Lecture 28 ADC 9
Lecture 28 ADC 9
40 60 80
of Communication 100
Systems 120
40
Lecture 28
Date: 13/10/2020
60
Dr. Pawan Kumar
Assistant Professor
Department of Electronics and Communication Engineering
National Institute of Technology Rourkela
80
Rourkela, Odisha
navigation symbols
Autumn (2020-21)
1 / 16
Part 4: Analog to digital conversion and digital communications
mm pulse-code
Differential 40 modulation
60 (DPCM) 80 100 120
• An alternative to PCM (pulse-code modulation)
• It can provide increased signal-to-noise ratio (SNR) for the same
consumed
40 bandwidth occupied as in PCM. Or, other way around, it
needs smaller transmission bandwidth for the same SNR as in PCM
• It mainly works on predicting the present sample using the previous
sample
60
• The prediction is better when we consider a large number the previous
samples for the purpose
• A better prediction results in smaller error sample
80
• Smaller error sample input to the the quantizer results in smaller step
size and, hence, smaller quantization error navigation symbols
2 / 16
Further discussion on DPCM
• In mm 40we could visualize
the last lecture, 60 80DPCM 100
that in 120
we can transmit the difference of sample m[k] and and its estimate m̂[k]
obtained from the previous samples (d[k] = m[k] − m̂[k])
• At receiver, we add the received sample d[k] with the estimate m̂[k]
40
obtained from the previous received samples to get the present sample
m[k] (m[k] = d[k] + m̂[k])
• At the transmitter and the receiver, the same algorithm is used to
estimate m[k]: m̂[k] = Estimate[m[k − 1], m[k − 2], m[k − 3], . . . ]
60
• But, in actual, we transmit quantized version of the difference,
represented by, dq [k] = d[k] + q[k], where q[k] is quantization noise
• Hence, at the receiver, we quantized version of samples m[k]
80
(represented by mq [k])
• This diversion results in error in the signal reconstruction
navigation symbols
3 / 16
• Hence,
mm to minimize
40 this error,60
we need to do
80 appropriate
100changes at 120
the
transmitter end
• Instead of estimating m[k], we can go for estimating mq [k], the
quantized sample
40 we get the updated difference
• Hence,
d[k] = m[k] − m̂q [k]
• The quantized sample is (this relation is used at the transmitter and the
receiver as well)
80 mq [k] = m̂q [k] + dq [k] = m[k] + q[k]
navigation symbols
4 / 16
Transmitter
mm 40 60 80 100 120
40
60
Receiver
80
navigation symbols
5 / 16
Adaptive
mm DPCM 40 60 80 100 120
• The efficiency of DPCM can be further improved using this method
• This method allows adaptive step size. That is, it can be large or small
40
depending on whether the estimation/prediction error is large or small.
• This way, the number of bits needed for encoding is reduced
60
80
navigation symbols
6 / 16
Delta Modulation (DM)
mm correlation
• Sample 40 is further exploited
60 80 100
by oversampling 120
• Oversampling results in increased correlation between the adjacent
samples which makes very small prediction error
• This40prediction error can be encoded by a single bit
• It is very simple and inexpensive
• The transmitter and the receiver are same as in DPCM with a time delay
of T60
s in the feedback path – first order prediction
7 / 16
(a) Transmitter and (b) Receiver
mm 40 60 80 100 120
40
60
8 / 16
Numerical examples
Question mm 1 40 60 80 100 120
Suppose the Fourier spectra of signals g1 (t) and g2 (t), given by G1 (f ) and
G2 (f ), have frequency support [-100 kHz, 100 kHz] and [-150 kHz, 150
kHz]. Determine the Nyquist interval and the sampling rate for signals g1 (t),
g2 (t), g21 40
(t), and g1 (t)g2 (t)
Solution
• For signal g1 (t)
∗ Frequency, fm = 100 kHz
∗60Time period, T = 10 µs
∗ Nyquist interval, Ts = T/2 = 5 µs
∗ The sampling rate, fs = 1/Ts = 200 kHz
• For signal g2 (t)
∗ Frequency, fm = 150 kHz
∗80Time period, T = 6.67 µs
∗ Nyquist interval, Ts = T/2 = 3.33 µs
∗ The sampling rate, fs = 1/Ts = 300 kHz navigation symbols
9 / 16
• For signal g21 (t)
∗mm 40
Fourier spectrum/transform60is FT[g2 (t)] =80G1 (f ) ∗ G1 (f100
1 ). Due to 120
convolution, its frequency support will be [-200 kHz, 200 kHz]
∗ Frequency, fm = 200 kHz
∗ Time period, T = 5 µs
40
∗ Nyquist interval, Ts = T/2 = 2.5 µs
∗ The sampling rate, fs = 1/Ts = 400 kHz
• For signal g1 (t)g2 (t)
∗60Fourier spectrum is FT[g1 (t)g2 (t)]= G1 (f ) ∗ G2 (f ). Its frequency support
is [-250 kHz, 250 kHz]
∗ Frequency, fm = 250 kHz
∗ Time period, T = 4 µs
80
∗ Nyquist interval, Ts = T/2 = 2 µs
∗ The sampling rate, fs = 1/Ts = 500 kHz navigation symbols
10 / 16
Question 2
Determine the Nyquist sampling rate and the Nyquist sampling interval for
mm signals 40
the following 60 80 100 120
• Following the steps of finding
(a) sinc(100πt) the solutions for part (a), show
(b) sinc2 (100πt) that the Nyquist rate for parts (b),
(c) sinc(100πt) + sinc(50πt) (c), and (d) are 200 Hz, 100 Hz
(maximum of 100 Hz and 50 Hz),
(d) sinc(100πt)
40
sinc(50πt) and 150 Hz, respectively.
Solution
(a) g(t) = sinc(100πt)
∗ Finding Fourier transform using duality property, that is, if X(f ) is the
Fourier transform of x(t), then Fourier transform of X(−t) is x(f )
∗60For x(t) = AΠ Tt , we know the Fourier transform is
X(f ) = AT sinc(πfT)
∗ Hence, Fourier transform of X(−t) = AT sinc(πTt) will be
x(f ) = AΠ (f /T)
∗ Now comparing g(t) with X(−f ) = X(f ), we get A = 1/T, T = 100
∗80Hence, the Fourier transform of g(t) is
G(f ) = 1/T Π (f /T) = 0.01Π (f /100)
∗ The frequency support is [-50 Hz, 50 Hz]. Hence, its bandwidth is
fm = 50 Hz. This suggests the Nyquist rate should be fs = 2fm = 100 Hz
navigation symbols
12 / 16
mm 40 60 80 100 120
40
60
80
navigation symbols
13 / 16
mm 40 60 80 100 120
40
60
80
navigation symbols
14 / 16
Question 4
A compact disc (CD) records audio signals digitally by using PCM. Assume
mm 40 60 80 100 120
that the audio signal bandwidth equals 15 kHz.
(a) If the Nyquist samples are uniformly quantized into L = 65,536 levels
and then binary-coded, determine the number of binary digits required
to encode
40 a sample.
(b) If the audio signal has average power of 0.l watt and peak voltage of 1
volt. Find the resulting signal-to-quantization-noise ratio (SQNR) of the
uniform quantizer output in part (a).
(c) Determine
60 the number of binary digits per second (bit/s) required to
encode the audio signal.
(d) Forpractical reasons discussed in thetext, signals are sampled at a rate
well above the Nyquist rate. Practical CDs use 44,100 samples per
second.
80 If L = 65,536, determine the number of bits per second required
to encode the signal, and the minimum bandwidth required to transmit
the encoded signal. navigation symbols
15 / 16
Solution
(a) Nyquist rate fs = 2× signal-bandwidth =30 kHz.
mm of levels,
Number 40L = 65, 53660= 216 . This80gives number
100of bits 120
16
required to encode a sample as n = log2 (L) = log2 (2 ) = 16 bits.
(b) Finding SQNR
∗ Signal power P = 0.1 Watt and peak signal value is mp = 1 volt.
∗ SQNR for uniform sampling can be obtained using relation
40
SQNR = (3P/m2p )22n
∗ Number of bits required is, N = 44, 100 × 16 = 705, 600 bits per second
16 / 16