Stacking the Tracks
What happens when we use all of inputs of the Behringer X32 as a line-level digital interface?
Now that we’ve seen that the X32’s XLR preamp channels are very clean at 0dB gain, the next question is “what happens when we stack a whole bunch of them together?”
The answer is going to take a bit of dB math, and a bit of terminology. In audio, there are correlated, and uncorrelated signals. The concurrent sound of a trumpet and piano playing are considered to be uncorrelated sounds because their wave shapes and harmonics don’t naturally add together. The correlated signals of sine waves used for testing were exact clones and therefore will combine in a way that things in air played by humans will never do.
I fabricated a -108dBFS sine wave by processing a -90dB gain reduction on the -18dBFS reference signal. Then, I made 31 clones of that clip and summed them into a single bus. The correlated output of 32 signals at -108dB added up to a peak of -72dBFS (about 26dB SPL in my room) that was softly but clearly audible. When merely 8 of the clips were played together, the signal was right at -84dBFS. It was right at the threshold of what I could hear without turning off the mixer. This is the behavior of correlated signals strongly adding together. The math for uncorrelated signals says that a set of 8 signals at -108dB’s would add up to be -99dBFS. That sits at a physically impossible -1dB SPL as my monitors are calibrated, and is very encouraging for future mixes.
My monitors as currently calibrated at the listening position give me an 80dB SPL output from a mono signal at -18dBFS. With four of the -108dB clips playing simultaneously, I could not hear the combination, and the peak display in Voxengo SPAN says -90dB. After a few more trials, it was pretty consistent that turning on the 9th clip made me aware of the signal, confirming the -84dBFS threshold for my situation. It’s a really weird experience to work with signals that are too soft to show up on the -90dB range of the individual track meters, and to even have the the batch of 32 at full cry being too soft to show up on the bus meters’ -60dB scale!
Looking forward, this all means that when 16 tracks have uncorrelated (also known as incoherent) signals, the -108dBFS 3rd harmonic will be -96dBFS (2dB SPL as calibrated) and a full set of 32 channels of uncorrelated noise at -108dBFS will show as -93dBFS or 5dB SPL, still a full 9dB below my audibility threshold.
Remember, I’ve been harping (ha, ha) on the 3rd harmonic because it’s the strongest of the harmonics. A 3rd harmonic is also musically related; an octave plus a fifth above whatever note is played. The even harmonics, 2H, and 4H, are musically desirable while the other even ones 6H, 8H, 10H are generally so small, that they wind up being benign anyway. As a side note, if you’ve ever taken a good look at the drawbars of a Hammond organ, the 4’ is like a 2H, the 3H is like the 2-2/3’ drawbar, and the 4H is like the 2’ drawbar.
My particular X32 mixer has had some modifications. I bought it in March 2012, so I was well out of warranty when I discovered that the rising noise of the 24-volt fan would not be easily overcome. There are zillions of silent 12-volt 120mm fans from which I can choose, but 24-volt fans are rare and expensive. The particular part that Behringer uses in the X32 is manufactured in Japan, is rated for 100K hours of usage, and costs over $50 in single unit quantities. I opted to keep the fan and deal with the problem in the acoustic realm. I covered the internal slanted metal doghouse with a highly effective constrained layer product called GTMAT Quadro. It’s sort of a super version of the popular DynaMat sound deadening material used by car stereo enthusiasts. The mixer also sits atop a 1-inch compressed fiberglass pad that’s the full dimension of the mixer. Actually the feet of the mixer sit on the frame of the 'tray' and the gap between the bottom of the mixer and the fiberglass pad is about 1.5 inches. The combination makes the mixer almost a non-issue in its contribution to the noise floor, but it’s not completely silent.
In summary, the preamps in the Behringer X32, when run at 0dB gain, show a soft 3rd harmonic contribution that is best managed by keeping the input signals in the range of -18dBFS. At this level (which is also desirable for the purposes of post production with plug-ins) all of the best features of a line-level mixer are readily available with no cautions or compromises necessary. I am attempting to get equivalent test data from a Midas M32 in order to compare these results with the higher-priced version of the X32.
In upcoming installments, I’ll reveal the software settings and processing involved in this set of measurements, and quite a lot about how some popular preamps measure up in this environment.
EDIT: I've updated the pictures at the left side to show all dB values from 0 down to -120dBFS. This should also make the charts useful for other discussions. (TAG 20180821)
- Ted Gary of TedLand
May 30, 2016