Adaptive 60 Hz Noise Cancellation

Weekly Updates

Weekly dispatches concerning the design and implementation of our system.

We wrote a software simulation and GUI in python for our noise cancellation algorithm. While it worked very well at 44KHz, the adaptive component is not quite tweaked to our liking yet, and it struggles with fewer samples.
We designed all of our hardware, ordered parts, went shopping ourselves, and started building the circuitry for our microphone input. We are using a four-pole lowpass Butterworth filter, which is a cascade of two LM358 opamps.
We soldered and tested the ECE476 Prototype Board.

We wired the audio bass boost amplifier to power our subwoofer on a protoboard and tested it. Once we verified that it worked, we soldered it on a small solder board for longevity.
We wired a 4-pole, lowpass Butterworth filter and tested it. In the design, we didn't realize we needed a +/- 9V power supply on the opamp. So far, we have one on the protoboard and one that is already soldered to a solder board. We need to "make" one more and put it on a larger solder board.
We wrote code in gcc to blink an LED on the ECE476 Prototype Board and also to accept an analog input. We also got the microcontroller to program through using gcc and a "broken" STK500 board that Bruce says we can keep. Our next step is to MUX the analog microphone input from two low-passed speakers.
We also tested the microphones that came. So far, we don't have a good feel for them---the output and +Vs are both on the same pin, and the output signal is very small. Further testing will ensure that we can use these. We also need to take out the DC coupling by using a small RC highpass filter at the output of the low-pass filter before sending this to the microcontroller.
We would like to examine the eigenvalues of some of our batch sampling of audio information to determine the approximate order of mu. According to some papers, mu should be .

We got crunching with our code this week. We have tested mini programs that can mux analog inputs and we implemented a simple FIR filter that can run in realtime with over 250 poles (with 980 samples/second).
We determined the worst case delay through our circuit. We learned that the lowpass filters we designed have an inherent delay on the order of 1.2 ms for one and 3.5 ms for both.
We have the mics working--the trouble we were having with them seems to have been a result of the audio amplifier we were running them through.
Our website is now "compiled" into HTML, so adding new pages and maintaining a menu structure should be straightforward.
We started working on a 60Hz noise cancelling circuit as an intermediate step in our project.

We fixed some circuits so that they don't take as many batteries and repowered the audio amplifier with a 4.3 V DC source from the wall. Now we have a better testing environment for the audio cancellation.
On Friday, we had a good bit of luck with an adaptive FIR filter that canceled 60 Hz noise. We are trying to make it a more general filter that will have a longer weight vector and a more adaptive filter design. We would also like to sample at a faster input frequency (use clock/8 instead of clock/64).
In order to create a signal to cancel, we created an adder circuit that took in two signals. We then canceled the output of their sum and fed the error back to the mega32. We are finalizing a design for this board and think that if we can get the adaptive signal to work, we can use this as a prototyping design. Bruce also plans to bring in an AC transformer for us so we can get a good reference signal.
We also worked some more on simulations in Matlab just to verify that our design would work. We still it will.