Telecommunication engineering

Pulse Modulation

BSc in Electronic and Electrical Engineering 1999-2000

When intelligence is transmitted at ultra high frequencies, power requirements of the transmitting equipment is an important consideration. One method of decreasing the power is to reduce the intelligence to small samples. As a result, only proportions of the intelligence are transmitted and the "pulse modulated" wave is off most of the time. A sufficient number of samples are required to allow the total intelligence to be reconstructed. Mathematics can prove that a signal sampled at twice its frequency component (known as the Nyquist rate) can be reconstructed at the receiver to a high degree of accuracy. A method of providing this type of transmission is classified as pulse modulation.

There are different methods used to pulse modulate a signal. Each method gets its name from the way samples or pulses are varied to represent the intelligence. Common types of pulse modulation are pulse-amplitude modulation, pulse-width modulation, and pulse-position modulation.

Pulse-amplitude modulation is the process in which an intelligence signal is sampled at regular intervals, and each sample is proportional to the amplitude of the intelligence at the time of sampling. The amplitude of the intelligence signal at the time of sampling (pulse high) represents the transmitted information. Transmission does not take place for the remainder of the sampling period. This On/Off feature allows transmitters to operate on a low duty cycle. It also lends itself to a technique called time division multiplexing TDM in that the time intervals between pulses can be filled with samples of other messages to transmit different intelligence over the same channel. Each PAM signal occupies its own time slot within each sampling period. The receiver must be synchronised with the transmitter to properly sort samples of PAM signals and send them to the appropriate demodulator or detector and to retain transmitted signal quality. A receiver uses a synchroniser circuit to lock in a sampling frequency with the sampling frequency used at the transmitter.

Pulse width modulation PWM is the process in which the intelligence signal is sampled at regular intervals and the pulse width is proportional to the amplitude of the intelligence at the time of the sampling. The pulses are equal in amplitude and each positive edge of the pulses is the same time apart. The negative edge of the pulses is not.

Pulse position modulation PPM is the process in which fixed width, fixed amplitude pulses vary their position due to the change in amplitude of the intelligence. One method used in the generation of a PPM signal is to use the negative edge of a PWM signal. The PPM pulses vary their position due to the varying pulse width of the PWM pulses.

Sampling and signal recognition.

Sampling and signal reconstruction are two important considerations when transmitting and receiving information using pulse modulation techniques. The ability to reconstruct the intelligence depends on how often the original intelligence signal at the transmitter is sampled and how sharp the frequency response of a low pass filter network is at the receiver.

No single sampling signal frequency or frequency response of a low pass filter network performs best to reconstruct the intelligence. It takes combination of both.

Selecting a sampling signal frequency is defined by a principle. (Nyquist) Sampling at a rate that is equal to or less than twice the maximum frequency of the intelligence signal should be avoided. Sampling at these rates will distort the reconstructed intelligence signal.

The generation of the three types of pulse modulation begins by identifying the intelligence signal with a sequence of pulses representing the intelligence at a particular time. A simple electronically controlled switch shows the concept used to sample the intelligence. A sampling signal operating at a low duty cycle controls the position of the switch.

The switch is closed for the duration of each pulse which allows the intelligence at that sampling time to become part of the output. The switch is open for the remainder of each sampling period making the output zero.

Another form of sampled signal that occurs in pulse modulation is called flap-top sampling. A first stage is used to sample and hold a level of intelligence. A capacitor charges to the level of intelligence when a switch is closed and holds that value when the switch opens and until the switch closes again to take a new sample. The result is a staircase signal. The staircase signal is then input to a second stage which is identical to the circuit used to naturally sample the intelligence (first example).

Signal reconstruction is the process of recovering intelligence from a sampled signal. In the receiver a low pass filter filters the sampled signal and outputs the reconstructed intelligence that is undistorted replica of the original intelligence. Two factors governing the ability to reconstruct the intelligence are the sampling signal frequency and frequency response of the low pass filter. The sampling signal frequency must be greater than twice the maximum frequency of the intelligence. The cut-off frequency of the low pass filter must be high enough to pass the maximum intelligence signal frequency but low enough to reject sidebands frequencies of the sampling signal frequency.

Aliasing occurs when sideband frequencies of the sampling signal normally outside of the intelligence signal frequency band overlap. The sampled signal does not approximate the original intelligence signal and the low pass filter reconstructs an entirely different signal.

Pulse-code modulation (PCM)

In digital modulation systems the intelligence signal is sampled. After sampling, the intelligence signal is digitally coded. This code represents the sampled amplitude of the intelligence signal. The binary coded signal is sent to the receiver in serial form and decoded to produce the original intelligence signal. A sufficient number of samples are required to allow the intelligence signal to be reconstructed accurately.

There are different methods of digital pulse modulation. Each method is named for the way samples are changed to represent the intelligence signal. A common type of digital pulse modulation is pulse code modulation PCM. Other types are differential pulse code modulation, delta modulation, and adaptive delta modulation.

The pulse code modulation process begins with an analogue intelligence signal being sampled at regulars intervals. The analogue value of the sampled signal is converted into a series of bits by means of binary coding. This process is accomplished by a PCM encoder circuit. The series of bits is transmitted over a single line to a receiver. At the receiver a PCM decoder circuit converts the series of bits into an approximate copy of the analogue intelligence signal.

The PCM encoder consists of a sample/hold circuit, an analogue to digital converter and a parallel to serial converter.

The quality of a PCM signal depends on the synchronisation of three separate processes - sampling, quantization and coding. It is the function of a PCM encoder circuit to provide each of these processes.

Quantization is the process of converting the voltage level of the sampled amplitude to the voltage value of the nearest standard level, or quantum. The total range that intelligence signal amplitude covers is divided by the ADC into a number of standards levels, called quantum. Coding is the process of representing a particular quantum of the analogue signal with a binary code.

The baud rate ( R ) is a unit of signalling speed equal to the number of bits transferred per second. In PCM, one baud is equal to one bit per second. The bit time Tb of the signal is the pulse width of one bit. The baud rate of the signal is equal to the inverse of the bit time.

The PCM signal bits that are transmitted to the receiver side of a PCM system must be converted into the original intelligence signal. A circuit that performs this conversion is the PCM decoder.

A PCM decoder consists of a serial to parallel converter, a register, a digital to analogue converter and a low pass filter.

Synchronisation between the serial to parallel converter and the register is necessary to ensure that the correct bit combination is loaded into the register. If synchronisation is not present, bits of one word may be loaded with bits from a different word. This would result in an incorrectly decoded signal. The amplitude difference between the original intelligence signal and the reconstructed signal is called quantization noise.

Quantization noise occurs since the reconstructed intelligence signal approximates the original. The fidelity of the reconstructed signal improves as the number of bits and quantum increase. Because of this, quantization noise is not noticeable in PCM systems that use converters of six or more.

Companding (compressing expanding) is the process of compressing intelligence signal amplitudes before and expanding the decoded signal amplitudes after reception into the reconstructed intelligence signal. This reduce the quantization noise.

Another method of companding uses non linear ADC in the encoder and non linear DAC in the decoder. The non linear ADC has small step sizes for low voltages and large step size for high voltages. This allows to the ADC to quantize and compress at the same time. The DAC performs the opposite function.

Pulse code modulation multiplexing

A PCM communication system can use three different modes of operation, simplex half duplex and full duplex. Simplex is a method of communication where one station can only transmit data and the other can only receive data.

Half duplex transmission is a method of communication where each station can transmit and receive data, but not at the same time.

Full duplex transmission is a method of communication where each station can transmit and receive data at the same time.

Multiplexing is the transmission of two or more messages over a single channel.

Time division multiplexing permits the transmission of two or more signals over a single line by using different time intervals for the transmission of the intelligence of each message signal. A frame is a group of bits.