100% found this document useful (6 votes)
3K views

DSP Module 5 Notes

The document discusses digital signal processors and Von Neumann architecture. It describes the components and design of Von Neumann architecture machines as well as how digital signal processors work. Digital signal processors differ from Von Neumann architecture in that they have specialized hardware units like a multiply-accumulate unit to improve performance of digital signal processing operations.

Uploaded by

Kesava Pedda
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (6 votes)
3K views

DSP Module 5 Notes

The document discusses digital signal processors and Von Neumann architecture. It describes the components and design of Von Neumann architecture machines as well as how digital signal processors work. Digital signal processors differ from Von Neumann architecture in that they have specialized hardware units like a multiply-accumulate unit to improve performance of digital signal processing operations.

Uploaded by

Kesava Pedda
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

01

Modul 5: Digt Siqnal otous


Ttheduction (Ds) ro-
Dnlke HECLobRotaseets H(co ConluoleDiaTal Sig
ssesAhase Specda e r s et Sequia bpralions Such as
FFTAlkaing, Gonvolctinm 4 CoAslalien, dud bloeg bas proshghasdsrai
heyetaDS Roanots kea dilfwnt dedicld
asditic
B Kaooiug Ds DRocas%R archetues,l t Aave u ka
dJ [heue beton Von-Numaun tbalasd architietue

Von-Neumaun Aichitusu VsHasvod ArdiLut


Vou-Aleuwmun Azulluu Hetward Mchlida
t a tturitLal design based 1 a Hodatn armpe ancktee-
OnAud-pRogpaun Cerpau ertptte basud en Haruarel Hak-
elay bosf lompata Moda
Rt es Satysicl Nanot es Sepeiale dmemot addrs
addiess Ansucn data o uslulions 4 data
Data Easlus 4 sllicn 3. Data tausus 4 s ucor
atcku lanuot bi þeulorind nuul-ethes Can be pszotuad at za
-lauetLsL Sanatiu.

4 E c o n o }usuclions l 2xuion uyuc Lend a


Seltiv Blrs. astuy
5. Szutter o iuiuuclien Tates mota S Sslen oLuskachion Tates s
puslueie ael Äuskudion gicle
6. Siples Costuel uit dsign devepG. Crue lui goto busas
moe

-meut ne is Cheapes 1as Complicold g CstH,to desig


1UAdinpusanaB mmpdeu, la-|7 Usd Miaeinuolaas
-tops, cemsal puje piemeis Scakc applcine_kt ep,
(02

have loo Pe g Aachucta Vou- Aleuans Ha


V ahetctuus
u this t us disGs bot t Aachiteluorsb Crili
ttisAppkicalions Such as qenunal purpee ploGssns o Ocki-
Ven- Newmaun 4 Dgit Sigal lckass ohordor
tce
GenetaMiaopeda sser base e lon-Nounany Aseliteeue

dey FogRo2
sáluy
Addsbs

Arlme PAogpon
nput/outit urt data manu

Prta ba.

Tkegude luous ou Vou-Neumann basd qeneea purpo


icaopidsghArditrclune lorsislg o
data
Singu Shard Metnpty o Rogkan
S d a l bus o menioty alces
3. Anauthenata nit
o a m Corte t
2Th proGsso
atput detios
piotes n a Seria) yslin an tans t l g
lxealien yds
thu acles e fetehiug 4 exeulion ae Operutd by-fething
an usuchin hem mame ) detpdig- t Va he piegtamn
03

luit instucher .The sRotss 4 ron


Cortse 3 finall txecetng as

An t q u helors.
Sene ossd prds
UUL _ suion
n a Serial ashion, At
Cas Tmst lht to

feteh xutan SeGend stayu Klait Sat

T gue cluasly inticats, , t th geh eblais t op.code


Rem hne t Certuej lut oill"'deted ta isluclien 2 ddn-
mihe t t heialin
Net, tu esecien yda ill deledle iotuaien 4 medil Te
Conlut e egists oma
t n e te Saiutaen 2 Comtd, rotss belteh rotulinctin4
Cernt
la pouyo oprals ene s u l i o at a tima u aSuial asr
becawse it has Sugl Shuasd memdty fot botc pAagaam data

Digta Siqual ploaua based a Hauwnad Azcttu


Adds PReaan
Ccrt t
ogau Mert-Adeus bus-

Data meuo Addaess bus.

|Artnac PRepawM Daa


ptputht
davios
PM Dbus
DMd a t Ls

Autiplza SSajt
t
04

F.q
-ANclitito Consising
Sepiat Aeuuly oReam dala dan
a Sepuat bus (Addus 4data) fet y
3 A ahmetic lluit
4 pR9Ram Corito uit
S Tuput otput deuas Additanal MAC hit luits.

h DS potsse bas uo psate mehuc bu s o pega-


mCodu3 otus s o data
2 Ttas lso umois Seprali Tso addsus buse &lkoo dála
buss ae Ud.
Bott th Teneis hav ai ey lemduns l to jlogtany D
dhbus dati Tnonao bus leszacti
So, tis_pietsseh Can ezh isiuahey 4 data ~u paale at
he ha
PAosssha, au addtiona t Gled ultiplin A Accurmulalo
(MAC) ehich is Ud r digata_j:ltaing o4ezalien as au haidsa

dung d tg
haiquu Shoros u iioved Speed o Ds sotuso cpialin

Rth Extin
OVlap
oVula teteh Execut
(Ppenig).
05

Call asipeling oealen.


1e Ds
podsso Knsbtnus bu txcadior Cyel ktul also Jethiug
Ht neat studien to'e exatut.d» Kahich ntatas ploasig Sal

Cnna uubrs pholossets Digitel Sgol PAlasses


Insudon Multipl.clock, yel 0 dubuclwns a tuuit
Ttquiuo etli) 0ua-in Siugl Gyeh e ta
iusttadio
RDustuuduen 2eten ustudien sPazall lxoadion
Cxulie Lonys>eguulial fossib
3Meueias Sepaai data 4 ptegto Sepezat dala y piogta
menmdy s abkut. meipt plsat
4Cmputatenaaitnly he ut ie ALU The Seeat. Computa
Eena i t Au, MAz t .
SAduskdata |Addass dala buy au Addsess daa kus a not
bus Nuttteriug Hultiplexad mul tipnod
6OpeiandFtk Operand fetke Huultiple Gfeiands u be
Sequial Hetkad S ultaneust
ppeliu uug peramed Ppeluig s mbiia
eaplcat. by bu 9uu Tgis tartagh utudien a s a
Sappdt pypeluing 4tulion Cache
8. e chip Namall n chip Gcha mrn Plongia daa mametis au
- on Chip 4 also in
Chip Mamo psnt, man memolyE Puit
chip.

-Appkcation Genutal pupoi Spe R<3kp| Squa piouginadaag pio


lo Cost sexpnjve highy Rxpensive
Sas dsign Conpicat in dusiq
(06

Dig:talSirPhotesscs Iasd Lbat.


Uls Ola
e specéa doktat.J Ds pAotase hatdhnta.
Multiplies Qud Acowmulaton (tnc) a s i t 3 ldus ewmales
Hultipli aud Auwmudator (Mac)
ha Ds potasses ses t MAC, a Sheuia hasdsoau lait ot
enlaucig tt Gpud dgt) kltuiyg
MAC Luit fuigoms multislalan acunulaien pRos
Basce HA luit Consists o multiplu, adde 5 acwnulaley: Jhka
Sess au bape stnclne t MAC
Opeaun epeiand he figuu shons anp:a) haidon
MAC. multisi Aas a paiy i p
YRgistey Sagatas, eash holding te l6-t ilp
t utiplia,
Muabbplis The lesalt et multipkcalan is
acunultd u te 32-bt aCaumulaleA
Acceulat let he eut Ragistu Aeds t
32 deuel teiiom Saa t o Aluwmulte
RzltF

RShitus
b Sevenal bt fo eaeh
ata oond.
10 Sped up Sueh salion, th pruia hadsar i t Uaits
dosigre o peryn Scalig operal.
Sspeccally cligitltang, unnt ovufland, a Sealing
Cfnalion guiud
Stalen is faltud sag Sli/ting ellods
IScaliulron a Seugus.
07

Scaliu dosn. This opuatien l t data To Ma t


Aroction
, Sa,uet aas
t

dividiug4 dt. data by 4 Aaasig tttoction fut


Sat. S 3 bit data ol, = 310 sshiigst oto1,ight
1
0o o 3 I.S>
Scalig-Up his opualan sijt data to di ft, hanm a
Hultiplytug ta data b *
Sa 3bit dah: o1 3p Skiyting OlI t At
10 bjo 3xa : 6 10

i bos Blitta fsfotaes Soalug obtalion


3 Addis uensale
he s Ros Jenesáts h addesses each dalun on
Aia data bya to be fietessed using au medhanism Callod Gcada
ofuagTkaus Lalors desozb t basse muchanisu o-
tucudas bing to a
butfa hag
K t igt data Sampls

ln4) H , a poiula is Ud h9lach points to


xln-1)
tle asest Sauple dat) in) as
ttn-s)\ Sheion u +gulu
slan ta next kauple is oblainus t
aln-3)
aln-s)
oldust sapl i ln-7) s ghud out
u s Sampl data is placad 1(n-7 at
Such a t n-j) biems loalien utant Sanpla
tis aoss Certiws Sait data Supl,Suh lat only ne
kotton t Cucular bta ued, to upthdA.

t lorsidur, a 2bit Caraulat brlfe krtk a datalns (a,b.c,


ef ADC as hensm intigue
08

Port FIro
Post .GuntE
dala Contois
equivalat C data a, b,cd i k in

Poiuting to d os uut
data
Past
data

Cuxant
e dt okau e lomes n , a2
huivalntt

e Dat
oltest dta 4 poiuta
Poiut b Poiut to e.

d + dat IhnComas n, b
pusud out as Zt
e
et Dat aldst data 4 Pbiuta
Poiut t a s u
Cust data
Du al ti Casus, l aclar bfa
o
locatien is Updatd Piula
iut Xocalizn es G u t lcat
But in FtFo, r e data pusad, u FIFo has t petu hea
Data moVenut to get utet Leatior u peiuta peiut a

ircont Ds tRoaso . flooting priut Ds #losso


dpit DspRocei ied poit
thase Ds pioases Cpu ehs Hauiplalien oj data siug
utrge artmatie
Au Ovehus oTMS320CS4x ptoaso- dveleped by Texag
n s t u t s disume os kxadoit Ds ocassd
Th Shouos an tyria THsocs+x fxa Poiut Dsp archta
09

regram Program
controt unit

datamemary address b
Ddata remoryaddrens buy
E data memory address bus

Data
Program
memory memoryy

Program memory data bus


C data memory data bus
D data memory data bus

E data memory data bus

inputiOutput Arithmetic
devices logic unit Multipliert Shift unit
acummulator

It Censisti o RAM, RoM, 4 data buss, 4 addus bues, Conpataliane


[luit kte AW,HAC,sdit tuit, 40 bit actumulalev,
dala
RoM Used opiegian Code, iustucen
C D dati mancky addsuss buges 4 C D data mamel data buj
deal oit tthiu data sem t data mambiu lalile t E data munau
addess bus t t .mionmcly data Lus a osed t Mow data snlo
dala mamety
E cly data bus Cau accs5 Io devias
ALU Aas iudpenedet accunlatoty (4obit) GpalL qestng
G0 bit addile
cltiplr Cau feth T data flom C D mamo data bus 4 Aht
Aata Via E Aata manAbs So ttat it apadle t lave 76t X
I1bit Hultiple ata
T o Spea up mamey a c s 2adduss genealots Q plaseatt. ou
oi pAelan makot ou ol dah muo
10

Teales o MS320 CS4x


|Advou Ce Multibus Asehtectau Asit 3 Sepuat l6-bt datamemet
Buses aud DueMogtan memoly bus
40 bit ALU 40bit Slita'2 indlaeudest 4o-bit actunulalots.
7x11 bit Pakalle Multipla fo Ha Ohetalion
o adaduss geutalots itk g auilan Yglas 2(AR AUZ.
studens ixtk a 32-it donq oo1d opeiaud
sudens itk 2 ot 3 Opetaa Teals
AAhoneke iustudiong Nit pasall lood u
aoiuman lou por Cenkuuptiory aschitelau
& otngoint DS pRocessos
u Thase DS piotussois Cpo sotus Oporali6u data sing
TAaxttug poiut 'attmaliz
t e e bRodsss kups gt rid Hes udh s evetlanos,
Tound-o exas, nalicn tudk, ate aud isik ligh Spea ale.lalion
PcisinaRovids indly bividenmet b dueles iuplemeut Ds
algeithn
atiqu Slus au tipicl aselitet TMS320c3x daveloped
bTexas2smat disussed as Heabg poit Ds plocssd
The plocasse ras laagn menmoly Space4 s equippe kk duo
aitess on-hip mumois
Tla Roga Cacha is mployd to ehante ta uliom
} (omon
Au Sepesat memot.uss f deta buys t t Jot DMA E erotuy
OfeAalon
Cor Cueut 2l0,4 Cpug perzphesal acess Sueh ás Seaiol at, o pit.
The Cpv Comlains tte lootiag foiut/iatae multëpliy , an ALi
elich is ued s opuati bok idaza 4ldtiug Poiut aitnte
A bane Sliu , tunal bures, didiatd Auxilany Pista Ari
- matia Leik (ARAU)
RAM RAM
FOM
Prograrn tiocx 1 BAock
XRDY Cache
blcck 0
(1K32) (132 (4K32) JoSTRS
TOSTRB (64x32 KR
XRIW
Oata b u
X-
KD31-0 KA12
XA12-0 MSTHS
MSTRB DAA
RDY CPU
INT3-0 Intege Integer
Address generators
IACK Floating-pointFloatingpoint Control tegisters
XF1-0 Multuplier ALU
MCBLI MP 8 Extended-precision
registers
x2ickIN Address Address
gensrator0 generaor 1
V 8 Auiliary registers
SHZ
12 Control registers

tn u bit nlg, 4
e rulbipli opeata Sug ye mulbipkcatien
32 bit oatiug poiut Valuis.

ALD.

T SpesiaJuclcrs e t ajsus pucud 8 exded 4o-b:t phrse


Keajiatos t maiutaiuiua acuiaty {eatiug Deiut luulg.

meli d niuo Tpdak kagu g Vaiablu uby uiuasing


oa
eatws TMS320C3x
allensiua Opuattn q hwnbus Wika
1: I5 Heot ig foiut (apabill
l a r g duomi langs
aThoiotingbeut fuatk o Sutptld: A Lbit {loting poit a t
* 4o bit eland.d Preuisin eanat
A 32 bitloidiug Poiut fotuat,5 ba
otAu t dikent past e a l i n givun n explaalien a
Awit fo hat aso.
12

Taco oit dud Jlovting Poittea


ined Peinto uat
it Apesui data 2's omplamat uug fonat 2
aipelas data usiug intge aihnale.
Kaug a
Jkpusets oula a Vely nats d a
ntcqey vauber, a Xoblem Such as Owln 0 data maip.lalo,
naOcCus

t Coridg
us bogn hoitt 2s Cmplmat ppoeilatian b Conidiag
3 -bet als"Cemplemeut decima) n u r in Taug - 4 to 3)

Dcml lumber 24mpanuil


3 oli 24 Cmpemait of +V numbu
Bane u biay but
60 2 Cnnplonut -Ve umbe
000 is otaincd tng Coplenut
1U
2
-3
-4 00

enept et OVesle Occus dutig multipla


Snanpl a) ax-) L) 2x-3)
Sela a) axl-1) Tu 2 lenpumant Ooolo =lI1o
2 0lo O0 xOo) emove to Catinded Sigu b
ol e get lo oich 4 otking bt
-2 "ohich is illay lagl4t 3)
13

b) xt3) 2- 010 0lox O The asoolonut o oolo


= lolo
3 OII
OO0
T lernovug loo extrdod bi
kOo get olo budh s
but e Oaut(-6) .*This inditat, Ofons Jas Ouid in mult
phcalion olith s out
q ug -4 to 3)

Pastial Sdutien to owuleme bello du to multplicalimumber


ined oiut epiesadalion s b lsg Sgud actiona
Sheme Calld as -forhat

-eat Baadena numb Jouat alld a-ftmat. f


I t s availasla as - 2 , Q-3, Q-, Q-IS et.

mst lmmeny he
-15pimat is 03 ed- peint DSp
imdemattio is slpn beles.

Binay foiut
The 4igm Shoros -Is ; data ase n a Sigu maquitud Joi

Sgu
ANt t Siqu bit, ta dot nptas tlu binant poiut.
Ractina ang umbe lm -tol
dividcd auto 2 uaa, eadh hit Sze
ang is a

J h mot mgctive umbis -1


The most Positive numb l- s
14

n s on diu to-ISJdat
ind Siqud psutakien o ta clcma) nwnb 0-S6O123
Q-IS
Sol -i5 Apiescitatirn s OHaiud bHuutiplying 0:S6ol23 x 2

o times 4- noting Casy 9eneset.d as dlma) paut Ac


MsB to lL8
he get .Sbol23 X2 egt
I120246 I(ks8) l 00ol1l|oNOolo

Sa, tt iveu aumber s tve Q-Is tRasealalen is done itout


2s lomlmut uit O as Sgn bat.

0.S6ol23 .1000ll||o l| oo lo
Not:eu ued to multipl a ls-tisy
FR te Siqed -Is plesitalen Ja dacial Tumb-0: l6o1a3
Selu st AutQ-olt by Cnsddelina+VeTUmbe

he get
lbol23 = 6.00lolooolllU|O
hu yt a t g 23, lontlit, Q-Isouat beemes

0-bol23 1loOoo00o1o|

Sobvethage iolung
d t Saed G-s hpeseallen or ta dacina) ubey e
a)0-25603 -0.2-166|23 )-0356712) d)o.4798762

Pholplenns on Q-s ta dednenumbe


Crvet the Q-1S Sigud mbu l |lolol|o0000lo
Se Siua tha umbes is megatsa, aplaing-2s rplamut yelels
0.0ololo0o||||||0
5

datimo aumbuu as

2Convet Is nunbe O.loool||DIDojo to


Stqud duiual nuyb

24S+64s4+2?4204 z+"=O.S60120
Solve tesepons
Convt ta Sigd Q-Is Mumbus t ecina wnbuss
a)1.0lololmojõool0 b)0.00looolII1ollio 0.llolo) D0Ol000l
d) 1o1000|00 10I

Aobems Ou Addiug troo bes Q-IStat


-Ad e tioo ntm s an -Is elauat 1lolou do0oo l0
ael O. Joool|) 611ool0
olol|| doooolo

eult |s 0 01lool0ollolo0
clecinal ey 3+26427t 0.4oo024

SolvthusAod
1eld th tlenotug o -IS numbers
a) lolo|D I boO0O| +0:6lboo l1||ololo
) b.0ololololo o000) 0.6|0)0 IIolooIO

e 0 0iO)O10100000 +1o1 olotiIOloolo


d 0.0o)olo10|000 00* 01OloT01001o
16

Tloatiug poiut tnat


t fsusut' numl Wsiug aymatiMa 1otinol ot)
4 a xpenut n add.licr tha intap frust rsts
data Uiai Hlootiug Peit aaitmae
It pieti a"wid ang g g dala Jo aveid oveln

loting toint Tnn Auploy.laken as


Jcard Fat e
ven by
kohu H t Hantisna (actionas part -totaat)
E M e lapenut
dlocting point nuinb s l6 bit umber in otich
n
2 bit a asiged f a Hautisa
4 bi ailgus apnast
4bit per 1z-bot Hantistg kt Siga bit

Siuu, l bt maxtima n i t bt" lngr -1 ),tt r u n


b Owquud të expenut emlaes clranie Xauge The bigg
uub biz diignatj to enpenit, Jangu dgneuniag

Eaaple
Cnvet each q t l e o n deima Tbes to a eating poiutt
aO.lko1230 ) -20.140529
nuab Uiug fouast

Sdn: a) o.lLel230
Sal devern 0-1Lol22o Ol4el230/T2 - O640492 i t
Ou erperent q - 2 fo (+v runber)
.
17

2
0.640492 X 2
Jolwe 2s n.wt 1o
jaete -2 u enpent
Qud Hautisa totud ing -
o.b40992 s al
1 mT
te t l01O0ollUU)Not; Solve by xti t t d G1arnt
leis Casadng Hantissa 4 ênpenaut, 1e gt

Ioolol0oo|||||
b) -20 4 30S27
Scal doon -20.430527 -20.4 30527/ = -0.638454
ith aM tnpenent 0+s r -Ve munb)
he mumbi à xllu as -0.6384S4 X2
aluse 5 C n enponat Fotus Value is 0lo) aud
Maudissa e O-6384sh s jund i -tat t Jt
6lo100ol|ol
l o takiug 2 mlrmeut t get -0-638454 = l0jbli|0olo
Nos Ca Cadiug Hautissa Qud exboneurt to unt -

0.63845Y xa
e get ololloo|0olo

Sdve ttus ploblens


Corvet each te olleroing dsima ruubes to a oatiug poiut
b e wiug tt ahat.
a) o.llol23s L)-lo.430S27 25568922) d) -o.6789o3

You might also like