Zero Sum Game
Let’s talk about signal distribution. Some mix engineers send a Left-Right (LR) board mix to the PA, whereas some prefer to send Left-Right-Sub (LRS) and still others prefer Left-Right-Sub-Fill (LRSF). Readers hoping for me to advocate for one practice as superior in this article will be disappointed, although those interested are directed to the examination of aux-fed subwoofers in Chapter 11 of my book. Instead, I would like to take a more critical look at what I believe is one of the most commonly overlooked questions in modern sound systems engineering: how the front fills are driven, and how that decision can affect their level relative to the rest of the system.
We can begin by examining a configuration where four console matrices are provided to drive the sound system: Left, Right, Sub and Fill. This is more or less standard fare for any situation in which the FOH engineer’s console is driving different PA systems from day to day.
Left and Right are simple enough - driven directly from the main stereo mix. The front fills and subs may be handled in a variety of ways, and let’s focus specifically on the Fill feed for a bit (don’t worry, I’ll explain why as we go). There are basically 4 prevailing approaches here. The Fill feed is either:
a mono sum of the main LR mix
a mono sum of the main LR mix, -6 dB
one side of the main LR mix (say, just Left)
a completely separate mix tailored for listeners closer to the stage (say, more vocals, less drums)
The interesting thing here (and by “interesting” I mean “potentially problematic if we’re not paying attention to it”) is that these four methods result in different signal levels being delivered to the sound system.
Adding It Up
Summing the main LR mix to mono produces a signal that’s approximately 6 dB higher in level than either of the contributing signals (Main L and Main R). Of course, L and R aren’t perfectly correlated, so we can’t expect a perfect 6 dB of summation, but in a practical sense, in the live sound reinforcement world, L and R tend to be “pretty darn correlated” because of the drawbacks of using wide stereo mixes in live venues. I tested a variety of various FOH engineers’ board mixes and found that summing Left and Right reliably produced a level increase of just about 6 dB, and at least 5 dB in all cases I tested. Almost certainly, the “money channels” - lead vocal, kick, snare, etc - are panned center in the mix and therefore those elements are emerging with a full 6 dB of summation when the LR mix is mono summed.
Thus, it’s reasonable to assume as a starting point that summing a LR stereo mix into a mono matrix will result in an approximately 6 dB level increase. This is the justification for case number 2, summing L and R into the front fill matrix at -6, which counteracts the effects of the summation, resulting in a signal level that matches the individual L and R signal levels.
Some FOH engineers who make use of various “stereo widening” processes in their mix wish to avoid any potentially undesirable side effects of summing those processes to mono, and so opt to drive their fills from a single side of the LR mix, say just L (case #3 above). In this case, we’re operating at “unity” again, since we’re not summing multiple signals together.
We can’t make any sweeping statements for case #4 - a separately mixed front fill feed - since we have no idea what the FOH engineer may want to do, and so the resulting level could be different in every case.
So What?
If you’re keeping score at home, this means that front fill drive levels can vary significantly from act to act, with a 6 dB difference between the two most common approaches (#1 and #2). Two reasonable questions at this point might be “So what?” and “Isn’t this also true for the subwoofer feed?”
Yes, it is, but the subwoofer feed is, in a sense, “self correcting,” because the mix engineer can hear the subwoofers during the show and can therefore make any desired level adjustments until the mix is balanced to their liking (or ask the systems engineer to do so), which is not true for the front fills in a typical situation. If the system was tuned and level balanced by sending the same (mono) generator feed to every zone in turn, which is common practice, that means the system will therefore achieve its desired level balance when the mixing console sends matched signal levels from its matrices as well. If the front fill feed for an act is derived of a LR summation with no attenuation (which for shorthand I refer to as “LR0”), the front fills are now 6 dB louder than they were intended to be when the system was tuned and pronounced balanced, and there’s a good chance the engineers located at FOH can’t hear it.
6 dB is considered the “upper bound” of acceptable level variance in many contexts, in which case we’ve lost the game right out of the gate, but there’s a more sinister side effect as well: since every 3 dB level increase halves allowable exposure time under NIOSH REL, the listeners covered by the front fills accumulate four times the amount of sound exposure as listeners farther back. While folks near mix position might enjoy an hour-long performance at, say, 100 dBA, the folks down at the barricade are now enduring a crushing 106 dBA, and have reached the same sound exposure dose after just 15 minutes.
Alternately, we could align our systems in such a way that we presume the front fills will be an LR sum, and set them to achieve the desired level when driven with the extra 6 dB. In this case, anyone driving sending either the LR sum attenuated by 6 dB (which I abbreviate as “LR-6”), or a single side of the mix, now has front fills that are 6 dB too low, which - given the proximity to subwoofers and stage bleed (acoustic drums, guitar amps, etc), is a significant enough deficiency to prompt complaints and people unable to hear the show as intended. Certainly not as concerning as the opposite case resulting in hearing damage, but both are situations we’d like to avoid. How?
Down In Front!
Of course we are responsible systems engineers and we walk the coverage during the show and adjust as necessary, but it might take us a song or two to get down there, and in many situations, front fill coverage can be extremely difficult to accurately evaluate during the show, as it requires wading through the most packed and enthusiastic section of the audience. I also like to ask the FOH engineer how they’re producing their FF feeds, when the situation allows.
6 dB is a lot of error right out of the gate, so we need a method to reduce it to acceptable levels very quickly - especially in festival situations with a lot of acts, a lot of consoles, a lot of showfiles, and a lot of mix engineers sending us LRSF feeds derived in different ways.
When I align a sound system, I inject my analyzer’s generator directly into the processor and route it at equal level to all zones. (Figure 1, below).
Figure 1 - Generator distribution to all system outputs
We start everyone off on a level playing field and use the output gain for each zone to balance the system as needed. (Astute readers will note the 18 dB of attenuation on the generator input, which I won’t be diving into here, but is discussed in Chapter 15 of my book.) There are a number of benefits to driving the DSP directly rather than routing the generator through a console, but most relevant to this discussion is that it avoids introducing level offsets as a result of the console’s Panning Law into the system alignment, as well as steers clear of all the potential routing complications and considerations inside the console.
Once we have the system tuned to our satisfaction, we have a very useful starting point: as long as all inputs to the system are at the same level, the level balance between the system zones should be preserved. Thus, when the first act of the festival kicks off and our DSP input meters look like this:
Figure 2 - Smooth Sailing
we can be reasonably certain that the front fills are in the ballpark of the correct output level since their input level is matched relative to the main LR signals. A FOH engineer sending us a Front Fill feed that is Left + Right with no attenuation (LR0) looks more like this on our DSP:
Figure 3 - Down in front!
A glance at the metering tells me that I can knock the FF feed back 6 dB either on the input gain or in the routing matrix and quickly get things back closer to where we want to be, with a coverage walk to fine tune.
Note that broadband signal level on the DSP input meter isn’t telling us everything, particularly in the case of the FOH engineer applying a high pass filter to their FF matrix before they sent it to our DSP - the signal level could look lower than Mix L and Mix R but still be the appropriate SPL for those listeners, so the method isn’t foolproof - but it’s extremely helpful when dealing with chaotic festivals for keeping things on the rails.