The least satisfactory part of experimenting with microphones has been measuring their overall performance. Measuring the electronics is reasonably straight forward once the levels, impedances, and circuit operation are understood well enough. Acoustic measurements are something else. Turns out there’s a reason the big microphone manufacturers have elaborate and expensive anechoic chambers. Getting an accurate or even a repeatable measurement in a normal room with standing waves, reflections off of everything, and background noise is crazy.

So what have I been doing? There are measurements in some of my postings. With the caveat that my measurements don’t necessarily relate to any made elsewhere, I have been able with some difficulty to get repeatable frequency response measurements. That is, with a given setup on a given day, I can measure a mic, take it down and measure a second mic, then put the first mic back in place and get the same measurement as the first within a dB or two. At the low end below 40 or 50 Hz, things typically wander up and down due to room resonance and artifacts of the Fourier transform process. Above 10 KHz, there is often not enough level to overcome noise in the system, so there is often “grass” on the graph. Between 50 Hz and 10 KHz, typically the response is pretty solid and repeatable.

The first item is the speaker used to excite the mic under test. I’ve tried several, and the best are single full range drivers. For high frequencies, a small 35mm aluminum ball speaker from a “sunflower” iMac works pretty well, but is hard to get both the Mic Under Test and the reference mic aimed so both are on axis and equally illuminated by sound. For most purposes, a 5 inch full range speaker in a well damped enclosure made from a tea container works best. Working back toward the source, there is a small DC-coupled amplifier with flat response from 0 - 250KHz ± 0.1 dB which has been my general purpose shop amp since I built it 40 odd years ago.

I have also used a KRK Rokit 5 or Tannoy Reveal 502 amplified monitor system, but there can be issues around the crossover between woofer and tweeter.

A TASCAM US-144 mkII USB interface links the amp to a Mac Mini. The TASCAM is unmodified. Its noise and distortion performance is good enough that it doesn’t interfere with the measurements.

The Mac usually runs Faber ElectroAcoustic Toolbox software. Specific microphone test software hasn’t yet been written. For the Macintosh platform, the other measurement tool available is FuzzMeasure. The two are very different. FuzzMeasure is the easier of the two to get started with, but about equally difficult when complex measurements are made. It has one test mode. It generates a frequency sweep, a chirp, once. Then it calculates for a second and displays a graph of the result. You can select dozens of graphs from menus, ranging from basic frequency response to phase and harmonic distortion plots. Impulse response is also calculated. Sweeps can be short or long. The default is 1 second. Because the mic is muted shortly after the chirp, the theory is that reflections are eliminated. In practice, I’m not convinced that during the sweep, room response doesn’t have an effect. A second is certainly longer than it takes for my small shop’s first reflection. FuzzMeasure’s strong suit is its graphing capability. It plots beautiful graphs with lots of options such as smoothing and FFT artifact reduction. Its weakness is the single chirp measurement. There is no way to get repeated measurements of distortion while you adjust bias on an FET for example.

Faber ElectroAcoustic Toolbox is a collection of signal generators, FFT analyzers, X-Y scope, oscilloscope, level meters, etc. which can be bought one by one or as a whole suite. It’s a bigger initial step to get started with, especially if you’re not familiar with FFT terminology as most of us aren’t. FEAT works more like a traditional signal generator and oscilloscope / analyzer setup. The signal generator makes sine, square, triangle, and sawtooth waves, noise signals, and sine sweeps which can be modified by standard Mac audio units like graphic EQ if wanted. There isn’t much in the way of test signals this can’t generate. There are a bunch of displays like X-Y phase, multichannel oscilloscope, octave and 1/3 octave analyzer, spectrogram, meter bridge, FFT analyzer, and a dual FFT analyzer which is my favorite tool. Any of these tools can be used with any I/O that goes through the Mac’s audio unit system, so you can call up a special set of VU meters or octave level displays to use in ProTools or Logic for instance. You can run more than one instance of any tool. It’s a virtual rack of audio analyzers, displays, and signal generators. Much more accurate and powerful than the H-P analyzers I used to use. Where it’s weak is graphing results. The displays tend to look like an oscilloscope. Period. No smoothing or polish for publication. If you save a bunch of measurements, they can’t be plotted without compatible I/O hardware attached to the computer. For instance, measure a mic’s response to white noise at 88KHz sampling rate using a USB interface. If you take the data file to another Mac, you will need an external interface set to 88KHz to display the data, since a Mac’s internal A/D won’t run at 88. FEAT is designed and maintained by a scientist. It doesn’t make much effort toward ease of use.

The packages produce identical results when configured the same, as you’d hope would be the case. SuperMegaUltraGroovy and Faber appear to be one man companies. FuzzMeasure was acquired by Røde (microphones) in 2018 with promises of a bright future, but appears to have been abandoned. No support, no bug fixes, no updates. FEAT is also abandoned. Faber is devoting his time to iOS / M1 MacOS analysis tools. See:

https://www.faberacoustical.com

I cannot recommend either for purchase today, especially if you are running a system later than MacOS Mojave. Go with currently active software such as SignalScope.

This is getting a bit long, so more measurement in the next article.