We have said it before : laboratory measurement of loudspeaker drivers are necessary for speaker system design. For this reason we developed a tutorial article explaining how old and modern acquisition systems work in the time and the frequency domain. We studied the central idea of acquiring and storing the IR of a loudspeaker driver in order to produce (and store) the SPL AND the phase response of this driver. These two responses are actually what we need to design our crossover network via simulation software.
What point did we miss ?
Apparently the microphone is supposed to record the properties of the speaker to be designed as these will be perceived by the listener's ears. That is why we prefer to place the microphone on tweeter axis and at a significant distance. Ideally we should take these measurements at 2m but this would enhance measurement noise and reflections.
-Which exactly are these properies that change from one point of space to the other ?
Several (primary, not reflected) sound waves arrive at each point of space. At very low frequencies there only two sound sources, the woofer and the vent tube (if there is one). At very high frequencies only our tweeter driver emits waves. At mid frequencies where the woofer and the tweeter drivers cooperate, they both function as sound sources.
Generally speaking whenever we have multiple sources their waves arrive at the reception point with different sound pressure levels and propagation delays.
That is why each driver has its own SPL and phase response at the reception point for every reception point in space. If we want our lab measurements to catch the right SPL and phase responses for all sound sources our geometry must be kept :
What we mean is that no propagation distance should be altered even slightly as we change connection cables between measurements, nothing should change the position of the speaker enclosure or the microphone itself. Needless to say that our enclosure should be final with all its drivers, vents and binding posts mounted (no holes left unoccupied) even if some of them are left disconnected. In a similar sense the enclosure volume should be damped with some material prior to measurement.
The figure above illustrates all three IRs measured for a three-way speaker. We can clearly see that as expected the (red) tweeter IR arrives at the reception point (microphone) first. Second comes the (light blue) midrange IR and last is the (yellow) IR of the woofer.
If we go back and study the geometry of the typical speaker presented we can easily isolate all three propagation paths to the microphone. Usually it is the tweeter that is closer to listener ears. Propagation paths differ by a few millimeters or centimeters. Even these small differences produce phase delay differences that are involved in phase responses and can not be neglected during design. That is why geometry should be kept fixed among different driver measurements so that these differences do not change during our measuring process.
The IRs presented above will easily generate (through FFT) the correct SPL and phase responses needed by our crossover network design software.
What we mean is that the propagation paths presented in the figure above (with our microphone in the reception point) should be similar to those involved in a real world listening room with the human listener placed about 2 to 2.5m away from the speaker at tweeter axis approximately. Indeed we can't measure at such a distance. But if we measure at 1m the propagation path differences (that affect crossover design accuracy) will not change significantly. Thus measured phase responses at 1m will allow for a successful design.
-Why not measure at 10 or 20cm away from the speaker's baffle as suggested in a previous tutorial on lab measurements ?
No we shouldn't because at such a distance propagation paths will have bigger differences so driver phase responses will not be suitable for our simulation software which develops targets for real-world listeners.
-So why engineers use near-field measurements ?
This is a different case. These measurements are for very low frequencies only. They are not used by crossover design. They are only used to extend the conventional SPL response (at 1m) towards the very low frequencies. After this 'correction' is applied a new phase response is generated in the low frequencies also. This phase response for the woofer driver (+vent) is not used by crossover design and to some extent is erroneous. But this issue is far away from our discussion.
-Can we summarize ?