Digital Signal Processing

This lab was basically a continuation to the subject Signals and Systems which we studied in the 5 th semester. The DSP lab brought in more clarity to Signals and Systems as we simulated Convolution, Correlation, Fourier Transform and FIR Filters which was studied in the previous semester. It also helped me understand the significance of using the DSP board over the microcontrollers we’ve been using. The DSP board used was the TMS320 and used the Code Composer Studio (CCS) software. The various sessions kept to understand the immense computation power DSP has fascinated me. The simulations done brought in more clarity to the Digital Communication subject that we’ve been studying. The sensor interfacing made me realize that the board has various applications and can be used in projects. Brief knowledge about flash memory was also attained which showed how the board can work as a standalone system.   Overall the lab was wholesome and quite useful!

In this experiment, we studied the various sensors interfaced to the DSP board. The analog signals thus detected by the sensors are converted to digital signals using ADC on the board and thus we get the desired response. The various sensors interfaced on the DSP board are Distance sensor, LDR, Hall sensor, Audio sensor, Temperature sensor and Optocoupler.   Distance sensor measures the proximity of an object to the sensor, LDR measures the light intensity, Hall sensor senses the presence of a magnetic object, Audio sensor is used to capture the voice, Temperature sensor detects how hot or cold the surrounding is, Optocoupler interconnects two separate electrical circuits by means of a light sensitive optical interface.

The demonstration of this experiment was first provided to us in our classroom where the importance of using DSP boards was emphasized. We were taught how with the knowledge of the internal hardware, we can configure the board according to our needs. In the lab we first performed DAC where we saw various signals such as Ramp signal, Triangular signal and Sine wave. Thus, the Analog signal can be observed on the DSO. We also observed two sinusoidal signals with different phase values. In the DSP board used in our lab i.e. TMS320f28335, we have one external 12-bit DAC and two 8-bit  ADCs.

In this lab, we were first asked to implement a 4-bit binary counter followed by a 3-bit counter. We implemented the codes for the same on the DSP board and saw the LED’s glow from 0-15 as in case of 4-bit counter and 0-7 in case of the 3-bit counter. After this, we were asked to read up about the flash memory. We learnt how a code can be loaded into the DSP board’s flash memory using the GPIO pins 84-87. These GPIOs are pulled high by resistors R3, R4, R5 and R14 to code 1111 thus jumping to Flash memory. To load code in flash memory, Texas instruments itself provides the code to execute flash programming. Now once the desired code is loaded into flash memory, we saw how it is not needed for the DSP board and JTAG to be connected to the PC, but a mere power supply from the adapter is all that it takes for the code to function. This thus eliminates the need for a PC to execute a code and makes the DSP board a standalone system.

In this experiment, we learnt about linear FIR filtering methods i.e. OAM and OSM. These are two algorithms that are used to process real-time signals. This is used over Fast Fourier Transform (FFT) because in case of FFT, the entire signal needs to be available at the same time in order to receive the output which isn’t feasible in real-time signals. Here’s where OAM and OSM come into play. In the lab, we implemented the required codes on the DSP board and saw how real-time signals are processed. Overlap-Add Method (OAM) Here the input sequence is x(n) and an M point sequence h(n) is considered. The long input signal x(n) is further decomposed into L point sequences. With the formula N = L + M -1, where N is assumed to be a radix 2 value, we get the value of L. After decomposing we perform Linear convolution of each decomposed signal and add all of them to get the final output signal. Overlap Save Method (OSM) In this method, again we decompose the signal in the same way ...

Fourier Transform converts a signal from time domain to frequency domain. Fourier transform is extensively used in signal processing. In this experiment, we performed Discrete Fourier Transform (DFT) and Fast Fourier Transform (FFT). DFT In DFT discrete time signal is converted to discrete frequency representation.   The signal converted to frequency domain can be converted back to time domain using Inverse Discrete Fourier Transform (IDFT). However, DFT is a slow process and works well when the signal length is not too large. FFT Again, in FFT, the signal is converted to frequency domain, however, this method outranks DFT. This is simply because in this method the input signal is decomposed into smaller signals thus enabling faster computation. Also in this method parallel computation takes place thus making it a faster process.  

Correlation is the measure of similarity between two signals. In this experiment, we performed auto and cross-correlation. In auto-correlation, the two input signals are the same whereas, in case of cross-correlation, we compare two different signals. Next, we calculated Carl Pearson’s coefficient of correlation. This coefficient gives us the degree of similarity between two signals.

This was the first experiment we performed in the DSP lab. First, we got ourselves accustomed to the DSP Board, learnt how it is better than microcontrollers for signal processing. The software to be used was CCS. We then created our project. The next task was to run a code. We used the convolution code as our initial codes and observed the results. Convolution is basically combining two signals to form a third signal. There are two types of convolution, linear and circular. In case of linear convolution, the size of the output is equal to L+M-1, where L and M are the lengths of the two input signals respectively. In Circular convolution, the size of the output is equal to the size of the largest input signal.  

In this experiment, we simulated various techniques on CCS version 4. To get a hang of simulation, we first simulated a normal sine wave. Next, we simulated a Harmonic wave of frequency 60 Hz. We then simulated the various modulation techniques i.e. BPSK, QPSK and DPSK. Also, we simulated a few line coding techniques i.e. Manchester Coding, Duo-Binary Coding, Modified Duo- Binary coding and Duo-binary coding with precoder. This experiment was very helpful as it helped it us visualize and understand things better.