How can a stereo pair of figure-of-eight microphones invert the phase of a signal?
The classic coincident crossed pair stereo microphone configuration consists of two figure-of-eight mics mounted as close together as possible and angled at 90 degrees to each other.
One mic points to the left of the sound stage, the other to the right. When the recording is played back over conventionally-positioned stereo speakers, you will hear images of sounds coming from all points between the speakers, not just from the speakers themselves.
So imagine you are recording a radio drama using such a crossed pair of microphones only. The actors can move around freely in front of the mics, giving the recording a sense of space and movement.
But they can only remain in front of the mics. Imagine lines taped out on the studio floor extending directly outwards from the diaphragms of the mics, both at the front and the back. The studio floor will have an enormous X laid out.
If the actors remain in the upper quadrant of the X, the stereo image will be good. In fact they can also perform in the lower quadrant of the X. When played back over speakers, they will sound exactly as though they were in front of the mics, since figure-of-eight microphones are bi-directional.
But if they go into either of the side quadrants, there is a problem.
An actor in the left quadrant will cause the diaphragm of the left mic to vibrate from the front. At the same time, he or she will cause the diaphragm of the right mic to vibrate from the rear.
This produces signals that are opposite in phase. As one rises in voltage, the other falls.
When reproduced on speakers, the cone of one speaker will push out while the other pulls in.
This in turn pushes against one ear drum and pulls on the other. There is nothing in nature that causes this effect, so the brain has no way of making sense of it and it sounds really odd.
Signals that are out of phase in this way are always to be avoided.