DSPP PROCESSOR

The  hardware based demo experiment,  we were very excited as it was demonstrated by a senior who knew the dsp processor quite well.basic Instructions required for Arithmetic, Logical and Shift Operations were demonstrated. The changes in the values of the registers before and after execution of each operation were observed. Code Composer Studio was used to compose the code.

DSPP APPLICATION

A group experiment to find on technical paper and one patent in signal processing on a 1D signal . The application that we selected was 'Noise Reduction Using Adaptive Filters'. The group members are Kapil Rawal, Prerana Sarode,Chinmay Upadhyae,vaishnav dandge and Harshit Shukla.

Summary

A new adptive algorithm named affine projection is used rather than traditional methods for noise cancelation like LMS and NLMS ,the output is compared on the basis of signal-to-noise  ratio.The result showed that affine projection algorithm gave high signal to noise ratio then LMS and NLMS.

window function

Writting the code on C was not that tough but implementing the same in scilab was not a cake walk.But finally it was implemented .In this experiment,we used Hanning window for LPF and HPF filter Design ,by taking Pass band Attenuation (Ap), Stop band Attenuation (As), Pass band Frequency (Fp) in Hz, Stop band Frequency (Fs) in Hz, Sampling Frequency in Hz as an input. we plotted magnitude spectrum and phase spectrum and verified the value of Ap and As in pass band and stop band from the magnitude spectrum.

winspec

FIR Filter Design Using Frequency Sampling

The coding for this was done in SCILAB.The digital filter is designed  using frequency sampling method. For LPF and HPF filter Design we have taken Pass band Attenuation (Ap), Stop band Attenuation (As), Pass band Frequency (Fp) in Hz, Stop band Frequency (Fs) in Hz,  Sampling Frequency in Hz as an input.  I have plotted Magnitude Spectrum and Phase Spectrum and verified the value of Ap and As in pass band and stop band from the spectrum. From magnitude response it is clear that number of lobes increases with increase in order.To obtain the Magnitude and Phase plots. h[n] values of the signal and the magnitude and phase plot was obtained using SCILAB.


fir-fsm

DESIGN OF CHEBYSHEV FILTER

This experiment was same as the last execpt in mathematical equations , The  design was implemented for Analog Chebyshev filter using BLT method.Taking Pass band Attenuation (Ap), Stop band Attenuation (As), Pass band Frequency (Fp) in Hz, Stop band Frequency (Fs) in Hz, Sampling Frequency in Hz as an input. magnitude spectrum was plotted and verified the value of Ap and As in pass band and stop band from the spectrum.


chebyshev

DESIGN OF BUTTERWORTH FILTER

The new thing with this experiment was that we used scilab situations came were problem occured with synatx but final we found out solutions to it. We implemented Butterworth filter using attenuation and frequency in stop and pass band and sampling frequency.By using blt method we got the normalised filter for it and denormalised according to the filter .Thus filter was designed after taking its ZT. From the graph we concluded that there are no ripples found in either stop or pass band and with increase in order the slope becomes sharper.



BUTTERWORTH

FILTERING OF LONG DATA SEQUENCE

We were able to implement this quiet well as compared to last three experiments. In this Implementaion of Long Input Sequence using Overlap Add Method and overlap save method. length of input signal and input signal values as an input are taken from user.Taking length of impulse response and signal values of impulse response as an input from user. Here long input sequence is broken down into small sequences of equal length, then using zero padding and FFT corresponding X[K] is obtained. Then all these X[K] are added to get final output. Overlap add method is used to process real-time signal which have no defined end. overlap save method is used to evaluate discrete convolution between a very long signal X(n) and a finite impulse response h(n).

OAM 

OSM

FFT

 In this experiment  FFT of 8 point signal was performed .Problem occured while implementing the logic but that was rectified later.Here the length of input signal and signal values were taken from user. X[K] using FFT and x(n) by IFFT was calculated. In FFT calculations takes place in parallel manner. Also there are less number of complex additions and multiplication, hence the number of calculations is drastically reduced. Therefore FFT is much faster than DFT.

FFT-1 
FFT-2
 IFFT-1
 IFFT-2

DFT

In this experiment  DFT of 4 and 8 point signal was taken. We have developed basic functions for taking input and displaying .So for each new experiment we just have to implement logic as a new fuction only.The DFT was calculated and effect of zero padding on magnitude spectrum was observed. length of input signal and signal values from user. calculation of X[K] by DFT and x(n) by IDFT and  magnitude spectrum of the signal was plotted.Change in 4 point signal by appending 4 zeros at end.As the number of points increases, the resolution of magnitude spectrum increases.

DFT
IDFT

DISCRETE CONVOLUTION AND CORRELATION

In this experiment some prerequisite were neede of C .we used mathematical tools for linear convolution , circular convolution and linear convolution using circular convolution functions and then implemented it. The main thing was the implementation of logic that has to be applied for writing code, after developing the logic code was executed and the obtained results were verified with the results obtained from mathematical formulation.In linear convolution if both the signals are causal then the output result is also causal,circular convolution gives aliased output.

Correlation is mathematical tool used to find similarity between two signal.The output signal is in the form of pallindrome, when signals are same this can be used to find erroe in a signal .

CIRCULAR
LINEAR
CORRELATION