# Sensitivity Testing Work

## Offline Tests 4/14/10

I attempted to align the short recordings (0.08s) made with the scope signal sent to the speaker with the appropriate region in the much longer MEMs bitstream captured from the FPGA (~10s). Trial 1 has clear success in the correlation sequence, whereas Trial is not clear at all. The codec signal also appears to have much less detail than the demodulated bitsream from MATLAB that has arbitrary precision.

 Trial 1 Trial 2

--EvanEttinger 18:19, 14 April 2010 (UTC): Another note is that I had to do some tricks to get the bitsream signal @ 2.4 Mhz down to 25.6 Khz which is what the scope records at. Since this is not a simple integer division, I first downscaled the bitstream to 32 Khz, then using interp I upscaled this signal to 128 Khz using a factor of 4, and downsampled again with downsample() to 25.6 Khz by a factor of 5. I don't think this makes much of a difference though.

## Initial Tests (Pre 4/14/10)

Below is the results of the original coherence tests between the MEMs microphone and two different Analog mics in the MAE lab. These test are done in an automatic noise cancellation (ANC) tube with the speaker playing white noise about a foot or two away from the microphones.

 Coherence

--EvanEttinger 20:25, 14 April 2010 (UTC) Here's the equation for the empirical coherence in the frequency domain (note you need i=1,...,k measurements to average over): $C(f) = \frac{|\sum_i X_i(f)Y_i(f)^*|^2}{\sum_i |X_i(f)X_i(f)^*| \sum_i |Y_i(f)Y_i(f)^*|}$