I remember being about 12 years old and reading an article in my mother’s Reader’s Digest, in which readers submitted their “pet peeves” – things that, although insignificant, really bugged them. The “winner” of the article was a woman complaining about different cars having the gas cap on different sides.
One particularly annoying audio bugaboo for me is the fact that the terms “polarity” and “phase” are so often used interchangeably when they represent two markedly different concepts. This is not an attempt to single you out, Dear Reader, it’s an offense that has been committed by nearly every major gear manufacturer and software developer out there.
While a complete exploration of the two concepts would fill an entire book, let’s take a brief look at each concept, in a nutshell.
Polarity is binary: in or out, positive or negative, up-going or down-going, hot or cold (that Katy Perry song is actually about polarity, not phase).
When a positive pressure wave hits our microphone diaphragm, it creates a positive voltage at the mic’s output, if the mic is in polarity. Once this signal gets to the amplifier, it creates a positive-going voltage and current, which moves the loudspeaker cone towards the audience (positive direction.) If a polarity reversal occurs anywhere in the chain, the loudspeaker cone would instead move inward (away from the audience) and would be out of polarity with respect to the signal.
If we place a snare top mic and a snare bottom mic, hit the drum once, and zoom way in on the recorded waveform in our DAW, we will see mirror images: most likely, the bottom mic would have a positive-going transient (because the drum head first vibrates towards the mic) while the top mic would have a negative-going transient (because the drum head first vibrates away from the mic.)
Trying to combine these two signals at the mixer would result in a cancellation, because one signal is trying to move the loudspeaker cone out, and the other in, which results in degraded performance.
Imagine you and a friend on opposite sides of a door, both pushing on it. You’re not going to get too much done. For these two signals to properly sum at the loudspeaker, we need to invert polarity on one of them. (That’s that little Ø button.) This literally flips over the waveform, turning all the negative voltages positive, and vice versa. It’s the same as switching the hot and cold pins on an XLR cable. In fact, that’s how a polarity inverter adapter works.
Phase, on the other hand, refers to a point in a wave’s cycle.
Since we’re dealing with complete, cyclical oscillations, phase is measured in degrees, with a full cycle being 360 degrees. But the cycle speed depends on the wave’s frequency, which is a function of time. So while we can simply invert an entire signal’s polarity, we cannot “invert phase” on a signal – there’s no such thing unless the signal is a pure sine wave of a single frequency.
Why? Because to cause two signals to be 180 degrees apart, we need to introduce the proper amount of delay: half a cycle. And the amount of time which is “half a cycle” is different for every frequency.
For example, 100 Hz takes 10 milliseconds (ms) to complete a cycle. If we combine a 100 Hz wave with a 5 ms delayed copy of itself, it will completely cancel, because they are 180 degrees out of phase, and the peaks of one wave negate the valleys of the other. However, the same delay introduced into a 200 Hz signal would cause a 6 dB boost, because 200 Hz goes through a complete 360-degree cycle, so the phase angle between the two signals is now one of summation, not cancellation. They’re one “wavelength” apart.
The nitty gritty part is that all of the frequencies in between 100 and 200 Hz will have completed various portions of a full cycle in the 5 ms period, and so will have various degrees of phase offset. So there is no “phase” operation we can do that will affect all frequencies the same.
Time After Time
Let’s look at just one application from the real world: A bass amp that is miked and also running direct (via a DI). The signal from the DI is in the electrical domain the entire time, arriving at the console practically instantaneously. However, the mic signal will be a bit later because it needs to be amplified, created by the amp’s speaker, and then it needs to propagate through the air to the mic’s diaphragm where it’s turned back into a voltage.
This takes time. Not a lot of time, but some.
If the mic is a foot from the cabinet, this is approximately 1 ms of delay (sound travels at about 1,130 feet/sec depending on temperature and altitude). This means that the two signals, if summed at the console, would create a comb filter effect, where certain frequencies are in phase and certain others have a phase offset. The resulting pattern of alternating cancellations and summations is called a comb filter because it looks like the teeth of a comb on a linear frequency analyzer.
At 1 ms of time offset, for example, 1 kHz will have gone through 360 degrees and therefore will be in phase. At 500 Hz, it will have gone half a cycle (180 degrees) so there will be some cancellation. The important thing to remember is that intermediate frequencies exhibit various degrees of phase shift. Phase offset can be anywhere from 0 degrees up to 360 degrees (or more, if you want to count multiple full cycles), and this offset is different for each frequency contained in the signal.
A common way to “fix” the mic/DI issue is to flip polarity on one of the signals and find out which one “sounds better.” This doesn’t fix the comb filter, it only moves around the peaks and nulls (things that were previously 180 degrees will now sum, and perfect summations are now turned into 180-degree cancellations).
Rather, the correct fix is to delay the DI signal by the amount of time it takes for the mic’s signal to propagate (1 ms). This is easily accomplished on most digital consoles, and as a result, the two signals will be completely in phase at every frequency, whereas using a polarity inversion just moves the phase cancellations around.
Let’s say you’re exceptionally experienced with miking amps and knew this already. Why should you care? Everyone knows what you mean when you say “flip the phase,” right?
After discussing this point with some members on a popular pro audio forum, the argument that gave me pause was that most engineers do know the difference but it’s just habit to use the terms interchangeably, as we “all know what they mean to say.” Reflecting on that argument, I figure that it may very well be the case a large portion of the time, which at least allows me to rest easy.
That said, I don’t think it’s a justification. Why is “that’s how it’s always been done” a reason to keep doing something? We would never move forward.
However, I stand firmly (but perhaps foolishly) against the winds of convention for a different reason: Education. In any situation where our terminology might be viewed or interpreted by individuals trying to learn our craft, whether by training on a mixer that we manufacture, using one of our software plugins, or reading a technical document we author, it’s in the best interest of the audio profession as a whole to use technically accurate terminology.
While it may be true that pros know the difference between polarity and phase and automatically “select” the proper concept in context, how is a person who is learning supposed to know what’s being referred to? Polarity and phase are probably two of the most important concepts in the whole field of audio engineering and, for this reason, it only makes sense to properly refer to them, particularly when we’re around interns/work study students, volunteers, etc.
One day I was in their position, and some day they may be in mine. And in the words of that one guy from that one movie: “You never know who might be listening.” Or was it a John Lennon song? Something about listening to the children?
Sure. Let’s go with that.