Does anybody understand panning laws?

You may be sorry you asked...
 
Panning laws govern what happens when a monaural sound moves from left to right in the stereo field, which can be different for different pieces of software. In fact, not knowing about panning laws can create significant issues if you need to move a project from one host to another. Panning laws may even account for some of the online foolishness where people argue about one host sounding "punchier" or "wimpier" than another when they loaded the same project into different hosts.
 
Panning laws originated in the days of analog mixers. If there was a linear gain increase in one channel and a linear gain decrease in the other channel to change the stereo position, at the center position the sum of the two channels sounded louder than if the signal was panned full left or full right.
 
To compensate for this, it became common to use a logarithmic gain change response to drop the signal by -3 dB RMS at the center. You could do this by using dual pots for panning with log/antilog tapers, but as those could be hard to find, you could do pretty much the same thing by adding tapering resistors to standard linear potentiometers. Thus, even though signals were being added together from the left and right channels, the apparent level was the same when centered because they had equal power.
 
But this "law" was not a standard. Some engineers preferred to drop the center level a bit more, either because they liked the signal to seem louder as it moved out of the main center zone, or because signals that "clumped up" around the center tended to "monoize" the signal. So, dropping their levels a little further created more of an illusion of stereo. And some of the people using analog consoles had their own little secret tweaks to change the panning characteristics.
 
With virtual mixers we can create any panning characteristic we want. While that allows a high degree of flexibility, it also adds a degree of chaos that we don’t really need.
 
Sonar takes the whole process further than most DAWs, with six different panning options, which you can find by going Preferences > Audio > Driver Settings. In the descriptions below, "taper" refers to the curve of the gain and doesn't have too radical an effect on the sound. The six options are:
 

• 0dB center, sin/cos taper, constant power. The signal level stays at 0dB when centered, and increases by +3dB when panned left or right. Although this is the default, I don't recommend it because of the possibility of clipping if you pan a full-level signal off of center.
• 0dB center, square root taper, constant power. This is similar, but the gain change taper is different.
• -3dB center, sin/cos taper, constant power. The signal level stays at 0dB when panned right or left, but drops by -3dB in each channel when centered.
• -3dB center, square root taper, constant power. This is similar, but the gain change taper is different.
• -6dB center, linear taper. The signal level stays at 0dB when panned left or right, but drops by -6dB when centered. This is for those who like to hype up the sides a bit at the expense of the center.
• 0dB center, balance control. The signal level stays constant whether the signal is in the left channel, right channel, or set to the middle.

 
You can actually see the results of choosing different pan options by comparing meter readings for different pan law settings given a constant input signal. (Note that pan law is a global setting, and can't be set individually for each track.)
 
Regarding which law to choose, as laws go this particular one is pretty unspecific. In fact, if you compare DAWs, most default to a different law. Seriously.
 
This can become a real problem when you move a project from one host sequencer to another – unless the selected panning laws match. I often wonder if when some people say a particular host sounds "punchier" than another, the "punchy" one boosts the level when signals are panned hard left or right, while the "unpunchy" one uses the law that drops the level of the center instead.
 
For example, suppose you move a Sonar project to Cubase. It's going to sound softer, because Cubase drops the center channel to compensate, while Sonar raises the left and right channels to compensate. Conversely, if you move a Cubase project to Sonar, you might have to deal with distortion issues because signals panned hard left and hard right will now be louder.
 
But where these laws really come into play is with surround, because here you're talking about spatial changes among more than just two speakers. Bottom line: Be consistent in the panning law you use, and document it with the file if a project needs to be moved from one platform to another.
 
Personally, I go for the tried-and-true "-3dB down in the center" option. I designed analog mixers to have that response, and so I'm more than happy to continue the tradition within the various sequencer hosts. Also, this is one option that just about every host provides, whereas some of the more esoteric ones may not be supported by other hosts.
 
Finally, remember that the pan law you choose isn't just a matter of convenience or compatibility, although I've stressed the importance of being compatible if you want to move a project from one host to another. The law you choose can make a difference in the overall sound of a mix.
 
This is less of an issue if you use mostly stereo tracks, as panning in that case is really more of a balance control. But for many of us, "multitrack" still means recording at least some mono tracks. I tend to record a mono source (voice, guitar, bass) in mono, unless it's important to capture the room ambience – and even then, I'm more likely to capture the main sound in mono, and use a stereo pair of room mics (or stereo processing) that go to their own tracks. And if you pan that mono track, you're going to have to deal with the panning laws.