIinente”.............................................
`l-4.1 DigimlFiltetDesignPregrnnm.
`l-5.lnIIntlue1u:3rE.1pemnents ll
`
`I4..'-l. fllherfiei‘tware.
`
`.-
`
`..
`
`Eamqemmmamum—
`
`ll
`
`1. anin; to Use the Herthure Intl S-IIftIIIIre 'llnnll hy Generating II Sine We've ....................
`
`I3
`
`2.], Some Basie Utilin Files, Pregame, and Environment 1iIr"II|I'iIaI:I1es Required by the 'l'MSflilC 30
`Assembler C Compiler and Linker ................................. . .....................
`2-1. MethndlfurGeneratingaSineWaw........
`..
`1. 3. Methed 1 fed: Generating e. Sine Wave—Using lIitei-rupts .....................................
`1.4. Methnd 3—DMA fl‘nm a Table ...........................................................
`
`13
`l?
`14
`16
`
`

`

`Euler-ti:
`
`3- DIEIHIFIHE” uuuuuuuuuuuuuuuuuuu emu-rue uuuuuuuuuuu I-rv++++++111-IIIu.-...-..u-rr-r-rr-r-rr-r-rr-r-r-r-u-
`
`3 i. Discrete-Time Cnnenlutian and Frequency Responses .......................................
`3.2. Finite Dnrntinn Impulse RespnnselFIR‘I Filters
`.
`.
`.
`
`3.2.1. Elect: Diagram f‘cr Mes-t Common Realization ........................................
`3.2.2. Twn Metlintln tier Finding the Filter Einefiieients tn Achieve a Desired Frequency Reap-erase .
`.
`.
`3.2.3. Using Circular iinfl‘erii tn Implement l-‘lFt Filters ......................................
`3.2.4.
`iinw tn Make lll'l Assembly Language {‘nnvnlutinn Functinn that It. Called from and Returns
`
`3-3.
`
`infinite Dnriilitiri Impulse Reap-unit: [HR] Filters.
`3. 3 I Rnalizalinna Terr HR. Filters...
`. . . _ .
`. _ .
`
`. . . ..........
`. .
`
`3.32. APE-gramIhrfletigningllRFillels
`3.3. 3. Twn lI'Ietlmds- finr Measuring a Phase Respnnae .......................................
`3-4. Hardware Details and Lahciralnry Experiments far I-' [II and Illi'. Digital Filters .
`.
`.
`.
`.
`.
`.
`.
`.
`. _ .
`. _ .
`. _
`3.4.1. Same l-lartiware and Heft-ware Details fer the E‘I'M ....................................
`
`3.4.2. Measuring the Time to Execute a Program Segment by Using the Benchmarking Capabillljl'
`at" EVMSD ......................................................................
`
`3.4.3. Lahnratcry Experiments fer FIR Filters ..............................................
`3.4.4. Lul'I-nratery Experiments fer IIR Filters ..... .
`.
`.
`.
`.
`.
`.
`. ..................... .
`.
`.
`.
`.
`.
`. .....
`3.5. fiddilinnalflefercnces ......................................
`
`- TheFFTandPnfilnrsplctrumEstlmltiu“IIIIIIII.I.I-I-I-I-I-Int-I11111:iIIiII11111111111111+++++rrrrn
`
`4. |. The Discrete- Time FnIIrinr Trnniil'urm .
`
`4.1 [JuaninelnwFunctinnn......... _.
`4. 3. The Dinerele Funrier Transfnn'n and. IIS Inverse .
`4 .4. The Fast Fourier Tranal'nnn .
`.
`.
`.
`.
`.
`4. 5. Lining, the PPT In Estimate a Putter Eipeelt‘um.
`4J5.LaberalnryErtperimenta.........__.__._........
`
`.
`
`.
`
`.
`
`.
`
`..
`
`4.t:.l. FFTErtperimcnts
`
`435.2. Experiments Eur Fewer Spectrum Estimatien
`4.7. Additinnal Refit-renew .
`
`5. Amplitude MuduIItI-nn .....................................................................
`
`1. Thenrelicnl Deseriptinn nffimplitndt: Mndnlatinn ...........................................
`
`5.1 .I. Mathematical Funnula fur an AM Signal: ............................................
`5.12 . Example I‘nr Single Time Medulatinn ...............................................
`
`5..13. Thefipeetrumet'anhMSignal
`
`5.2. DemndulatinganMI-I SignalbyEnvelnpeDeteetinn
`
`5.2-1. Muare- Law Dctnntlulfltinn ni‘flIM Signals ...........................................
`
`5.2.2. HilbertTranal‘nmIaantitlIeCemplenEntrelnpe...................
`5.3. Leburatnry Experiments fer AM Mndulatinn and Dematlulatinn ...............................
`531 MakinganhMMndulatnn.
`5. 3-2. Making a Square-Lav. Ent'eleipe iffieteeleir ............................................
`5.3.3. Melting an Eneeinpe Deteetnr Using the Hilbert Transferm .............................
`5.4. Additim‘tal Reliereneea ..................................................................
`
`29
`
`29
`
`3a
`
`313
`
`3i!
`
`32
`
`$331
`
`42
`
`4ti
`
`43'
`
`49
`
`4"}
`
`5|
`
`5 I
`
`5 I
`52
`
`53
`
`54
`
`33%
`
`fit!
`
`I55
`
`55
`55
`
`{36
`
`Eli
`
`tie
`El?
`
`it?
`E19
`
`59
`
`3'11
`
`'21
`
`

`

`Cements
`
`IIiIH
`
`ti. Double-Sldehlod Suppressed-Currier Amplitude Modulltion and Coherent Detection . .
`
`. . .
`
`. . .
`
`. . .
`
`. .
`
`'73
`
`Er. I. Mathematical Description of the DSBSC-AM Signal ........................................
`Ei..2 TheldealC'oherentReocitrcr
`. ..
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`..
`5.3. The Costa: Loop asePreelical Approach toEoitei'enl Demootmllnlitiri
`5.4. LIIIIIIIorIIItorgII Exercises and Experiments for the Cosme Loop ...................................
`5.4.]. TheoretieelflesignEIIereiees
`[3.4.1. HerdwercEIIperimenls
`5.5. AdditioneIRel'crenees......................................-.....-..-..............-...
`
`7. Slngle-Sldehend Medulltlon Ind FrequencyTreniletlon
`
`T3
`T3
`T4
`to
`'I'T
`TS
`TI!
`
`T'jl
`
`T‘I'l
`'i'..I Single-Sidehondhlloduiutore...
`SE]
`2.2. CoherentDemoduletionofSSBSigoals-..-..-..-................-..-..-..............--.-
`3.3. Frequency'lranhlatton 3|
`7.4.
`I.IIhI'IrnI‘or_I.I Experiments ................................................................
`SE
`7.4.1. Melting :tn SSE Modulator ........................................................
`S2
`7.4.2. CoherentDemodulnlionol'nnSSBSignal..-................-..-..-..-............-.-
`SE
`15. AdditionalRel'erenoes.-..-..........-..-.....-.........................................
`SJ
`
`3. Frequeneyllr‘lodulalion ................................. .......... ......... ............+....
`
`35
`
`.
`SJ T’I.IIcFMSignaltInIiSomeol'lteProperlies.........
`Ell Def'nltlonoflnstanuneouhFrcqueneyandlheFMSIEnul
`S.I.2. Single Tone FM Modulation .......................................................
`S.l.."-. NerrowflendFMiI-locluietion......-...............................................
`
`S.l.4. The Bandwidlh ofen FM Signal ....................................................
`3.2. PM DemodulelionbyaFtequency Dieeriminnlor ......... . ..................................
`SJ. UsinguPhuse-LocketlLoopl‘orFMDemorluInIioI-I
`34. Laboratory Experimentit forFrequeney Modulalion ..........................................
`14.4.1. Experimentally MeeeuringtheSpectrumol'en FMSignel
`3.4.2. FMDemoduiationLleingeFrequeocyDiscrimioeror.............-.....-...............
`3.4.3. Using a Phase-Locked Loop for FM Demotluietion ....................................
`8.5. Additional References ..................................................................
`
`SS
`HS
`So
`Hfi
`
`So
`Sti
`S?
`SS
`H9
`9t}
`1it!
`St]
`
`'9. Pseudo-Random Binary Sequence: and DIII Scrambler: ....................................... 9|
`
`‘3
`
`|. Using Linear Feedback Shift Register: to Generate Pseudo-Random Binary Sequences ............
`ELL]. The LinearFeeIiheek Shift Reg-inter Sequence Generator
`. .. .
`.
`.
`.
`.
`.
`.
`.
`. ..
`lii|.l.2. TheConnection FoiynomiolondSequencePeriod
`“ill Propertiesol'htlmtimulLengti'I Sequencer.
`9.2. Seli‘SynehronizingDataScrambler:
`9.2.1. Tthcremhlcr
`9.2.2. The Descrernhler ................................................................
`
`SI
`'92
`91
`Sr]
`'93
`93
`‘14
`
`94
`'15
`
`'15
`
`'35
`
`9e
`
`9.3. "l'heorcticni and Simulation Exercisce for Shift Register Sequence Generators. and Scrumhlere .......
`‘13.]. Exercises For It Shift Register Sequence Gene-rotor with o Primitive Connection Folynominl .
`.
`.
`9.3.2.13itetei5ee for It Shift Register Sequence Generator with on lrredueihle but not Prlll‘litl'I'L'
`ConnoelionPelyoomiel.
`..
`
`..
`
`'3.3.3. Exercises Put a Shiflliegislier Sequence Cienerator witmhoRodoeihleC-‘onoeelionPolynomial.
`
`9.4. Additionalkefereneee.
`
`

`

`"1'
`
`Eel-teats
`
`HI. Introduction to the RSI—232E Protocol altl a flit-Error Rate Tester ...............................
`
`It?
`
`1t}.l. The EIA BIS-232E Serial Interface Protocol
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .........................................
`
`.......
`1:12. Hit-Error Probability for Binary Signaling on the Additive. White, Gaussian Neise Channel
`10.3. The Haflel Datatestii Bit-Error Rate Tester ...............................................
`
`9?
`
`99
`9";
`
`ISA. Bit-ErrorRate Test E1per1tl1ent ......................................................... lit-El
`155. Additional References ................................................................. 1D2
`
`II. Digital Data "Transmission by Easels-ml Pulse Amplitude Modulation {PAM} . . . . . . . .. ............. 1113
`
`. . .
`
`. ..
`
`ItiS
`lfl5
`HIS
`It'lti
`Iflfi
`III“.r
`III?
`
`NJ. ficneratflescription ofaEaseband Pulse Amplitude Modulation System .............. .. .
`II.2- BasebandShapingandlntersymbollntcrference
`1|.2l. TheNyquistCriterinnfnrNaiSl.
`Il.2.2. RaisedCosineBeaehandShapingFiIteI‘a
`il.2..5. Spittnng the Shaping‘oetween the Transmit and Receive Iiilters.
`11.24. EyeDiagranIs
`Implementing the Transmit Filterby anInternal-stint; FiIte'I' Bank .
`.
`.
`.
`.
`.
`IL].
`I l.4. Symbol Error Probability for a Channel with a Perfect Frequency Response and Additive CIIIussinn
`Noise ________________________________________________ . ............................. IDS
`
`.............- .
`
`. -.
`.
`
`ll.5.SynnbolCIocltRecoIrc1y.
`IDS
`ILEI,S1rnulal1onandThenrettcalEaerctseafarPAM
`1H]
`Il.ti.l. Generating Flour-Level Pseudo-Random PAM SyrnImis .............................. HI]
`ll.e.2. Eye Diagram for It PAH Signal Using a Raised Cosine Shaping Fitter ............. . .. .
`.
`III
`I I.ti.3. Eye Diagram In: a FIRM Signal Using a. Square-Rout rII'RIIised flu-sine Shaping Filler .....
`I
`I
`I
`II. 5.4. TheoreticnlError Probability ntaPAM System ....................................
`Ill
`11.2.HardwarcEaercisesforPAh-I.
`III
`
`III
`II'I'..I GenerattngaPAlu'lStgnalanelEyeDIagrem..-...
`112
`II 2.2 Testingthefiqttare-LawannholflinekFrequeneyfieneralor
`1.13-
`“.13. DptiortalTeaInEJt-ereise
`11.S. Additional References ................................................................. 113
`
`12. FundamentalsofIJIIIIlratureAmplltutlehlednlatlon ..++. 1'5
`
`l2.|. AE-asic QAIH'I Transmitter .............................................................
`12.2. TwnConstcllationExamples
`I2.2.I. T'Itc4a4lfi-PeintConstellatioa .................................................
`l2.22. A4-F'ni1'tt FmtrvPhaseCnnstellatinn ...............................................
`
`I15
`IIEI
`lle
`Ii'il
`
`12.3.21 Modulator Structure Using Passhand Shaping Filters ...................................... HS
`124 IdealQfiMDemnduIatIon I21}
`11.5 QAMModulatorExperiments .......
`I21}
`l25.l. Stepsiel-‘ollowin MakingaTransmiucr ..................... . .................... Izt
`l2-5.2- Testing‘t'ouITransmittccr ........................................................ l2]
`l2_5.3,
`flptifl't'lal Exercise—Tasting "I’cIIIr Transmitter by Sending to a Commercial Mnden‘l .......
`I 23-
`.I'Itlditionat References ................................................................. 125
`
`12.15.
`
`15. filial Receiver I. General Description ofConIplete Receiver Bloclt Diagram and Deralls ofthe
`Symbol Clock Recovery III! Dther Front-End Suhaytems ..................................... . 127
`
`15. I . flyewiew nfa DAM Receiver .......................................................... 127
`13.2- Details about the Receiver Front-End Subsystems .......................................... 12';1
`
`

`

`Communication System Design
`Using DSP Algorithms
`
`1With Lahnratnry Experiments
`fur the TMSSZIJCSD
`
`

`

`
`
`Brief Overview of the Hardware
`
`and Software Tools
`
`The purpose of this initial experiment is to introduce you
`to the main features of the hardware and software tools
`
`that will Ire used in this course. A 1rrsrtt'iety of signal pre-
`cessing and emnmunicatien system components will be
`implemented by writing t2 andt'er assembly language pro-
`grams for the TMSJEHEJD DEF. The DSP communicates
`with the analog world through ND and Dirt cent-«errant
`andwilhahaatPCthIoughtl‘reEvhitFCdatahtbs. The
`DSP and converters reside on a card called the Evaluation
`
`Module [E‘u'h-t} that plugs into a PC slot. It is important
`to have a general picture ofthe hardware platform so you
`will understand how to write programs to aceentplish the
`desired tasks. First, a very brief history of DEF chips
`is presented and some typical applications are dismissed
`Then, the important feanrrea of the TMSJEDC3D DEF are
`described. Heat. a hloctc diagram of the Ey'ht is discussed
`followed by an inn'oduction te some of the software tools.
`Finally, you will he guided throufl'r some very simple err-
`perirrtents to learn the basics of how to use the system.
`it.” you finish these experiments before the end of the lab
`periudt go on to ILlhrrpter 2.
`
`Details of the TMSflflCJfl architecture and insane-
`
`tiecrt set and of the EVh-Il board will be introduced grad-
`ually in the first few experiments as they are required.
`Complete details for the TMSSEDCSD. E‘r'h-t. and Ti soft-
`ware can be found in the Team Instruments manuals when
`
`required. (The ability to read manuals is an important skill
`to acquire.)
`
`There are many things to be learned initially and you
`may feel cyenrrhcln'red. Please be assured that the major-
`
`ity of your classmates feel the same way. 1tier-y shortly you
`will get otter this initial hurdle and find the errperirnents
`interesting and challenging. As the semester progresses.
`we will build up a bag-ef-triclss that can be used in fel-
`lewing experiments so they will almost seem to get easier.
`
`1.1.
`
`Sense DEF Chip History
`
`and 'l‘yp'real hpydieafierrs
`
`Digital signal processing chips {DSP‘sl were intro-
`duced in the only lfitfifl’s and have caused a revelation
`in product design. Current major DEF manufacntrers
`include Tessa htstrurrrents [Tl], Motorola, thtT1 .I'tlfta-
`
`IOg Dot-ices. NEG. SGS-Therttaent and Rockwell Interna-
`tional. DSP's differ from ordinary microprocessors in that
`they are specifically designed to perform the sum of prod-
`ucts operation requires in many discrete-time signal pro-
`cessing algorithms rapidly. They contain hardware paral-
`lel multiplies. and functions implemented by microcode
`in ordinary microprocessors are implemented by high
`speed hardware :in DSP‘s. Lewvcost DSP’s have made
`it more economical to implement functions by digital sig-
`nal processing techniques rather than by hard-wired ana-
`log circuits, particularly for audio hand applications litre
`speech compression and telephone line modems. Some of
`the advantage results from the fact that integrated digital
`circuits are very reliable and can be automatically inserted
`in boards easily.
`in addition. programmable DSP‘s can
`implement complicated linear and nonlinear algorithms
`and easily switch functions by jumping to different we
`
`

`

`lL'ltrapler I.
`
`tions of program code. The complexity of'tlie algorithms
`is only limited by the imagination of' the programmer and
`the processing speed of the DEF. Once the program is
`perfected. the chip t'unction does not change with ago un—
`less a very rare failure occurs. {in the other hand. analog
`components require more board space. sometimes must be
`trimmed to the correct values alter insertion. and change
`with temperature and age. Analog circuits are designed
`to perform specific functions and laclt the flexibility of
`the programmable USP approach. Another advantage is
`that small changes in the DEF function or bug lines can
`be made by changing a Few litres of code in a REIM or
`EPROM. while similar changes may be very difl'tcult with
`a hard-wired analog circuit.
`
`DEF techniques were used earlier but required build-
`ing your own processor with many TTL MEI chips. in-
`cluding cascadabie ALU sections and stand-alone multi-
`plier chips.
`at
`typical system contained over IIIIII MEI
`chips. These systems were big and csp-cnsive to man-
`Ltl'acture because of the large chip lt.':t:ttt..ll'll'. they consumed
`many watts ot'power and required cooling fans. The most
`popular First-generation DEF chips.
`the NEC pPD't'r'Eil
`and Tesas Instruments Tusaaotu. became commercially
`available in. late Will These chips perfonncd lot-bit in-
`teger arithmetic at the rate of 5 million instructions per
`second {MIPS} and had limited internal RAM. RUM. and
`lit.) capabilities. They initially cost about Elf-till and now
`can be bought for about $2. They reduced the chip count
`by a very large 1:rercent.age1 resulting in smaller circuit
`boards usirtg significantly less power. more reliable sys-
`tems. and reduced manufacturing complexity and cost.
`DSP's have continually evolved since diey were first
`introduced as "r'LSi
`technology improved, as users to-
`qucstcd additional functionality. and as competition arose.
`More internal MM and ROM has been added and the
`
`total address space has been increased- additional func-
`tions have been added.
`including hardware bit-reversed
`addressing for FFT’s. hardware circular buffer address-
`ing. serial ports. timers. ulna. controllers. and sophisti-
`cated interrupt systems including shadow registers for low
`overhead cmtteitt switching Analog Devices has included
`switched capacitor filters and sigma-delta MD and Dirt.
`converters on stun: DEF chips. Insu'uetion rates have in-
`creased significantly. State-of-Ihe—art integer DSF's liltc
`ti1e TMSflthSs series have recently arrived on the mar-
`lret that can operate at 57-" MIPS and cost as little as 53]
`in quantity. The speed increase is largely a result of re-
`duced geometries and improved CMGE technology. The
`introduction of CMDS technology after the that couple
`of generations of DSlt's significantly reduced power con-
`sumption. also. DSP‘s using lower supply voltages are
`ttow being built. applications like telephone line modems
`that required at least two HSP’s and an additional ordi—
`
`nary microprocessor acting as a oontroller as little as five
`years age. can now be implemented using a sirtgle DEF
`and at lower cost!
`
`Another milestone in DSP history occurred amend
`toss when nTl'er introduced the first cornmtrrciai float-
`
`In IEEE. T1 shipped initial
`ing-point DEE the DSFJE.
`samples of the Tlv'iSftlilICZlil at a price of stone to be-
`gin its first-generation TMSdEflCEs floating-point ESP
`family. Both processors have a 32-bit word length. The
`WSEEIIJCSU is currently offered with clock speeds of 21.
`33. and sit] MHz. The suggested resale price for a 33 Ml—lr.
`rmsaaucao is new star in quantity. a stripped down
`version. the TMSEEDCEL followed and has a suggested
`resale price of' $53- for the 33 “Hz version- Recently. Tl
`started its second-generation floating-point DEF family
`with the TMSEEtlti-ltl. which contains extensive support
`for parallel processing.
`Some new generations of DSF‘s are now beng de-
`signed with video applications in mind.
`in the Fall of
`little. T1 began shipping samples of its next generation
`TMS’lztlcai-t DEF series. The initial member of this
`
`the TMSJEIIJCSD. contains four advanced-integer
`series.
`DSPs using 32-bit data and rid-bit instnlction words. one
`RISE processor that
`includes a floating-point ttt'tit. Still:
`words of Shall-t. a crossbar switch. a video controller.
`
`It has special instructions
`and extensive lit] capabilites.
`designed For pisel processing. Ti claims the chip can
`perform two billion operations per second.
`It is projected
`that the chip will soon sell for around Hill} in quantity.
`
`Floating-point DSP's are now being used in many
`applications hocattsc oF their case in programming and
`reasonable price. However.
`integer DSP‘s are still be-
`ing chosen very often because they are cheaper. some-
`what faster. and use less power. Cost is a major con-
`cern in the competitive corrutiercial market. particularly
`for mass-produced products like modems and hard dislr
`drives.
`lliare must be Ialcen with integer DSP‘s to scale
`signals to avoid overflow and underfirrw. However. this
`is not much of a problcnt in many digital communication
`system applications where signal power levels remain rel-
`atively constant. Floating-point DSF‘s automatically per-
`form the scaling.
`
`DSP‘s are used in a wide variety of offline and real-
`
`time applications. Some typical areas and specific appli-
`cations are:
`
`telephone line modems. fax.
`I Telecommunciattons:
`cellular telephones. speaker phones. ADPCM trans-
`codcrs. digital speech interpolation. and answering
`machines
`
`I l-“oioet‘dpeech: speech digitisation and compression.
`
`voice mail. spcalter verification. and speech synthesis
`
`

`

`Hriefflvtotvlflv Elf the Hardware and Software Tools
`
`r Automotive: engine control. antiloek brakes. active
`suspertsion. airbag control. and system diagnosis
`
`a Control Systems: head positioning servo systems
`in disk drives.
`laser printer control. robot cantral.
`engine and motor cornrol. and numerical control of
`automatic machine tools
`
`I Military: radar and sonar signal precessing. naviga-
`tion systetns. missile guidance. HF radio frequency
`modems. secure spreadepectrum radios. and secure
`voice
`
`imaging. ultrasound
`hearing aids. Mill
`- Medical:
`imaging. and patient monitoring
`
`a instrumentation: spectrum analysis. transient anal-
`ysis. signal generators
`
`an Image Processing: HDT‘t-I'. pattern recognition. im-
`age enhancement. image compression and emails-
`sion. 3-D rotation. and animation
`
`real-time DEF applications are limited to
`Clearly.
`cases where the required signal sampling rate is suff-
`cictttly less than the DEF instruction rate so a renewable
`number of instructions can be performed between same
`pies- For esarnple. a wideband radio frequency {RF}
`signal with a high carrier frequency can net be directly
`sampled and demodulated with a DSP. However. when
`the bandwidth of the RF signal is sufficiently less than the
`instruction rate. analog front-end circuits can he used to
`demodulate it to baseband inphase and quadrature corn;
`poncnls. which can then be sampled at a rate equal to the
`bandwidth and processed by a DEF. altemativety. art ana-
`log filter can be used to form the Hilbert transform of the
`RF signal and then the original signal and its Hilbert trans-
`form can be sampled at a rate equal to the bandwidth and
`processed with a DEF. DSP’s have been used extensively
`in audio frequency applications. where many instructions
`can be perfon'ned berwecn samples. However. they are
`being used to process increasingly wide-band signals as
`the instruction rates of new generations increase.
`fit very important attribute of the DSP approach is
`the flexibility of a programmable device. a. timely ex—
`ample illustrating an ideal application for this flexibility
`will now be preset-tied.
`in June 1994. the ”U approved
`flte final draft of the use telephone line modern recom-
`mendation. This modem can transmit data at rates varying
`from 24th] to asses bitsfsec in muhiples afloat] bitsfsoc.
`a variety of modulation schemes are used during hand-
`shaking and annual data transmission. During the initial
`handshaking phase. binary. continuous phase. fretpsency
`shift keying is used to exchange information about mo-
`dern capabilities using the vs protocol. Later. the called
`
`modern. usually referred to as the answer modern. trans-
`mils a signal called the answer lotto to the calling modern.
`In the v.34 case. a small sinusoidal amplitude modulation
`is impressed on the answer tone to identify it as a v.34
`modem. so the receiver in the calling modem needs to
`perform the function of an envelope detector.
`In another
`phase of the handshake. a channel probing sequence con—
`sisting of a sum of sine waves is transmitted first by one
`modem and then by the other. The calling and answer
`modern transmitters must synthesise this signal and the re-
`ceivers must process it to estimate channel characteristics
`such as frequency response. noise level. and nonlinear-
`itinl.
`r'tncthct‘ special sequence is transmitted to rapidly
`adjust adaptive equalizers in the receivers. During still an-
`other portion of the handshake. hinary differential phase
`shift keying {DBPSK} is used to exchange configuration
`infonnation between the modems. Quadrature amplitude
`modulatim't {ta-AM] is used during normal data transmis-
`sion- Transmit and receive signal separation is achieved
`
`by using adaptive echo eaneellers- Additional tasks that
`must be performed during normal data transmission are
`scrambling and descr‘at‘r'thlir'tg the input bit stream. map-
`ping the scrambled data bits ta transmitted signal points
`by a technique called shell mapping. trellis encoding. non-
`linear preceding. and salt decision Viterbi decoding.
`fill
`these tasks. and more. can now be performed by a sin-
`gle stateaf-the-att 50 MIPS integer DEF.
`In addition.
`as a result of intense competition. all the ‘v'..l-t modems
`
`just now beginning to teach the market can also act as at
`least sis older modem types and inclttde Fol-i modes. All
`that is required to include new functions is to "add a few
`more lines of code and. possibly. scone more memory."
`These modems will cost the consumer 54ml or less. Fro-
`
`grammablc DSP's have made it possible to manufacture
`
`products with tremendous complexity and sophistication
`at prices that the ordinary consumer tJ-tlfl easily afford-
`Finally. to maintain a balanced viewpoint.
`it should
`be pointed out that the DEF" approach is not always the
`best solution to a problem even if a USP can accomplish
`the task. For example. a commercial not radio signal.
`which has a carrier frequency on the order of I MHa. can
`trivially be demodulated by a sitnplc envelope detector
`consisting of little more than a din-dc. resistor. and capac—
`itor. as engineers. you she-ind always look for the most
`reasonable and ccottotnical method of solving a design
`problem.
`
`L2.
`
`Introduction to the TMSJIIICJD
`
`Floating-Point DSP
`
`The experiments in this book. are explained for the
`Texas instruments TMSHHCED floating-point DEF. A.
`floating-point DEF was chosen for these experiments to
`
`