Microphone CalibrationKnowing the sound pressure that is being delivered to the animal is an important measurement that allows comparisons between different animals, and also in comparing results, such as noise-induced hearing loss or age-related hearing loss, with results from other labs. While calibrating the sound pressure delivered from the speaker systems is not always trivial, but it is also not complex. Routines now built into ACQ4 can manage the calibrations, and provide a quick way to check the microphone, calibrate the speaker and check the speaker. Different calibrations are used depending on the speaker selected. The calibrations are done at spot frequencies using tones, and the software uses a spline interpolation to determine the necessary attenuation to achieve a specified sound pressure level. The calibration takes place in 3 steps, and takes about 10-15 minutes to complete. Calibrations should be checked at least once a month, if not every time the system is used.
About the sound pressure calibrationsIt is standard procedure to use a 1 kHz tone at 94 dB SPL as the reference. 94 dB SPL corresponds to 1 Pascal. The microphone sensitivity is referenced to this value, and for each microphone (with a serial number), the calibration is given on a sheet that comes with the microphone. In our case:
ACO Pacific 1/2" model 7012 microphone (SN 39279): 16.22 mV/pA, -35.8 dB re 1V/PaACO Pacific 1/4" model 7016 microphone (SN 9945): 3.55 mV/pA, -49.0 dB re 1V/Pa
The 1/2” microphone has a fairly flat response curve (+/- 1dB) up to 50 kHz. The 1/4” microphone should be fairly flat up to more than 80 kHz.
Care and Feeding of the MicrophonesThese microphones are laboratory instruments. The microphone elements should always be stored in their appropriate cases between uses, and the cases should be stored in a dessicator (this is a “to do” for us!). The microphone system consists of the microphone itself, the preamplifier (which, for the 1/4” microphone, includes the adaptor), a connector, and cable to the main preamplifier. The system should be disassembled after each use and placed in it’s foam compartments. The preamplifier section (approximately 1/2” dia silver tube, about 4” long) has a cap that should be carefully screwed on over the connector. Treat the microphones with resepct! They are expensive. Never force any of the connections, cross thread when assembling, or pull on the cord.
DO NOT DISASSEMBLE the microphone itself! This could expose the element to dust, dirt and oil, which will throw the calibration off and may ruin the element.
Step 1: Check the microphone sensitivityThe first step is to verify the microphone sensitivity against a standard; we use the ACO Pacific 511E standard at 1 kHz (actually about 995 Hz) set at 94dB, with a “fresh” battery. This standard was checked against a Larsen-Davis 800B reference sound pressure meter in May, 2010, and was found to be within 0.5 dB. We calculate a microphone “sensitivity” based on the voltage out from the microphone preamp (set to 20dB gain); this will be value in dB re 1V/Pa. The sensitivity/transfer factor may be calculated at this page: http://www.sengpielaudio.com/calculator-transferfactor.htm. Now any 1 kHz signal can be directly compared and it’s signal calculated. DiffdB = 20 * log10 (Vsignal/Vreference@94). Then dB SPL(signal) = 94 - DiffdB. Once we have the microphone sensitivity, we can reference all measurements to the microphone output voltage, under the assumption that the microphone is flat as indicated by the reference curves provided by the manufacturer. Remember, this only holds over the range in which the reference curve indicates that the microphone has a constant sensitivity. The program presents 1-second tones through the speaker (possibly with some additional attenuation, depending on the speaker output), and measures the voltage waveform. A “cosinor” function (basically, a single-frequency FFT) is used then to determine the mean voltage amplitude. From this, the sound pressure is calculated. The data are stored in a Matlab ”.mat” format file, but with the extension ”.cal”. [The file contains a structure “CAL”, which is loaded into the program to retrieve the calibration.]
Select the microphone to be checked in the drop-down list. Make sure that the microphone is on, plugged into the left channel of the preamplifier, that the preamplifier gain is set to 20dB (switch on back), and that all cables are connected (the output of the preamp goes to IN2 on the RP2.1). The microphone is then inserted into the coupler in the calibration source, and the source set to emit 94dB SPL. Make sure that the microphone is firmly set all the way into the coupler, and that the coupler is firmly set all the way into the calibrator unit.
Hit the button “Microphone Calibration...”. After a few seconds, there will be a printout in the matlab command window that will tell you the results of the measurement. The numbers should agree within a few dB of the sensitivity values originally provided for the microphones. Since the microphones are a few years old, their sensitivity may have dropped a few dB and this is normal. If the sensitivity is more than 4 dB from the values listed above, recheck all connections and make sure that the voltage out of the microphone shows a 1 kHz sine wave.
Notes- The Etymotic ER7C that we have seems to have an incorrect calibration; I need to look into this (the 94dB reference source seems to be way off). It was used to calibrate the ABR system in 2007, and that calibration was held constant until 4/30/2010. There is therefore a significant (and obvious) error in the calculation of the stimulus level in the ABR program for that period. This error is 30 dB (the signal was 30 dB lower than what we calculated). Makes sense to me.
- The signal from the ACO microphone and amplifier is somewhat noisy (spikes) and has a lot of low frequency rumble. We need to process it digitally to get an accurate reading. The cosinor analysis seems to very effective in accurately measuring the amplitude of the periodic signal, while not being perturbed by noise or low-frequency rumble.
发表于 2018-9-17 20:13:08