After a day of troubleshooting on Sunday, the team decided to meet as many times this week as possible to conduct testing of the device.
Monday:
The team met at John's house again to continue where they had left off on Sunday.
This day went much more smoothly and they were able to conduct a preliminary round of testing with no fxlms and a full round of testing according to the Success Criteria.
The preliminary round was conducted as a kind of control against which the team could compare the role of the fxlms algorithm to mark any improvements it made.
For both tests, each frequency (50, 100, 150, 200, 300, Pink Noise) was played at 4 different volume levels.
The full test included measurements at 12in, 32in, 32in offset by 8in, and 12in offset by 8in.
Test recordings of a Lawnmower and AC unit were also tested with one trial at each distance.
Monday's Tests consisted of a total of 104 different Measurements.
Average attenuation across this dataset was -0.47dB (an increase of 0.47dB).
This result is due to the devices negative performance at 200 and 300Hz. 200Hz noise was increased greatly by the device. This is most likely due to resonance of the wall as the 200Hz tone was blocked much more than other tones.
The device performed best at 150Hz
Trials using test recordings of an AC Unit yielded no attenuation of noise.
An average attenuation of 2.1dB was measured for the test recording of the Lawnmower
Tuesday:
The team met on Tuesday to continue testing. This time 5 transducers are used spread across the surface of the wall.
These are wired to yield a final nominal impedance of 8ohms:
(16ohm||16ohm + 16ohm||16ohm)||16ohm = 8ohms
Test Trial Time-Lapse Video:
The results were greatly improved with the average attenuation of 2.075 dB over 104 data points.
This time the device attenuated for all sine signals but the 200hz tone.
Again, best performance was measured at 150 Hz
The device also performed well at 300Hz
Spectrum of Lawnmower through wall no ANC
Spectrum of Lawnmower through wall with ANC
Spectrum analysis of trial audio recordings of the lawnmower show a 1.7dB decrease of the 94Hz peak. However, magnitude of the upper harmonics is increased substantially.
After a full day of testing, temperature of the power amp heatsink was taken
Thursday:
The team met for a third time to conduct testing of the device on a residential window.
The same procedures for the previous wall tests were followed
Average attenuation across 104 data points was 2.67 dB. Negative attenuation (enhancing noise) was measured at only 9 data points and never exceeded 1 dB
In this application the device performed best at 50Hz with an average attenuation of 13.6 dB
An average attenuation of 2.1dB was also measured for the AC Unit noise source.
Spectrum Analysis shows a reduction of 6.1dB at the 33Hz Fundamental
AC Unit Window No ANC
AC Unit Window with ANC
John
During the initial testing of the device, John noticed a fair amount of harmonic resonance coming from the plastic transducer enclosure. In order to reduce this he constructed an mdf enclosure to seal the back side of the transducer.
Reducing the higher frequency resonance greatly improved the devices noise reduction capabilities.
After the first round of testing, John installed a DPDT switch to allow for on-the-fly phase inversion of the transducer terminals. Although phase inversion is possible using sigma studio, the sigma studio function block reduces the output level of the DSP. The physical switch makes comparisons between output polarities quicker and more accurate.
John also installed an extension cable for the reference mic in preparation for the window test.
John also finished installing the arduino and touchscreen housing onto the main chassis.
Hunter
The majority of the worked conducted this week was teamwork. Updates to the microcontroller and touchscreen program where also added, now that the ANC design was complete and ready for trial testing.
Below are two of tests to confirm that the programs button functionality for displaying calculated values is working when initiated and turns off when user presses the off button. The first video is a simple test to show the Vrms measurement of a AA battery being displayed on the TFT touchscreen, after the on button has been engaged.
Although this test for Vrms is technically invalid, since rms voltage is only possible to test with AC voltage, not DC. This "test" using a AA battery is simply to check that the calculations within the program that will be calculating Vrms to verify that the program is updating correctly to help prepare for the next meeting.
In preparation for the next team meeting, the analog input voltage needs to be tested that to verify it is measured accurately from the microcontroller's input. This is because the AC signals peak voltage is used to calculate the Vrms value, which is required for the additional calculations needed for measuring the SPL-dB that is coming from the MEMs microphone. The calculations will be explained in more detail in next weeks progress log, as the reference voltage also needs to be determined to calculate the SPL-dB using the microcontroller.
To prepare for the ANC test trials, the microcontroller program for measuring the SPL was tested by using an electret microphone with a built in amplifier. The video below shows a cellphone application for measuring SPL-dB to compare with the microcontrollers vaules viewed from the display and Arduino serial monitor.
An Excel document was also created in preparation for the test trials:
