The reason being that certain kinds of crossovers produced a phase shift that amounted to 180 degrees at the crossover frequency.
So connecting the drivers anti-polarity produced phase-coherent response through the critical crossover frequency.
I have often wondered about this... Perhaps should have Googled it...
Thank you for the education Richard.
Set up another speaker (correctly wired as a reference) and mic, and record both mic's in stereo, just look at it on a scope.
If these speakers are in the same room, you can play a high frequency tone through the system and walk between the speakers to hear if there is a zone of cancellation.
I did this plenty of times but I looked for phase cancellation at low freq!?? Typo?
In the situation you seem to be describing, phase may not be necessary to match at all, and an out of phase signal may actually be more correct than an in phase one. The reason being that the wavelength of sound traveling through the air is very short.. For instance a 1 kHz tone has a wavelength of just over a foot in room temperature air. That means the phase reverses every 6 inches or so. Unless your listener is going to be in a specific spot, being this picky about phase isn't really worth it.
From personal experience often the best performance out of a multiple speaker system installed in a building is obtained simply by a trained ear walking the site and reversing speakers where there seems to be a phasing issue which might be removed by flipping the speaker phase.
In large live sound systems such as those used in a concert, multiple speaker locations are used, and the signal to each speaker is actually intentionally delayed such that the signal is exactly in phase with the main speakers on or near the stage.
This is right but assuming it's all about single driver 110v line speakers, yes you may as well be sure they are all in phase.
The reason they sound shit (inaudible if it is a PA) is that the differing delays to each spot from multiple sound sources ie. multipathing.
Best, if possible to minimise the number of sources and make them louder, so less null zones.
The other one to fix inaudibles is to minimise hard surfaces and also somehow also minimise the resonances of each room.
I believe the OP is talking about a distributed sound system with a large number of tranformer coupled, ceiling mounted, speakers. It is important that they be in phase with each other, or there will be "suck-outs" as a listener passes between the acoustic overlap points between them. In my installation business, I used a Sencore/Terrasonde SoundPro/Audiotoolbox. These units have a phase checker that uses a special waveform (like what EricTheNorwegian mentioned) that is triangle on the positive side and sine on the negative, with a 1khz frequency. With these units, one just walked around the room, with the acoustic measurement microphone on a boom, and hold it in front of each speaker. The toolbox would give you the polarity, along with the certainty of correctness, instantly. You could do a room full of speakers in minutes. One can still find these units on auction sites for very little money.
Another thing about 70 or 100 volt distributed systems. They are often wired in a daisy chain fashion, so it is very difficult to electrically isolate one speaker without pulling it (What the OP is trying to avoid). Getting close with a microphone acoustically isolates the speaker, and is usually the easier way to check out a finished install.