Phase Distortion - Speaker/Amplifiers

In speakers, it also refers to the ability of drivers to appropriately synchronize push/pull.

http://www.thielaudio.com/THIEL_Site05/Pages/FAQs/faqtimephase.html

Here is a layman explanation. It is full of speculation (much of which is doubtful) but the basic explanation is correct and there are some further links.

phase Distortion

IMHO phase distortion is way over rated.For sure there are extremely subtle audio effects but that is true of any change in listening space. One must remember that we hear frequencies NOT waveforms - changing phase can alter how the audio signal behaves overall as it interacts with the room - multiple reverberations will all cause phase shifts.

The key is to have gentle change in phase as a function of frequency. A marked 180 or 360 degree shift at a specific point in the spectrum will indeed be audible across the midrange as it will change the timbre of certain instruments very slightly.

The author's complaints about B&W are conflating two very different effects - by moving the crossover to 4KHz, B&W are dramatically altering the midrange into a narrow beam of sound with reduced room reflections (think a flashlight versus an ordinary lightbulb) and this effect will be far far more audible than any slight change in phase. This is an example that shows how easy it is to jump to false conclusions in audio due to the fact that many interrelated things are going on and what one hears differently after changing a setup can easily be something other than the presumed cause.

Although this subject is way over rated (much less important than many other terrible deficiencies in most loudspeakers), if you want the best then get an active speaker!

It appears that phase distortion is primarily the result of crossover networks in multi driver speakers which is why headphones and single driver speakers have less phase distortion.

On the surface it seems that a network that results in 180 degrees (2nd order?) of phase distortion could be corrected by simply by reversing the wires on the driver itself. It must not be that simple.

Any audible difference results from the individual drivers not all firing at exactly the same time.

In all cases the amplifier is "seeing" the cumlative effect of entire speaker so what does an amplifier do with a phase shift? It seems that if an amplifier is really just transmitting a series of + and - pulses it's already operating at extremes so why would phase shifting matter?

The phase shift is relative to the phase at the input.

When I think of the phase relating to speakers my understanding isn't much beyond swapping the speaker wires around such that the left and right channels would be firing opposite of each other.

technically the correct term for this would speaker polarity (& not phase). The negative on the speaker itself connects to the negative binding post on the amp & this terminal is really circuit ground of the amplifier output stage. So it's cable that allows current to be sunk by the amplifier (the cable connected to the red binding post allows current to be sourced into the speaker).

It appears that phase distortion is primarily the result of crossover networks in multi driver speakers which is why headphones and single driver speakers have less phase distortion.

yup! this is a very large contributor to phase distortion. You get phase distortion from mainly two parts - one, the x-over network (as you wrote) & two, from the mechanical properties of physical driver itself. The cone mass, spider network & how all of this is connected to the metal structure of the driver creates some driver impedance vs. freq & also driver sensitivity i.e. how good a certain driver is to reacting to low level signals & how faithfully it reproduces these low level signals. On the other side, how well is the pistonic action maintained for very strong signals before cone break-up. This will also generate phase distortion. You can see why esoteric material drivers get very expensive very quickly.

Re. headphones & single-driver speakers - the other thing in their favour is mass of the driver which is often lighter than the conventional cone driver. Lower mass implies higher probability of reacting to low level signals & the less the phase distortion.

On the surface it seems that a network that results in 180 degrees (2nd order?) of phase distortion ....

"phase shift" & *not* phase distortion. A 2nd order system has 180 deg of phase shift. Phase distortion, as the name implies, is a form of distortion i.e. a measure of imperfection & is measured in dB. Phase shift is amount of shift of phase vis-a-vis the input signal & is measured in degrees.

corrected by simply by reversing the wires on the driver itself. It must not be that simple.

yes, you'll correct the phase shift by reversing the wires but this correct will be valid at the x-over frequency only or a very narrow range of frequencies. it will not be valid over the entire operating freq range of the tweeter/mid/woofer because freq & phase responses of the x-over network & that of the physical driver are not constant/flat over that same range.

In all cases the amplifier is "seeing" the cumlative effect of entire speaker so what does an amplifier do with a phase shift?

The x-over network is the dominating factor to the amp as the physical driver is "hidden" behind the x-over. Once the signal gets past the x-over network it enters the physical driver in whatever (less molested or more molested) form. So, phase shift occurs once current is sourced or sunk into the speaker by the amp is response to the music signal.
If the phase excursions are radical (as in many of today's speakers) the amplifier becomes very ill-behaved. Suppose that the phase shift starts off positive (which means a phase lead, which in turn means that the speaker looks inductive/like a choke to the amp) then voltage leads current i.e. when voltage = max/min, current = min/max. If the phase shift changes over a very short freq range & becomes negative (which means a phase lag, which in turn means that the speaker looks capacitative to the amp) then voltage lags current i.e. when voltage = min/max, current = max/min.
So, you can see if the phase excursions are radical, the amp is having to switch the polarities of the voltage & current almost instantaneously - the music signal is jerking the amp around. How nice do you feel when you get jerked around? And, how much will you tolerate before you react negatively? Same deal with the amp. Now it's easy to see why an amplifier would go unstable with a speaker which presents wild phase excursions or an amplifier would have a hard time driving such a speaker. You also see comments such as "amplifier-speaker synergy is missing".

you'll also notice that on most amplifier boards there will be an inductor at the output stage very close to the point where the signals go off the board to the speaker binding posts. What's this inductor for? Remember the fundamentals of an inductor/choke? V=L*di/dt? Current does not change instantaneously in an inductor - the magnetic field in the inductor coils resists this change. So, this inductor is actually a protection for the amplifier electronics itself for the case when the user connects it to a speaker with wild phase excursions. When the amp is asked to change from phase lead/lag to lag/lead, the voltages & currents in the amplifier electronics change smoothly due to the fundamental properties of an inductor. If this inductor wasn't present, the amplifier electronics would burn themselves out due to the speaker load freq & phase responses.

It seems that if an amplifier is really just transmitting a series of + and - pulses it's already operating at extremes so why would phase shifting matter?

the amp is not putting out +/- pulses during a music signal. what do you think, is music a series of logic 0/logic 1 data pulses? No! it's a continuous analog signal & the amplifier, if doing its job correctly, is followinng that signal up/down on a real-time basis. Even class-D amps output an analog signal (even tho' some of the inner workings are trend more towards the discrete-time domain).

Hope this helps more than the 1st time. I've rambled enough now & am going to shut up....