For acousticians and those engaged in research and development, the ability to measure attributes of rooms, speaker systems, and audio circuits is obviously invaluable. While the need for this capability might be less pressing for the average engineer, producer, or studio owner, we may still benefit from having the tools to make some basic measurements.
FuzzMeasure 4 allows us to measure and compare basic attributes like frequency response, group delay, and harmonic distortion, as well as a few more esoteric metrics with which I am less familiar - nineteen parameters in all. Paired with a computer (running Mac OS 10.10 or newer), audio interface, and measurement mic, FuzzMeasure can generate beautiful, high-resolution graphs (including especially nice FFT waterfall plots) and facilitate comparative analysis of different systems.
Download, licensing, and installation of the software was simple, and the user interface was intuitive enough that I was up and running right away, without any manual-reading. The requirement of Mac OS Yosemite may discourage some non-early-adopters; I brought my own recent MacBook Air and a portable USB interface to The Bunker Studio in Brooklyn in order to run some tests.
Upon launch, the user may choose one of six templates (pre- populated with relevant measurement parameters). At any time, these templates can be supplemented to add additional measurements. When choosing a new parameter to display, the menu from which you select offers succinct definitions for each option. That's a nice touch - educational and helpful in interpreting results.
In order to put the software through its paces in both the acoustic and electronic domains, I ran acoustic sweeps of our control room (using a Gefell omnidirectional measurement mic at mix position) as well as electronic tests of a few preamps (no microphone needed).
For the room sweeps, I started with the Advanced Acoustics template, which contains an FFT Waterfall plot (a graphic illustration of frequency response as it decays over time), Envelope Time Curve (useful for locating reflections in the time domain), and Energy Decay Curve (determines reverberation time for individual octave bands).
I'm no professional acoustician, and interpreting some of the more specialized measurements in a meaningful way was a bit above my pay grade, but I was easily able to admire the textbook waterfall plot of our Rod Gervais-designed control room, confirming that it's notably free from time-domain problems and excessive resonance, even at low frequencies. A minuscule amount of HVAC rumble was easy to identify and ignore on the waterfall plots and reverberation time measurements.
Toggling between three sets of monitors showed predictable, but illustrative results. Adding frequency response to the measurements confirmed that our ATC SCM45A pair extends much lower than our Yamaha NS-10M speakers, which extend lower (and higher) than the Auratones. Waterfall plots revealed time-domain performance of all three speakers (in conjunction with the room) to be exceptional.
For the preamp tests, up on the block were the BAE 1084 (a faithful Neve reproduction), the Rupert Neve Portico 5012 [Tape Op #49], the API 512, and our custom in-house built Drip Electronics REDD.47 tube preamp [#104]. Bearing in mind that our tests were only moderately scientific, we observed a lot of interesting results.
I'd always felt that enabling the EQ on the 1084 changed the sound somewhat, even if none of the EQ points were engaged. This subjective impression was confirmed upon test. With EQ bypassed entirely, frequency response was quite flat to 30 Hz and within 3/4 dB to 20 kHz. Simply enabling the EQ - without engaging any of the frequency points - revealed a slight peak around 3 kHz and a response that dropped over half a dB from about 300 Hz down to its natural roll-off on the very bottom. Additionally, distortion measurements showed "EQ in" to produce increased odd harmonics below 1 kHz.
We were unsurprised to find that the REDD.47 had the highest percentage of second harmonic distortion, but intrigued that it exhibited among the lowest percentage of odd harmonics, especially in the critical mid-band, as well as the flattest, most extended frequency response of any preamp tested. Our older 5012 Portico exhibited a small but clearly visible 60 Hz power-supply ripple on the FFT waterfall. While not audible enough to become a nuisance, it was enough to skew the harmonic distortion measurements around 60 cycles. This had us questioning the stock wall-wart power-supply implementation.
Group Delay and Impulse Response measurements showed a few interesting differences among the different preamps in phase shift and overshoot/settling time. In our moderately-amateurish tests, the REDD.47 at its 34 dB setting was about 20° forward at 20 Hz, which was similar to the API's performance (and slightly better than the 1084's). The REDD.47's other two gain settings showed about 5° more phase shift at the bottom extreme. The RND Portico was our best performer here, showing only 17° forward at 20 Hz.
Bearing in mind that I'm underqualified to interpret these results absolutely, the 1084 appeared to have a longer settling time with the EQ engaged than without. The Portico had the most textbook-looking impulse response, settling in about 2 ms, while the 1084 and REDD.47 both took about 4 ms or slightly more to stabilize completely.
One of the most interesting revelations was the impact of the Portico's Silkswitch- a pronounced low- mid hump and a roll-off at both ends. This didn't sync with my expectations, betraying some expectation bias in prior subjective impressions. Harmonic distortion was roughly the same for both (perhaps the barest hint of more third harmonic at low frequencies, and fewer upper- order harmonics, with Silk engaged). Time-domain performance was near-identical at both settings.
Running these sorts of semi-controlled tests on gear we use regularly was kind of broadly enlightening. Although I'd be hard-pressed to say that I'd directly apply this knowledge consciously, increasing general background awareness of the studio's equipment could easily pay dividends in a variety of ways, including better management of expectation biases. Engineers with a deeper knowledge of signals and systems would undoubtedly find the electrical measurements described here more meaningful, even if they're not at the level of precision that a $25,000 dedicated analyzer provides.
In terms of acoustic measurements, I could imagine FuzzMeasure 4 being very handy for people trying to optimize DIY spaces or for guerrilla project studios on a budget. In our case, it mainly confirmed empirically that which we already knew subjectively - our control room is sonically really good. But for less-optimized spaces, the potential to gather data in advance of problem-solving is vast. Coupled with a bit of basic acoustics knowledge, the information gained from these tests could help a user target problem areas with greater focus and specificity. Recommended!
Personal license $100 direct, commercial $500; www.fuzzmeasure.com
Brad Allen Williams is at bradallenwilliams.com