VViIu 1(nv1 JELSCK". filr er (163315.511: examiples for several win (1 ovvs .
xvi 1') (1 our ' 6. . AP dB
rectangular 0 .09 0. 82
Hann 0.0063 0.055
H alnining O . 0022 0 . 0 l 9
Blackinan 0.0002 0.00 1 7
Discrete lTime Iiourier 'lransform (DTZF
gg?)
m +49 4
E Viva $535. ha. MW (4 W
%A an. new cemj
baaHm sea 4.42 6s a W
M
xszwvd?
a. (ii-5+
a. 41234:. Ceramic-J
q 52.? 9 Lame:
UM a, A% 4" t? w (Art C) J? $3 a: 'm-*Sr\ My 53:51:15.3 +5 1114ew _ JAN Us
K Numbers: ' '
are they?
sicaily, a complex numb
v-based FIR design:
a (causal) lengthL GLP filter h[n]
Fy desired signedmagnitude response D(w) 6 R,
L
2
dM $4MMe
Eute desired
wdelayed iimp:lse response:
ca u 5:3 I d OW w [7:- '0. i 1 . l l
thprw
edOmain windowing 42> freq domain convolution-
madel: Fflmw, Limw%m smu 5:"
previous examples are special cases of
lationship between the input u[n] and desired signal
epends on the application, and
>21! is to design filter coefficients cfw_f*[l] to minimize
the meansquared error Ecfw_|d[n] y[n]|2,
My; anwwgwgm <73;ch
35M windowing a sinusoid
5Wf 1W,
H multiplicationmgroperty then says
1 7r . .
_ / X(63)117(63(_)d
7T 7T
3 E 3% gfgwofW)Wcfw_M
k W
.w fmw;
WW
magma u. f mg
Eggs? gag mejj
ausal NIength) rectangular window" w[n] know
H
W<jw) : 3
- 'WWEW wan?
WE
8%. 5"?jxffr
' 3 "W W5 aim
' 3 g at gag-gm cfw_a 3%? E )ieif
xgw : m' all: gavel; We jig
3 gt"? 3: gag-cfw_j
Wmmrmx h w
A; 65:" CC C ass WC
' EM? W xtwm . m Wumem
Imple. 3 Egg ht V ygmfm L w
t WWW
1* afx m
y[n] 2 0-523[n] + a:['n, 1] +~
:g and Reconstruction Summary:
uency domain, we saw that Sim: reconstruction
-FF, with gainT and cutoff 7% rad/sec, to 51.1912).
SC(j2) m X(-jQT), and thus when 1:05) is
cl to f2% Hz so that no aliasing occurs,
l, w ~s~n _ v E?
y, we have 23?
sa
Eiiitfat
I
as.
Q-
( -
. _ i "' a?
. _ ,I
r I
as Ham:
aiinurunll frequencies:
TFT allows negative frequences What are they?
~ .52.: 2112;
that a can be (g3,
Ias ad " i 211" a (SI"-Szc> 4%
gnal 639015 is a corkscrew: for 20 > 0, it spins
rclockwise and,
ems m input/output behavior:
timeinvariant (LT!) systems are a very important
.nvenient class of systems:
e input/output behavior of an LTI system is
npletely characterized by its impulse response!
tetime case:
3 impulse response of H is the signal
Twig
30 Syllabus Highlights: we?%
a materials at www.carmen.osu n
abus
tu re packets
:ompiled from ECE52OO and ECEGOO
authored by P- Schniter and L. Potter
nned lecture notes
ignments and solutions
uple test questions
:5 to online resources M aw" mums,"
W W B
I and Reconstruction:
PE, Camitkumw cfw_We 45 W $11M .
now, we have consudered continuous-time Signals &
1s separately from discretetime signals & systems.
ill now connect them throu h uniform same/ling:
weaned-gnu . 33f.
sch/z] = $(ILT) for n E Z
hafkaax
XKW amiw" m mrj ma: Wiw
INT ~- important properties:
mme $[n] JDTE; X033) and w[n] 315; W(ejw).
ity: , ._
04:1: ['12] i Bu) [n] 233: wXQEw if" cfw_gwgjm
gate symmetry:
a:[n] E R 4:?
modulation:
)lication/convolution:
W . fwwg
:c['rz]w['n] m 34"; a, XMQ
m2 38 E a .322 E8; u._ 33% i :83 5me NE
T. 74. $3 ia
a. ,4 \ :onstruction: M
II now detail the reconstruction of waveform
Vt from its samples cfw_m[n]vn.
ghout, We'll assume that 33(6) is a lowpass signal
r CTFT/DTFT derivation, we viewed sampling as:
w
sampling -~ example:
mmumw
se that Xc(j2) has the CTFT magnitude below,
'flmax u: 60 Hz_ cfw_MEQh mfMww %3
m
at sampling specifies that % > ng 2: gmgg
gures below illustrate the Nyquist zones and the
ng DTFT for several choices of sampling rate w.
is le
g - DTFT/CTFT relationship:
dejrstandmglimgging we need to derive the reiationshipm
TFTHX (gsand the PTFT X(eaww WM
w. XMNW, _m
x: 3 13 n W gigF 13
a. i ban k3; mi
2 xccnmd") Win? We
SQSD _
- -r- QYFT c5 steal
:all 800 the continuoustime sampled sig
Wwxmvmn .W
FT important transform pairs:
Wwwyw warm
TFT is often written in transform pair notation
a:[n]<+ DTFT <+X(ej)
important examples of DTFT transform pairs are:
mplex exponential:
jwon DTrr 21,? E g (a; ma '4" 2W3
6 E. Em
anec ker delta: maa 7WW
E; b N gflg"
E' m: mwk' . :2
H a cfw_mm5,
5-! x cm I: a g V j 3 >96:
Wkt Sham W
9: Mariam-" a. rv' makeg 52
WW W Wisp:
fuss.
:33 a xdr-i
x: 5, w
AV
3; m Wag a ,
Egg Knwm") W
Kg M 4px: Dug-cfw_9
Mmw M.) 2.; (Ll15.
' 1" :2. 4-9 <22.qu ems - convolution
9mm mm.
m 313% ag; icrete Fourier Transform (DFT):
*FT as a to a na :yze the
may content of :_ 7 nals-
*FT discussion wiH inciude:
:inwigmions of the DFT and IDFT,
erpretations of the DFT:
sampled DTFT,
discrete Fourier series,
,perties of the DFT, in
()0 Midterm Exam #1 Review
grin exaln will be held 9:1 01020521111 Friday, February 17 in Bolz 436.
T tables will be photocopied into the test booklet.
are allowed one handwritten 8.13 x11 sheet of notes, twosided; 'lhe sheet is to be
ultted with the ex
E4215.
InterPO'ation: Cggigg, ggg
o Interpolation goal: increase effective sampling ratefrom
11 Hz to % Hz. We can accomplish this in two steps:
X(ej)
ll
ll-4
M8
All
u.
A
8;
a la
:53:
V
V
' a ~ Mm g $5 A f t, W W: a 1
( (J > @955; f: 2' W
ECE 5200 Syllabus Highlights:
Course materials at www.carmen.osu
syllabus
lecture packets
compiled from ECE5200 and ECE600
authored by P. Schniter and L. Potter
scanned lecture notes
assignments and solutions
sample test questions
links to online
ECE 5362
Computer Architecture and Design
The Ohio State University
Department of Electrical and Computer
Engineering
Instructor: Yuan F. Zheng
Course Contents
Study computer organization from hardware point
of view and basic principles of computer
archi
Sampling and Reconstruction:
Until now, we have considered continuous-time signals &
systems separately from discrete-time signals & systems.
We will now connect them through uniform sampling :
x[n] = x(nT ) for n Z,
where T (in seconds/sample) is the s
Memory Locations, Address,
Instructions and Instruction Sequencing
Read pages 28-40
Memory locations and addresses
The simple computer is a good start to
understand computer organizations
We need to study
how data/instructions are organized in the main
Number Formats, Arithmetic
Operations, Overflow
Read pages 9-17, 336-339
Numbers, Arithmetic Operations, and
Characters
The simple computer is a good start to understand
computer organizations.
It raises a number of questions:
How are data/instructions
Bus Structures and Counter Design
Read 228-237
How to Connect the Registers Inside the
Processor
Fully connected
- If there are m registers, how many connections?
m
C = 2 (i 1) = O(m 2 )
i =1
- Problem: connections grow quickly; expensive
Bus structure
Simple Computer Example
Register Structure
Read pp. 27-85
Simple Computer
To illustrate how a computer operates, let us
look at the design of a very simple
computer
Specifications
1.
2.
3.
4.
5.
Memory words are 16 bits in length
2 12 = 4 K words of mem
Addressing Modes
Read pages 40-48
A CPU uses multiple addressing modes
the way addressing operands
When working with operands in memory, the question is
where to find the operands? It depends on the
addressing mode. There are multiple modes in carious
C