Meaningful Comparisons
By Greg Simmons
3 March 2020
In this issue I reviewed Røde’s TF-5 small diaphragm cardioid condenser microphone and made comparison recordings to share with readers. I put the TF-5 against two established favourites I’m familiar with: Neumann’s KM184 and DPA’s 4011. Between the three, I aimed to create a set of comparison recordings that allowed the listener to determine where the TF-5 sat on the price/performance scale. The process of making meaningful comparison recordings was harder than it seemed, with each method inviting uncertainties that rendered the comparisons meaningless.
RAISING THE BAR
For my review purposes the recordings only needed to highlight enough differences between the mics so that I could make informed opinions, but those opinions were also informed by a lot of background stuff: I made the recordings and knew what the instrument sounded like in the room, I’d experimented with different microphone and instrument positions, asked the musician to play specific things to highlight differences, exchanged opinions with others at the session, and even went back and did it again to confirm my initial impressions. All that background context is lost when the recordings are passed on to someone who wasn’t at either of the sessions. So if I was going to make comparison recordings for readers to form their own opinions, those recordings needed to be meaningful on their own. The standards of the recording quality, and the musician’s playing, suddenly got higher.
TRIANGULATION
How hard can it be to compare three microphones? Stick them side-by-side, match the gains and hit record, right? That’s okay if you’re miking from a distance, where a few centimetres between diaphragms will probably be irrelevant. But it’s not okay for a close-miking test. For the comparison recordings, I used acoustic guitar, and anyone who has miked an acoustic guitar from 30cm or less knows that moving the microphone a few centimetres can make a noticeable difference to the tonality of the captured sound. It’s the kind of difference that can render the comparisons meaningless: are you comparing microphones or miking positions? So the goal was to get the diaphragms as close to each other as possible. But how?
ONE MIC AT A TIME
Ideally all three diaphragms would be in exactly the same spot. That’s not physically possible without doing three separate takes, one microphone at a time, and that’s only going to provide meaningful results if the performer can give exactly the same performance from exactly the same position, every single time. Separate takes means separate performances, and that means it’s impossible to tell if tonal differences between recordings are due to the microphones or the performance. The comparison recordings are no longer meaningful.
THREE MICS AT ONCE
How about laying two of the microphones side-by-side, laying the third along the top of them like stacking logs, and taping them together? They all have tubular bodies of similar diameters, and when bundled this way each diaphragm would be as close as possible to each other diaphragm. There are two problems with this approach.
First, these are all end-address cardioids with ports behind the diaphragms. Bundling them together as described here means each microphone will have its ports obscured at two points (one point from each of the other microphones). Whether or not that will affect the tonality of any of the microphones is debatable, but as long as it is debatable the comparison recordings are meaningless.
Second, each of the three microphones has an outer diameter of around 20mm, so the closest any two diaphragms can get to each other is around 20mm centre to centre. That distance is unavoidable, of course, but could result in each microphone capturing a different tonality from the instrument if the relative orientation of the microphones is not considered. Which orientation provides the least tonal differences on the instrument being reviewed: vertical, horizontal or diagonal? Get it wrong and the comparison recordings become meaningless.
UNMATCHED PAIRS
I decided to focus on two microphones at a time: the TF-5 and the one I was comparing it against. I’m not comparing the KM184 against the 4011, so I don’t need them in the same test.
The next decision was placement. As any experienced engineer knows, the best spot for one microphone may not be the best spot for another. To avoid playing favourites, I used a consistent placement of 30cm from the guitar, perpendicular to the soundboard, with both microphones aimed at the point where the guitar’s neck joins the body – my preferred starting point when recording acoustic guitar. For the distant miking tests I pulled the microphones back while keeping them focused on the same spot.
For orientation, putting one microphone above the other (rather than side by side) yielded better tonal consistency during subtle performance movements of the guitar – the tonal differences between microphone positions was significantly smaller than the tonal differences between the microphones themselves, and therefore irrelevant. I left a small gap between the microphones to avoid any possibility that one microphone was blocking the ports of the other.
To ensure consistency between comparisons, the guitarist composed four short pieces to reveal different aspects of each microphone’s performance: long slow strums, fingerpicking, muted strums and full-bodied strums. Finally, I sent a blend of the two microphone signals to the headphones so the guitarist could ‘work’ the microphones; staying on-axis and subtly moving closer and further back as desired, just like any normal recording session.
MATCHING PERCEIVED VOLUMES
The final step in preparing the comparison files was to carefully match the perceived volumes of each pair of recordings. This was essential because the TF-5’s frequency response includes a shallow dip from 2kHz to 7kHz, falling to almost -1dB from 4kHz to 5kHz. This dip includes the range of frequencies that human hearing is most sensitive to, meaning the TF-5 will probably sound lower in perceived volume when compared to a microphone without such a dip. In contrast, the KM184 and 4011 both have a subtle boost in the upper part of that same range of frequencies. Any attempt at simply matching the peak levels or matching the waveforms consistently resulted in the TF-5 sounding dull and uninteresting in comparison to the others when, in fact, it was being reproduced at a slightly lower perceived volume. As any experienced engineer or hi-fi salesperson knows, if you want someone to prefer one sound over another make it louder. Without carefully matching the perceived volumes, the comparison recordings would always favour the louder of the two files and would therefore be meaningless.
Excellent summary of the troubles of A/B! One thought — since you have matched pairs of the mics you’re testing, why not throw in a few takes of “both the same mic”, so you can compare purely the difference in position. If you ran a blind A/B listening series with some of the mics the same and some different, you could determine whether the difference between mics or the difference between positions dominates what you’re hearing.