EE 5345/7345
Medical Signal Analysis
Instructor: Carlos Davila
Dept. of Electrical Engineering,
Southern Methodist University
Module 1: Sampling
Theory and A/D Converters
Homework Problems:
- Suppose
a signal x(t) is band-limited to 100 rad/sec.
What is the minimum sampling frequency in samples/sec which will enable
the signal to be recovered from its samples, x[n]? Sketch a block diagram
showing how x(t) would be recovered from x[n].
- Consier
a 4-bit successive approximation A/D converter. The full scale voltage is
Vref = 8V and Vin = 5.5. Find the contents of the
successive approximation register at each step of the algorithm.
- Sketch
the quantization characteristic for the A/D converter having 3 bits.
- Explain
why a flash A/D converter is faster than a successive approximation A/D
converter?
- What
is quantization error? How is it modeled?
- What
is the power spectral density? What is the relationship between signal
power and power spectral density?
- Explain
why an oversampled PCM A/D converter has lower quantization noise power in
the signal band than Nyquist-rate sampled PCM A/D converter?
- Derive
equation (11) in Aziz, Sorensen, and Van Der Spiegel and explain why the
Sigma-Delta converter has even lower quantization error power in the
signal band than oversampled PCM A/D converters.
Project:
In this project, we will investigate the characterstics of
pulse code modulation (PCM) and Sigma-Delta A/D converters. You can do this
using a programming language of your choice, though Matlab or Simulink is
recommended. If you have not worked with Simulink, you may want to visit The
Mathworks web page (www.mathworks.com)
where you can do several tutorials which will help you get started.
- Implement
a PCM A/D converter. Use a Gaussian white noise signal for the quantizer
input, x[n], having zero mean and a variance of 1. Compute the
quantization error, e[n] and estimate the probability density function
(pdf) of e[n] for 8, 16, 32, 64, and 128 quantization levels using a
histogram. Comment on how the pdf changes as the number of quantization levels
increases.
- Using
the results from part 1, experimentally determine the SNR for each of the
quantizers that were tested and comment on the extent to which the SNR
approaches the theoretical value discussed in lecture. Assume x[n] has
been sampled at the Nyquist rate.
- For a
128 level PCM A/D converter, experimentally find the SNR when the input
signal, x(t), is sampled at four times the
Nyquist rate. Here you will have to lowpass filter your Gaussian white
noise samples in order to model sampling at four times the Nyquist rate.
This can be done by designing a lowpass filter using the Matlab filter
design and analysis tool (fdatool). Compare the resulting SNR with that
which results from Nyquist rate sampling in part 2. Also, compare your
results with theory.
- Experimentally
find the SNR for a first-order Sigma-Delta Modulation A/D converter
assuming x(t) is sampled at four times the
Nyquist rate. Compare your results with theory.