Frequency response measurement - far field
The easiest high quality measurements in a home environment are gated measurements. The longer the reflection-free gate time, the lower in frequency the measurements are valid. A gate time of 1 millisecond (0.001 sec) results in a frequency resolution of 1/0.001, or 1000 Hz. Your measurement system may show energy below that, but everything below that frequency is not reflection free. The data then has points above this frequency that are linearly spaced, at 1000 Hz, 2000 Hz, 3000 Hz, etc. If you can manage a reflection free time of 5ms, you can get data spaced every 200Hz,
The reflection free time can be calculated with geometry. The best place to put a speaker in a particular room is in the middle of the shortest dimension, usually the height (commonly around 8' or 2.4m.) The speaker should then be placed at 4' high (1.2m) and roughly this distance from the wall behind the speaker - if you can't raise the driver, do the calcs below as if the smallest room dimension is 2x the distance from the driver to the nearest boundary. Now to choose a measurement distance. If the distance from the speaker to the microphone is X and the smallest dimension of the room is Y, then the path length for the first reflection (from pythagoras) is 2*sqrt((X/2)^2+(Y/2)^2) and the path length difference (PLD) is this quantity minus X. Plug in the numbers using a measurement distance of 1m and PLD=2*sqrt(0.25+1.44)-1 = 1.6m The time for sound to travel this distance is 1.6m / 345m/s= .0046 seconds, giving a frequency resolution of 345/1.6 or about 215Hz. You can do the calculation with measurements in feet by using a speed of sound of 1132ft/sec instead of 345m/s, and in inches by using 13580in/sec. With a measurement distance of 3m, your PLD drops to 0.84m, causing your frequency resolution to be around 400Hz. The maximum PLD occurs when the distance from speaker to microphone is zero. You don't want to do this because the measurements should be taken in the farfield, roughly 5-7 times the driver diameter (or the span of multiple drivers) at a minimum.
The measurements should be taken on axis and off axis. Crossing over above the point where the directivity narrows will cause crossover gaps off axis. The measurements used for crossover design should be taken in the intended box to include the effects of cabinet diffraction.
Nearfield or groundplane measurements can be used to extend the measurement frequencies below what can be done using gated techniques. Nearfield measurements must be corrected for diffraction, but groundplane measurements include this. These low frequency measurements can then be spliced to the farfield measurements.
D.B Keele has some seminal information on nearfield measurements that is a must read.