DC1 Ec
DC1 Ec
3 0 0 3 3 25 50 25 -- -- 100 3
Week Topic Taught Remarks
Faculty will make sure to remain in cabin during this time slot for the
session
Monday 04.00 pm -05.00 pm
Purpose
Reconstruction:
SAMPLE AND HOLD (S/H)
By using flat-top samples to generate a PAM signal we have introduced amplitude
distortion as well as a delay of τ/2. This is referred to as the aperture effect.
However, Sampling theorem still holds.
Reconstruction can be done by passing signal through ideal LPF and then through
equalizer filter which is inverse of filter used at sampling to remove the aperture
effect distortion.
For a duty cycle τ/Ts ≤ 0.1, the amplitude distortion is less than 0.5%. In this case,
equalization may not be necessary in practical applications.
Sample and Hold (S/H) is practically used in PCM: Input signal is continuously
sampled and each sample value is held until next processing unit (quantizer in
PCM) needs it’s value.
It also helps in reduction of distortion at reconstruction end.
Speech
- Telephone quality speech has a bandwidth of 4 kHz
(actually 300 to 3300Hz)
- Most digital telephone systems are sampled at 8000
samples/sec
Audio:
- The highest frequency the human ear can hear is
approximately 20 kHz
- CD quality audio are sampled at rate of 44,000
samples/sec
Video
- The human eye requires samples at a rate of at least 20
frames/sec to achieve smooth motion
PWM:
Generation: Generated by
comparing analog signal with
high frequency saw-tooth
signal.
Detection: Spectra of PWM
signal contains baseband
spectra; easily demodulated
by using LPF.
Applications: Digital Speed
control of DC Motor
• PPM:
• Generation: Passing PWM signal to Mono stable Multivibrator to trigger it to
generate a fixed duration short pulse on –ve edge of input pulse.
• Detection: Spectra of PPM signal does not contain baseband spectra. Needs
synchronization between transmitter and receiver. Detected by using product
detector (phase detector).
• Applications: Telemetry and position control systems
SQNR Improvement:
Overall SQNR of DPCM scheme = Pg/Pq = (Pg/Pd)(Pd/Pq) = Gp (SQNR)d;
where Gp= Pg/Pd (>1) and (SQNR)d is the SQNR for difference(error) signal
d(n). With ordinary PCM the achievable SQNR is (SQNR)d only as it is the
signal which is quantized. Here Gp is the processing gain (SQNR improvement
factor) due to differential quantization scheme in DPCM.
Accordingly, the objective in implementing the DPCM should be to design the
prediction filter so as to minimize the prediction-error power.
To achieve same SQNR, the DPCM needs less number of quantization levels
(less no of bits/sample) compared to PCM.
For voice signals: 3 to 4 bits/sample in DPCM gives same SQNR performance
as 8 bits/sample in PCM.
For TV signals: DPCM offers a saving of 18Mbps over PCM.
The step size Δ of the quantizer in DPCM is
made variable here: updated automatically
depending on whether prediction error signal
d(n) is large or small.
The prediction filter can also be made
adaptive by making the co-efficients of
prediction filter time varying at every sample
time using some adaptive algorithms (for e.g.
LMS – Least Mean Square).
This will reduce the prediction error and
allows to match with varying power and
spectrum of input signal.
Receiver uses the same adaptive algorithm
strategy.
ADM Transmitter
Practical Solution to
problems in DM: Use time
varying step size ∆:
Adaptive Delta Modulation
(ADM).
Use Adaptive Algorithm in
feedback path: Logic to
control step-size ∆
according to input signal
derivative.
ADM Receiver
Practically used Algo.:
Song Algorithm and Space-
Shuttle Algorithm.
Same Algo. at Tx and Rx.
DM: Difference between successive samples is quantized into 2 levels (1bit) : Quantizing
derivative of input signal and transmit derivative of a signal.
So, the receiver of DM needs Integrator (Accumulator).
Also, as signal is integrated (low pass filtered or smoothed) before quantization, slope
overloading becomes less likely.
The L-quantization level quantizer generates digital signal which is an L-ary digital signal.
It can be converted to binary digital signal by using binary encoding circuit (for e.g. successive approximation ADC).
Select the number of levels L = integer power of 2 (so that binary encoding can be done easily) = 2n
n=Number of bits used to represent a quantization level using binary code.
After binary encoding each bit (0 or 1) can be assigned a pulse (waveform) Line coding. (Illustrated in figure for L =
16).
SQNR with binary encoding = 3*22nPg /Am2 =3*4nPg /Am2
SQNR in dB = 10 log10(3)+10 log10(Pg /Am2)+10 log10(4)n = 4.77+10log10(Pg /Am2) + 6.02n
For any signal g(t) it can be generalized as SQNR in dB ≈ α + 6n; where α is constant depending on nature of g(t).
Additionof 1 bit in quantization improves SQNR by 6 dB: But it will also Increases Transmission rate, BW requirement
and memory requirement.
n= Number of bits used in binary encoding; now specifies resolution of PCM or ADC.
In some applications like reading data from sensors 8-bit resolution is sufficient; but in audio-video applications we need
10,12 or 16 bits of resolution.
30
Special Case: Let g(t) = Am cos(ωmt): A single tone sinusoid
Pg = Am2/2
SQNR = (3/2)(4)n
SQNR in dB = 10 log10(3/2) + 10 log10(4)n
SQNR in dB = 1.8 + 6n
W = BW of baseband (analog) signals
Sampling Theory: We can transmit a maximum of two pieces of information per second (2 bps)
per unit channel BW (1Hz) assuming noiseless channel.
Transmission BW required to transmit PCM signal: BT = nFs/2 Hz
BTmin = nW Hz (Fsmin=2W)
e.g. Telephone Speech Signal:W = 3.4KHz, Fs =8KHz, n = 8 bits, bit rate = 64 kbps, BT =32KHz.
SQNR increases exponentially with increase in transmission bandwidth: PCM allows exchange of
SQNR with required transmission bandwidth BT
Recall: SQNR = α + 6n dB
Increasing n by 1; improves SQNR by 6dB at the cost of increase in BT by Fs/2 (min. W): A
Good Deal!!!!!
A raised cosine pulse with excess bandwidth 𝑓𝑥 = 0.8 MHz is used to
transmit binary data with bit duration 𝑇𝑏 = 0.5 𝜇𝑠.
a) Determine the rate at which the binary data can be transmitted
by this pulse via the Nyquist criterion?
b) What is the roll-off factor?
1
Data Rate, 𝑅𝑏 = = 2 𝑀𝑏𝑝𝑠
𝑇𝑏
𝑓𝑥 0.8×106
Roll-off factor, 𝑟 = = = 0.8
0.5𝑅𝑏 0.5×2×106
35
Minimum Shift Keying (MSK): BW conserving CPFSK scheme: A very special
case of CPFSK.
It is Digital NBFM (Narrow band FM modulation index less than 1).
With this properties the transmission BW is reduced to 1.5Rb.
The binary pulses are rectangular in MSK.
If the rectangular pulses are passed through a special filter called Gaussian
filter and then passed through MSK modulator GMSK.
GMSK = Gaussian filter + MSK
• The spectral efficiency of MSK is further enhanced by GMSK.
Gaussian filter has unit impulse response and frequency response given by
Gaussian function has same nature in time domain and frequency domain
α is a parameter of Gaussian function which controls the BW of a filter (B).
GMSK
Designed based on the product of
the filter bandwidth (B) and the
symbol period (Ts). Ts =Tb when
number of symbols in modulated
signal = 2.
BTs = ∞ corresponds to MSK.
GSM (Global System for Mobile)
uses BTs = 0.3, which defines the
bandwidth of the Gaussian filter.
The smaller the value of BTs,
however, the higher the error
rates.
Sacrifices the irreducible error
rate in exchange for extremely
good spectral efficiency and
constant envelop properties.
QPSK Demodulator: