Noisy Data in 10-20 kHz frequency band

[ankit]

Hi

I am using a NI 4462 board for measuring SPL and doing Octave analysis. The problem is that I get a noisy SPL values from 10-20kHz band which are of the amplitude of about 10-14 dB. I know for sure that this 10-14 dB is not the actual measurement or the electrical noise where the microphone is placed. These measurements were done in a hemi-anechoic chamber. I am pretty sure that there is some signal processing aspect I am missing which is leading me to this loud noise floor.

I will hghly appreciate if someone can help me in trying to figure out this

Thanks

Ankit

[Sandra_T]

Hello ankit,

It looks like you are doing a sound pressure level measurement and conducting octave analysis in the 10-20kHz band with the NI 4462 board. Take a look at this DevZone article which can give you some ideas to reduce the noise floor.

Dithering, Layout, and High-Quality Components : Tools to Decrease the Noise Floor

If you are worried about field wire and noise in analog signals:

Field Wiring and Noise Considerations for Analog Signals


Other suggestions include:

1) Increasing the sampling rate. What is the sampling rate that you have set?

2) GND your analog input and see what the noise floor is, this will give you a more accurate measurement of what the noise floor is.


Regards,
  Sandra T.

Applications Engineer | National Instruments

[ankit]

Sandra:

Thanks. I am using a sampling rate of 51200 samples/sec. The frequency range of Octave analysis is set to 100-20,000 Hz.  An impedance source was connected to the BNC cable and it brings down the SPL values on all the frequency bands to zero and there is some electrical noise 10-20 kHz of the amplitude of 10-14 dB. I am trying to understand the GND concept involved in the data acquisition. I will highly appreciate if you can explain this (AI GND) by a simple vi.

Regards,

Ankit