Acoustic Absorption Panels – Part III. Tests
28 August 2020
Now that the panels are assembled, it's time to put them to the test. We've established that our primary goal is to reduce the effects of first reflections, and so we start with marking these reflection points on the walls and ceiling. First reflection means short distance for the wave to travel, and short distance means plenty of energy left, and plenty of energy left means loud, clearly discernable delays in the room. In short, these are the waves to target, and so the panels should be hung accordingly. We also mark three spots on the ground, loosely representing three critical listening locations in the room:
1. the engineer's "sweet spot"—the location at which the listener completes an equilateral triangle with the left and right studio monitors;
2. the centre of the room, give or take, for an overall idea of how the room sounds; and
3. the back of the room, representing what any musicians and producers lounging in the back of the room can hear.
In the interest of not confusing the issue, we ensure all three locations are kept on-axis (i.e. equidistant between the left and right speakers). An off-axis microphone placement would pick up sound from one speaker before the other, needlessly adding an additional variable to the mix.
Room EQ Wizard, a powerful piece of free software, plays a series of sine waves out of the studio monitors and then records them back through a measurement microphone. The mic picks up the direct sound from the speakers as well as any "remnants" of the sine tones that are left bouncing around the room. This "sine sweep" is repeated for each of the three locations, and done first without the panels, and then with, for a total of six tests. Thanks to the simple A-B comparison nature of the tests, any shortcomings in the speakers, microphone, software, etc. can largely be ignored. This is because whatever limitations might be present, would stay constant throughout all tests, and therefore eventually cancel out. In other words, all we're interested in is the difference between bare walls and having absorption panels up. With all else equal, these tests should be a telltale sign as to the actual effectiveness of the panels themselves.
Audio Interface Calibration
We begin by calibrating our audio interface. As mentioned above, it's not strictly necessary to check for minor flaws in a component if that component is present in all the tests. That being said, this calibration is a good way to familiarize ourselves with the software's measuring and graphing capabilities, not to mention a good opportunity to check that the soundcard is not acting totally out of whack. We connect one of the interface's outputs to an input using a patch cable, generating an external loopback—in my case channel 6 (circled above). The program then sweeps through the spectrum at -6dBFS and works out how the soundcard altered the response. It graphs the sound pressure level (SPL)—a measure of loudness—as well as the phase, in case any phase shifting occurred.
I am happy to report my Focusrite Saffire Pro 26 passed the test with flying colours (well, no colours, I should say). Within the audible frequency spectrum (20Hz-20kHz), the soundcard introduced very little colouration: the response only rolled off by about a third of a decibel at the extremes, with wild phase swings occuring only outside of this range too. The interface's response is saved, and will be automatically subtracted from all subsequent tests. Calibrating the system by a third of a decibel around 20Hz and 20kHz won't exactly make much of a difference, but hey!
For these tests we use a special type of microphone known as a measurement microphone. Actually, special is not the word, considering its sole purpose is to be as unspecial as possible. Measurement microphones—such as the Apex 220 I used (pictured above)—are omni-directional with a near-flat frequency response. That means they pick up sound with equal gain from all directions, and don't add colouration to the sound for the hell of it. Their blandness makes them terrible for recording music, but ideal for testing. Finally, as with any condenser microphone, we boost our input gain with +48V of phantom power.
For each microphone position, the interface's output gain needs to be increased such that the microphone picks up pink noise at around 70dB(C). The dB(C) filter is similar to the widely used dB(A) filter, except that it uses a flatter frequency response to judge loudness—a feature that makes it more suitable for acoustic tests than the dB(A) filter, which would compensate for perceived loudness at the frequencies the human ear is most sensitive to.
Once the loudness is calibrated, let the sweeping begin! A handy timer feature in Room EQ Wizard delays the sweep by a few seconds, offering just enough time to run off and duck out of the speakers' way. The entire frequency spectrum is then shot out into the room as a series of sine waves, with frequencies gradually increasing from DC (0Hz), all the way up to 22,050Hz (well past what is audible for humans, but conveniently half of the sample rate). Using the software's default 256k samples and a 44.1kHz sample rate, each full sweep takes about 5.8 seconds (256k samples divided by 44.1k samples/sec).
The moment we've been waiting for! I decided the efficiency of the sound absorption panels can be best gauged using three metrics:
1. Waterfall plots: the presence of each frequency in the room over time (measured up to 300ms)
2. Clarity C50 plots: the sound energy ratio (in dB or as a percentage) between the 'early part' (first 50ms) and the 'late part' (post-50ms) of a sound
3. RT60 plots: the time it takes a sound to decay by 60dB
One slight annoyance I came across when conducting these tests was the construction noise next door. I hadn't realized the neighbours were planning on making a racket with power tools all afternoon until the project was already well underway. Since each sweep only lasted a few seconds, I eventually managed to complete the tests during quiet intervals. What I found curious, though, was how little all those boisterous sounds nextdoor actually affected the results! Despite being very audible in the room, all that whirring and rumbling barely put a dent in the graphs when compared to the change to the room due to the panels. (see brown Clarity and RT60 noise plots below).
Success! And what a striking difference too! The waterfall plots show that without any absorption panels (left column), most frequencies across the 20Hz-1kHz range remain in the room. In contrast, just about all of the frequencies are tamed when the panels are hung up (right column). Furthermore, at all three locations in the room, the 250-500Hz region dies out after only about 150-200ms.
The clarity C50 and RT60 plots, on the other hand, show just how much of an effect absorption can have in comparison to microphone placement, or extraneous noise (as mentioned earlier with the construction). Though both plots are a little chaotic below 250Hz, it's what happens above 250Hz that really matters. After all, absorption of this kind can only be expected to target the low-mid and mid-frequencies—anything lower would require bass traps and larger, bulkier forms of acoustic treatment.
- An average boost of 15-20% (or 8-10dB) in clarity above 250Hz...
- RT60 decay time halved between 500Hz and 2kHz...
- nothing below 1kHz left bouncing around after 300ms...
Can't complain :)