Mini-Project-2 - Digital Oscillator Design - Oct 15, 2012 - 2 PDF
Mini-Project-2 - Digital Oscillator Design - Oct 15, 2012 - 2 PDF
Karthik Date issued: October 15, 2012 Final Report Due: November 12, 2012 (Monday, 12:15pm sharp) [-50% for late submissions] MINIP(2) problem statement
Before you peep at the problem statement, read the reference material from the Analog Integrated Electronics textbook. The current problem concerns mimicking a real feed-back oscillator. The system which you are expected to design must have the following structure:
w[n]
White noise, Variance =
B[ Z ]
[n] = Ax Sin( 0 n + x ) x
Synthesized sinusoid (Desired steady state response)
w2
A[ Z 1 ]
The filters, B[ Z 1 ] and A[ Z 1 ] are causal and FIR.
[n] = Ax Sin( 0 n + x ) is the desired response of this oscillator, where Ax and x are random x
variables (independent of the time index n). For any particular realization (sample function - SF) of the process w[n] , say wSF (1) [n] , the response of the oscillator will be,
SF 1 [n] = a1 Sin( 0 n + 1 ) x
In practice,
Task
1) Design the filters B[ Z 1 ] and A[ Z 1 ] so that the output of the system contains only a single frequency component (discrete sinusoid of a frequency of your choice). Keep in mind that the loop gain of this positive feedback system must satisfy, B e j A e j = 1 + j 0 at a single frequency 0 . For all other undesired frequencies ensure that your loop filter gain,
) (
B e j A e j
) (
< 1
2) Explain the operation of this system (why it behaves as an oscillator in not more than one page). 3) Evaluate you design (the following plots are expected) Magnitude and phase response of your loop-filter L Z 1 = B[ Z 1 ] A Z 1
DTFT of the oscillator output ~ x SF [n] (Use stem to highlight the region in the vicinity of the desired output frequency 0 ). Compare the observation ~ x SF [n] with a suitably generated reference signal x ref [n] (Example, Sin[ 0 n] ). Comment on the discrepancies (if any).
[ ]
[ ]
4) (Extensions) Now convert this filter into a tunable parametric digital oscillator such that by changing a single input parameter multiple frequencies can be synthesized.
30%
30%
3 Implicit assumptions proposed framework in the
10%
30%
Vary the noise variance and filter parameters and show the outputs. Run the simulation for several cycles and show both the steady state and transient responses. What factors decide the rate of convergence? Discuss stead y state behavior. Is this truly steady state?
--------------------------