Measurement Microphone Comparison

Audio and acoustics guru Ethan Winer has a lovely measurement mic comparison on his RealTraps website. Here are some of my own tests as an expansion of that idea. The mics used for these tests are the ones I happened to have in my kit at the time: two iSEMcon EMX-7150 (not sold as a matched pair) and three Behringer ECM8000. Variations below 50 Hz are lower-confidence due to the small nearfield monitor used as a noise source rolling off at LF (lower SNR in the bottom octave).

Part 1 – Shared Acoustic Space Tests

In this type of test, two microphones are placed nose-to-nose, facing each other, in front of a loudspeaker, and a transfer function is taken between them. In other words, the signals from the two mics are used as the two inputs for a TF measurement, and so any deviation from flat phase / unity magnitude indicates a difference between the mics.

This test method seems to be preferred by many because it largely eliminates spatial variables (hold that thought) since both capsules are inhabiting the same “acoustic space.” However, it can exaggerate differences between microphones because we are comparing their 90° off-axis responses, which tend to vary more than the on-axis responses due to a number of factors.

EMX-7150 vs EMX-7150, without calibration curves (pink) and with calibration curves (blue)

From here I used one of the 7150’s with cal file as the basis for comparison to the Behringer mics.

Clearly the ECM8000’s have higher sensitivity, by 3 to 6 dB depending on the unit.

With level normalization:

Part 2 – Substitution Tests

In a substitution test, the mics are placed in an environment with a loudspeaker source, one at a time, with efforts taken to achieve identical placement of each microphone. For these types of tests, I use a flat table with markings to help achieve consistent capsule placement with about 1/8″ or so. The transfer function shows the mic’s response vs the original reference signal. The response includes boundary effects, acoustic / spatial effects, and the response of the loudspeaker, so it’s far from flat. Rather, of interest in a substitution test is the difference between the traces. There is a higher error here than the shared acoustic space test but the results are still informative.

Two EMX-7150’s, with and without correction files (see upper right for legend):

All mics, substitution test, unnormalized (see upper right for legend):

All mics, substitution test, level normalized (see upper right for legend):

Part 3 – Some Perspective

This last data came from a live demo I did at the Albany Area Audio Human Meetup last week.

Here we see the response of a loudspeaker in a room, measured from about 10 feet away.

Pink: Measurement with no mic correction curve

Red: Same placement, with mic correction curve

Blue: No correction curve, mic moved one foot away.