By Robert Carpenter
I’ve heard it from a lot of people over the years, in some form or other, that how you’re aiming your microphones makes a big difference to how you’re going to tune the system. Whether it be someone claiming that they heard someone new to the game was angling their mics all wrong and “ruining” systems left and right, a tech being very particular about their angles to the point of getting out a digital level to check (yes, I have actually seen that), or someone asking about the difference between Free Field and Diffuse Field microphone calibration files. Everyone seems to know that if you don’t pay attention to the angles you use, the results are going to be all wrong, right?
The Rig
For this test, I had an Isemcon EMX-7150, a Beyerdynamic MM1, a Genelec 8030B, and an Audient EVO 8, plus my trusty copy of Smaart Suite (9.1.3.0 when I captured the data, 9.2.9050 while I’m displaying it to write this.) Microphones were set at the same height and distance to the speaker for all measurements, regardless of the angle being used, though with a tape measure so some variance is to be expected in high frequency phase response due to minor variations in timing. Distance to the speaker was held constant at 18” (0.46m). Height from the floor was kept to a fairly consistent 3’6” (1.07m) as that is the height that put the capsule between the HF and LF drivers of the Genelec in use on my desk.
The Data
We’ve got two sets of data to display here, one from a Free Field microphone (the 7150), and one from a Diffuse Field microphone (the MM1). Conventional wisdom would suggest that the 7150 should be aimed at the sound source, and the mm1 should be aimed upward instead to avoid high frequency buildup it’s not designed to correct for. As a result, we should expect to see very different high frequency response, according to some sources as much as 10dB of buildup!
This graph shows the MM1 at a variety of angles. The blue trace is 0° (straight at the speaker) the teal trace is 45°, the yellow/orange trace is roughly 60°, and the maroon trace is 90° (straight up). So, what do we see? The traces are essentially identical in frequency response until we hit around 6kHz, where we start to get a little spreading, but not really more than about 3dB until we’re past 16kHz, where the difference grows to perhaps 5dB by 20kHz. We also observe that the 90° trace is really the only one drifting from the pack here. The phase trace tells a different story, with the 0° and 90° traces riding together, and the 45° and 60° forging their own trail. By 16kHz we’re roughly 190° out of phase, so fully canceling compared to… what exactly? We still have data there, so we’re not canceling the test signal by being at the wrong angle, but measuring compared to another mic at another angle at the same spot we’d get a different result for the phase offset. I’m not especially bothered by that, since at 16k a half wavelength is 0.42” (1.07 cm), which is only slightly larger than the microphone capsule itself. Based on the distances involved, it seems that minor changes in position of the capsule are going to have as much or more impact than variations in angle. This also presents a question of the precision of placement for my own measurements, and whether small variation could be the cause of the phase discrepancy entirely.
This graph shows the 7150 measurements, with 0° in green, 45° in the pinkish-red, and 90° in the gold/brown. The frequency response here tells essentially the same story as above, everything sticks close together until we get to really almost 14kHz this time, and doesn’t drift more than about ±2dB at all even to 20kHz. The phase trace here is a bit worse off than the MM1, but for similar reasons I’m not really bothered by that. For phase, the important thing is that I’m getting consistent results mic to mic not that I’m getting 100% accurate results to what is happening in the room. If all my mics are “wrong” the same way, then I’m not going to make wrong decisions even if the data isn’t “right”.
One more graph, this time all the traces from both graphs shown together. Same colors as before. As we can see here, all the frequencies are tracking fairly close to each other, even between the two microphones. So, from a frequency perspective the angle really doesn’t seem to make more than a couple of dBs of difference at the extremes of high frequencies. This is a common kind of finding when comparing measurement microphones generally, so it’s not overly shocking to find the same holds here as well. We can see looking at the phase responses that the only measurement that has a truly severe separation from the group is the 90° data from the 7150. It’s the foreground trace here to highlight that its coherence is also not the greatest in exactly the region of concern here. Since I took these measurements a while ago as of this writing, I can’t speak to exactly what may have caused that, but suspect that it’s further from the “ideal” measurement position the timing was set at than the other measurements were, leading to the drop in coherence and the phase shift observed. It wasn’t a major concern when the data was captured because I was, at the time, focused more on the frequency response side of the equation than the phase response.
Conclusions
So, from what we’ve seen the frequency response only really varies between the various angles by a couple of decibels at most at the very high end, and that the phase remains tight overall as well. Is that enough to make a serious difference in the system tuning we end up with? I suppose that depends on your definition of “serious”, but let’s step back and get some context here. I’m working from a context of system tuning, not artistic mixing. Any changes are being applied equally to the entire mix, and to all channels, not to specific vocal or other levels within the mix. The commonly cited Just Noticeable Difference in audio amplitude is 1dB for trained listeners, and 2-3dB for untrained listeners (Mills, 1960) . This is actually perhaps even smaller, as low as 0.5dB, for high frequencies. What that means is that even in an environment where we can expect discerning listeners, the differences involved here are such that they may not be detectable to the human ear in an A/B/X test as actually being different from one another. As a result, I think it is safe to say that no system tuning is being “ruined” by such differences.
Bibliography
Mills, A. W. (1960). Lateralization of High‐Frequency Tones. The Journal of the Acoustical Society of America, 32(1), 132–134. https://doi.org/10.1121/1.1907864
Audio Nerd Book Club 