EXPERIMENT NO : 1

AIM : To study Sampling and Reconstruction of analog signal . APPARATUS : THEORY : Trainer kit Scien TECH ST 2101 Patch cords Oscilloscope

DATE

The aim of any communication system is to transmit information from one location to another. In case of voice communication, this information will be speech . The signal which contains the information to be transmitted is known as information signal and in the case of voice communication this will be a continuously changing signal containing speech information. The aim is to reproduce this information signal as accurately as possible, at the distant, receiving end of the communication system . The voice signals are limited to the range 300 Hz to 3.4 KHz, a 1 KHz frequency fits conveniently in this range and can be used to demonstrate and test many techniques used in communications. In analog communication systems like AM, FM, the instantaneous value of the information signal is used to hang certain parameter of the carrier signal. Pulse modulation systems differ from these systems in a way that they transmit a limited no. of discrete states of a signal at a predetermined time; sampling can be defined as measuring the value of an information signal at predetermined time intervals. The rate at which the signal is sampled is known as the sampling rate or sampling frequency. It is the major parameter which decides the quality of the reproduced signal. If the signal is

sampled quite frequently (whose limit is specified by Nyquist Criterion), then it can be reproduced exactly at the receiver with no distortion. The pulse amplitude modulation (PAM) system is analog system where the train of pulses corresponding to the samples of each signal are modulated in amplitude in accordance with the signal itself i.e. the height of the transmitted pulses vary with the amplitude of message.

Fig. 1

SAMPLED OUTPUT

Fig. 2

An information signal sent through an ideal switch which is operated by a control signal, isolated from the information signal, produces a PAM signal. When the switch is open, the

voltage is zero; when switch is closed the output voltage is equal to the instantaneous signal voltage. The sample width depends upon how long a switch remains closed. In practice, electronic switching is used. Sampling by this method is same as multiplying the information signal by a rectangular pulse train. Let the information signal m(t) of highest frequency component Fm (known as the base band signal) is applied to a multiplier along with a train of pulses of unit amplitude , width dt and period Ts . The output of multiplier is S(t)m(t) .

Fig. 3 As it can be seen from the above figure the multiplier output has some value as m(t) when the pulse occurs , otherwise it is zero . Rectangular waveforms can represented as summation of Sine/cosine waveforms of fundamental frequency plus infinite no. of harmonics. Since PAM is amplitude modulation of pulse, we expect the sidebands to be formed around fundamental frequency and each harmonics. If the sampling frequency is Fs, the frequency spectrum of the PAM signal is as shown in fig.4

PAM Frequency Spectrum

Fig. 4

As it can be seen from the figure, the spectrum of PAM signal is very much similar to that of AM signal, except the following 1. The PAM signal contains the spectrum of the base band signal unlike that in and where it is absent it is due to this fact we can recover the original signal. 2. In AM, a fixed amplitude carrier component is also present at the unmodulated frequency Fc. In PAM no such component exists in PAM spectrum. The information signal can be recovered from the PAM signal by using a low pass filter of cut-off frequency Fm.

NYQUIST CRITERION :

As shown in fig.5 the lowest sampling frequency that can be used without the sidebands overlapping is twice the highest frequency component present in the information signal. If we reduce this sampling frequency even further, the sidebands and the information signal will overlap and we can not recover the information signal simply by low pass filtering. This phenomenon is known as fold-over distortion or alising.

Nyquist Criterion ( Sampling Theorem)

Fig. 5

The Nyquist criterion states that a continuous signal band limited to Fm Hz can be completely represented by and reconstructed from the samples taken at a rate greater than or equal to 2Fm samples/second.

This minimum sampling frequency is called as NYQUIST RATE i.e. for faithful reproduction of information signal Fs > 2Fm. EFFECT OF DUTY CYCLE ON INFORMATION RECOVERY The duty cycle of a signal is defined as the ratio of pulse duration to the pulse repetition period . This ratio can also be expressed as %, e.g. the square wave has equal pulse and no pulse duration, and hence its duty cycle is 0.5 or 50 %. The duty cycle of the sampling pulses is an important parameter in PAM system. They govern the following important aspects: a) The narrower pulses allow us to time division multiplex many such PAM signals i.e. we can send many no. of PAM signals over same channel at a time . Hence lower duty cycle beneficial in this respect. b) The narrower pulses has wider frequency spectrum. Hence the wider bandwidth channel is required. c) Narrower pulses have less power as the power content of a pulse depends on its amplitude and width. During transmission and demodulation the inherent noise can play a major havoc on the low power signal. Hence a pulse of larger duty cycle is desirous for this sake. In practice an engineering compromise is made between narrower and broader pulse width taking into account the efficiency, requirement and inherent noise of the system. EFFECT OF SAMPLE AND SAMPLE / HOLD OUTPUTS If pulse width of the carrier pulse train used in natural sampling is made very short compared to the pulse period, the natural PAM is referred to as instantaneous PAM. As it has been discussed, shorter pulse is desirous for allowing many signals to be included in TDM format but the pulse can be highly corrupted by noise due to lesser signal power.

One way to maintain reasonable pulse energy is to hold the sample value until the next sample is taken. This technique is formed as sample-and-hold technique. The sample-andhold waveform looks as shown in fig. 6. Now, the area under the curve ( which is equivalent to the signal power ) is greater and so the filter output amplitude and quality of reproduced signal is improved . The hold facility can be provided by a capacitor when the switch connects the capacitor to PAM output it charges to the instantaneous value

Sample & Hold Waveform .

Fig. 6

A buffered sample and hold circuit consists of unit gain buffers preceding and succeeding the charging capacitor. The high input impedance of preceding buffer prevents the loading of the message source and also ensures that the capacitor charges by a constant rate irrespective of the source impedance see fig.7.

Sample hold circuit

Fig.7

The high input impedance of the succeeding buffers prevents the charge from the capacitor due to loading and hence the capacitor can hold the charge for infinite time, at least theoretically. However, small leakage current through the capacitor dielectric into +ve input of second buffer is always present which causes gradual charge loss. The rate of change of voltage with respect to time dv/dt is called as droop rate and is important parameter in sample and hold circuit design. ALIASING If the signal is sampled at a rate lower than stated by Nyquist criterion, then there is an overlap between the information signal and the sidebands of the harmonics. Thus the higher and the lower frequency components get mixed and cause unwanted signals to appear at the demodulator output. This phenomenon is turned as aliasing or fold over distortion. The various reasons for aliasing and its prevention are as described. A) Aliasing due to Under-Sampling If the signal is sampled at rate lower than 2fm then it causes aliasing. Let us assume a sinusoidal waveform of frequency fin which is being sampled at rate fs <2fm.

The low pass filter at demodulator effectively joins the sample causing an unwanted frequency component to appear at the output . This unwanted component has frequency equal to ( Fs-Fm ) . B) Aliasing due to wide band signal The system is designed to take samples at frequency slightly greater than that stated by Nyquist rate. If higher frequencies are ever present in the information signal or it is affected by high frequency noise then the aliasing will occur. This does not generally happen properly designed telephone network where speech channels are band-limited by filters before sampling. In control engineering and telemetry, however, out of band high frequencies either from source or due to noise pick-up can be present. In this case band-limiting filters, generally known as anti-aliasing filters are usually installed prior to sampling to prevent aliasing. As a principle, the system is designed to sample at rate higher than the rate to take into account the equipment tolerances, ageing and filter response. C) Aliasing due to filter roll-off Roll-off is a term applied to the cut-off gradient of a filter. No filter is ideal and therefore frequencies above the nominal cut-off frequency may still have significant amplitudes at a filters output. If proper sampling rate and approximate filter response is not chosen, aliasing will occur. D) Aliasing due to noise If very small duty cycle is used in sample-and-hold circuit aliasing may occur if the signal has been affected by noise. High frequency noises generally mix with the high frequency component to be present at the output.

RECONSTRUCTION The PAM system the message is recovered by a low pass filter. The type of filter used is very important, as the signal above the cut-off frequency would affect the recovered signal if they are not attenuated sufficiently. The simplest type of the filter is a resistance-capacitance ( RC ) filter . The pass filter and low pass RC filters are as shown in fig.8. The analysis of this filters becomes easier if we think of them as a.c. potential dividers. The reactance of the capacitor is frequency dependent with a high value at low frequencies and a low value at high frequencies .

Passive High Pass Filter

Passive Low Pass Filter

fig. 8

PROCEDURE : PART 1 Keep Sampling frequency constant and Vary the duty cycle from 0 to 90% 1) Connect the input of 1 KHz to the sampling circuit on the kit. 2) Keep the sampling frequency 80 KHz constant and vary the duty cycle from 0% to 90% and follow steps 3 to 6 for each value of duty cycle in step of 10%. 3) Observe the output waveform of the sampling amplifier. 4) Now connect the output terminal of the sampling amplifier to the input terminal of the second order low pass filter and the fourth order low pass filter and observe the output waveform. 5) Observe the output waveform of the sample and hold amplifier. 6) Now connect the output terminal of the sample and hold amplifier to the input terminal of the second order low pass filter and the fourth order low pass filter and observe the output waveform .

PART 2 - Keep the duty cycle at a constant value of 50% and vary the sampling frequency . 1) Connect the input of 1 KHz to the sampling circuit on the kit . 2) Keep the duty cycle at a constant value of 50% and vary the sampling frequency from 20 KHz to 320 KHz and follow the steps 3 to 6 for each value of sampling frequency .

3) Observe the output waveform of the sampling amplifier. 4) Now connect the output terminal of the sampling amplifier to the input terminal of the second order low pass filter and the fourth order low pass filter and observe the output waveform . 5) Observe the output waveform of the sample and hold amplifier. 6) Now connect the output terminal of the sample and hold amplifier to the input terminal of the second order low pass filter and the fourth order low pass filter and observe the output waveform .