EE2.LabA: Microphone Frequency Response
||2-4 members per group
||Pure Data (pd), v. 0.39 or later
||One week allocated.
Aims of the Experiment
- To determine the frequency response of a microphone under test, using a high-quality condenser microphone with a given frequency response as a reference. To observe the resonant properties of the recording space, and account for these in the microphone
- Mac Workstation
- Digidesign mbox2 USB audio interface
- AKG C414B (reference mic), Shure SM58 (test mic A), Behringer XM1800S (test mic B)
- Genelec 8030a (loudspeaker)
The experiment measures the frequency responses of two test mics using a high-quality condenser mic (AKG C414B) with a known response as a reference. A sinusoidal test signal will be synthesized and microphone responses to the sinusoidal signal measured
using Pure Data (pd): a real-time graphical programming environment or patcher language. A studio monitor with a nominally flat frequency response will be placed in the live room, and the reference and test mics will be mounted 1m in front of it. All
signals will be patched to/from the control room. Frequency responses will be measured over the approximate audible frequency range 40 Hz - 20 kHz, and all measurements will be calibrated against an estimated room response.
Figure 1: Studio floor plan
- Estimate the predominant room modes of the main studio below 200 Hz from the floor plan in fig. 1. The room is approximately 2.75m in height.
- Write down an expression to calculate the centre frequency of the third-octave frequency band above and below 1kHz. Write down the centre frequencies of all third-octave bands from 40Hz-20kHz. Why do you think third-octave bands are used for the kind of
frequency response measurements done in this lab?
Hardware and Software Setup
- Place the Genelec monitor on a stand roughly in the middle of the live room, and ensure that its fine settings are set to a flat frequency response. Make a note of its free-field frequency response, which can be found in the technical specification section
of the manual. Mount the reference AKG microphone approximately 1m away from the loudspeaker on a mic stand, and set the directivity pattern to cardioid. Connect the loudspeaker (using a XLR gender changer), whilst switched off, and reference mic, through
the mic inputs of the patch bay back to the control room.
- Complete the input and output connections between the patch bay and the USB interface.
- Now start pd, navigate to Pd-extended->Preferences->Audio Settings and ensure that audio I/O preferences are set to the mbox2 sound card. Load the measurement patch 'sin_generator.pd'. Ensure that the sinusoidal gain is set to a minimum, and then switch
the loudspeaker on. Adjust the gain of a 1 kHz sinusoid until a moderate listening level (without any perceivable distortion) can be heard in the live room. This should not be changed again as it could distort frequency response measurements. Now adjust the
reference microphone gain on the mixing desk, making sure that clipping does not occur.
Explanation of the test patch
Figure 2: Test patch for microphone response measurement
The LHS of the pd patch 'sin_generator.pd' (above) synthesizes the test signal that is fed into the loudspeaker. DSP I/O is turned on and off by mouse-clicking the
SOUND ON/OFF boxes. There is an option to output either a sinusoidal signal or a white noise signal, but we will be using only the sinusoidal generator. The gain of either can be controlled using a slider, and the sinusoidal frequency can be controlled
using the frequency_range selector and frequency slider. The RHS of the patch measures the RMS amplitude of the input signal. The
adc~ object performs an A/D conversion, based on the stereo output of the USB interface. The RMS amplitude of both channels can be measured using the object
env~ (output in dB). env~ is supplied with an additional parameter which specifies the number of samples over which to average the RMS amplitude. Keep this at 8192 samples (approximately 200 ms at a sampling rate of 44.1 kHz).
- Plot the known frequency response of the reference mic at 3-rd octave intervals between 40 Hz and 20 kHz (this should be on a calibration chart in the microphone box), scaled to 0dB at 1kHz.
- Now measure the response of the reference microphone over the same frequency range, and adjust the overall response so that a 0dB point occurs at 1kHz. Subtract this from the known response given in the technical specification. Is the result flat? Are you
able to observe any of the room modes calculated in the preparation?
- Move the reference mic 2 cm to the left and re-measure the frequency response. Account for any differences between the two measurements.
- For each of the two test microphones: mount the mic on a stand in exactly the same position the reference mic was in, and adjust the gain on the mixing desk until the response measured at 1kHz is 0dB. Now measure the frequency response of the test mic.
Calibrate this against the room response at this position to get the actual microphone response.
- Plot the frequency responses of the reference and test microphones, after accounting for room effects, all on the same axes, and comment on the individual characteristics/limitations of the microphones. (It will probably not come of any surprise that the
mics are in order of increasing price: Behringer XM1800S, Shure SM58, AKG C414B.)
- The frequency response of the monitor was not accounted for in these measurements as it was claimed to be nominally flat. Is this justified, and how could it affect the microphone response measurements?