The gated frequency response can be divided into two families, depending on the chosen axis between the mic and the driver:

The On-Axis with the Driver Center (OA-DC) is made by placing the mic in line with the driver under test center, at a distance, measured from the front baffle, that varies from 0.5 to 1 meter depending on the driver. When measuring the driver all the other one's must be disconnected, thus only the DUT has to be fed with the signal. Each driver is measured using this technique, keeping the chosen mic-baffle distance the same for all drivers; as a result we will obtain the frequency responses of all the drivers at their best dispersion angle (on-axis). Remember that the obtained phases of the drivers are not comparable with each other because the microphone location changes at every measurement, as well as the gating markers; therefore don't use the OA-DC phase to understand the driver offsets.

The next step is to measure each driver Off-Axis at 30^° - 60^° : this is performed by placing the mic-driver center to form one of those angles; I usually just leave the microphone always in the same position while orienting the speaker to reach the desired angle. These measures should be performed for each driver, giving us the frequency response at their worst dispersion angles (30^°- 60^°); the Off-Axis response is fundamental in choosing the right crossover frequency for the drivers used, so don't be lazy and do it!

In the On-Axis with the Acoustic Center (OA-AC) the microphone is placed at ear height, for a seated listener; since the tweeter is usually found at this height, is a common rule assign the acoustic center to the tweeter center. Of course this is a generic rule that works pretty well in any 2-3 way speaker; however remember that lots of possibilities exist, thus each project needs its own study. The distance mic-speaker, as well, has to be wisely chosen because the drivers integration depend on it; a 1 meter distance works pretty well with a 2-3 way system but, again, every speaker needs its own study. The OA-AC is performed with all the drivers working, giving us a complete view of the speaker response; but since this measurement is made by leaving the microphone always in the same position, as well as the gating markers are unique, the phase response of each driver is comparable. It's also useful to perform the OA-AC with and without the crossover to see what happens adding different order/cut crossovers.

But let's start to measure the On-Axis with the Driver Center using Speaker Workshop! We are going to use the same cables-hardware seen in the near-field measurement, just you will need an higher mic pre-amplifier volume setting, because of the increased distance from the baffle. With everything connected and the volumes set to get a 20-25k (depending on your card clipping level) on both channels of Speaker Workshop VuMeter, Sample Rate and Sample Size to the maximum that works best with your card as you tested in the SWSetup section, choose a REPEAT COUNT of 5 (in OPTIONS/ PREFERENCE/ MEASUREMENTS) or more, depending on your test-room noise and used Sample Size (I use 20 repeat), and check that the USE PREEMPHASIS box is unmarked; in the same window click on MARKERS and make visible the 3 choices:

Then click twice on and then on or click on MEASURE/ FREQUENCY RESPONSE/ ON AXIS; check SW VuMeter to be sure the values are almost equal for the left and right channel and in the 20-25k range; a new file will be created: "Focal A.On Axis", but it's not useful right now since we haven't set the markers yet! Once the volumes setting are right, click on or MEASURE/ PULSE RESPONSE and open the new created file: "Focal A.Impulse" .

As we can see the first impulse starts at around 8.5 ms and goes diminishing in amplitude till around 12.5 ms where a second impulse makes its appearance: this is the first pulse that reaches the mic reflected by the wall, whose distance is equal to the time length of the second pulse divided by 3 (expressed in meters) that is 12.5 (second pulse) – 8.5 (first pulse) = 4 / 3 =1.33 meters that, in my case, is the side wall distance from the mic. So we have to set the markers (the 2 vertical brown lines) to delimit just the first pulse; the first marker should be set just before the starting of the first pulse at 8.5 ms. but I suggest you to leave it at 0 ms, as I will explain later on; the second marker should be positioned just before the second pulse starts, at 12.5 ms in this case. To move the markers just click on them and hold the left mouse bottom and move the mouse to the desired location; use the zoom option to better see the chart. In this case we will have a 4 ms of useful data, that is not a lot but enough for our purpose: a bigger room with a bigger floor-mic distance would have delayed the second impulse starting. In the download area you can find an Excel file that allows to calculate when the floor bounce occurs knowing the microphone distances, as well as the expected impulse length and the useful lower frequency. To increase the time length, you could place the speaker laying on the floor with the drivers facing the ceiling, so that the first roof bounce will start later; however place some acoustic mat between the speaker and the floor since their closeness can be a problem. I actually prefer to test the speaker the way it will be used, so give a try if you want.

Now about the first marker: it should be placed right before the first impulse. However Speaker Workshop limits the resolution of the data looking at the sample rate and the time gate, limiting to 128 points up to a time slice of 9 ms (for a 48k sampling rate), then goes to 256, 512, 1024 with approximately each doubling of the time window; the only way to increase the number of data points is to increase the length of the gated measurement or, to some degree, increasing the sample rate. So it helps to move the first marker close to 0, as well as to increase the latency. You can check the number of data points (resolution) by right clicking in the response chart and choosing PROPERTIES, then go to the SPECIFICS tab. You can change this number and the graph will look slightly better for crossover modeling, but once you repeat a measurement SW will reset this number according to its algorithm. Do some tests moving the markers to have an idea of how it works.

With the markers set, we can repeat the on-axis measurement: click twice on and then on or click on MEASURE/ FREQUENCY RESPONSE/ ON AXIS; check SW VuMeter to be sure the values are almost equal for the left and right channel and in the 20-25k range; now open "Focal A.On Axis".

In the chart you can see three responses of the Focal woofer made at different distances; as you can see the curves that have the same shape are the red and the blue, that correspond to a distance mic-baffle of 80 and 100 cm, thus we can say that from 80 centimeters we are in the far-field measurement for this driver. Remember that every time the mic-baffle distance changes, you need to redo the Pulse Measurement as well as you have to reset the markers! The lower frequencies are not very useful because of the limited first pulse time length: in our case 12.5-8.5=4 ms that corresponds to 1000/4=250 Hz; however this is not a problem since we have the near-field response to take care of the low region.

Now it's time for the Off-Axis measure: leaving the mic in the same position used till now, rotate the speaker baffle to form a 30^° angle with the microphone horizontal line; then leaving the markers as for the On-Axis and without changing the volumes, click twice on then on MEASURE/FREQUENCY RESPONSE/30 DEGREE and a new file will be created: "Focal A.f30". Repeat the procedure rotating the speaker baffle to reach a 60^° angle and again, leaving markers and volumes as they were, click on MEASURE/ FREQUENCY RESPONSE/ 60 DEGREE and again a new file will be created: "Focal A.f60". To go deeper, you can perform this measure at 45^° as well, even if SW doesn't have this option: place the speaker to form a 45^° angle and click on MEASURE/ FREQUENCY RESPONSE/ GATED and rename the "Focal A.Gate" in "Focal A.f45".

Now open "Focal A.On Axis" and add to the chart the two new files created to have a complete view of the driver response at various angles:

As you can see the Focal acts very well at 30^° making possible a high cut crossover; the last chart On-Axis response looks a little less “sawtooth alike” then the previous (512k point) I showed since I reached the 1024k points of resolution increasing my card delay (the DMX Fire allows to do it from its control panel) and putting the first marker at almost 0; also a sample rate of 96k helped to achieve it.

Now let's do the On-Axis with the Acoustic Center. Measuring all the speaker, with its crossover, will give us a response that combined with the near field one will show us the complete speaker response, like the ones we see in the magazine.

 What I will show you now, instead, is a procedure based on the technique described by David Ralph, to measure each driver response to correctly import them in a simulation program, with the aim to preview the filters to be used.

Place the microphone as I described before (look at the drawing at the top of the page), and start with a distance of 1 meter; connect all the drivers and be careful with the amplifier volume since the tweeter doesn't have a crossover yet! Now click twice on and then on or MEASURE/ PULSE RESPONSE; check SW VuMeter to be sure the values are almost equal for the left and right channel and in the 20-25k range; open "Focal A.Impulse" and set the markers as you already know; with the markers set, click twice on and then on or click on MEASURE/ FREQUENCY RESPONSE/ GATED; check SW VuMeter to be sure the values are almost equal for the left and right channel and in the 20-25k range; a new file will appear: "Focal A.Gate". Since all the drivers were sounding, rename the file in "Focal A.Gate Total". Now just leave connected one driver at a time and without moving the mic nor changing the volumes set and the markers, repeat the procedure for each driver, remembering to rename each driver file to easily recognize it; so in the end I came out with 3 files, since the system is a two way: "Focal A.Gate Total" , "Focal A.Gate FOCAL", "Focal A.Gate ATD".

To give a better look at the chart, you can do some smoothing by clicking on TRANSFORM/ SMOOTH and selecting a 1/8 octave then APPLY to see the result immediately; if you press CANCEL the smoothing will not be applied while clicking on OK will make it permanent. I don't suggest to do the smooth since it alters the Phase Response. Import the single drivers responses in your simulation program, as well as the all drivers sounding one (import it as a virtual driver or as a reference response, depending on the used software). Now it's time to talk about offset: it is the distance between the acoustic centers, on the 3 space axis. If we take the tweeter as reference (therefore X, Y, Z =0) we can easily measure with a ruler the existing offset between the tweeter and the woofer both on the horizontal X axis and on the vertical Y axis. To measure the Z axis depth offset is very difficult, using a ruler; however knowing its real value is not important for the simulation: what we need is the relative Z offset, so that we can correct the woofer phase response to obtain true filters predictions. The depth offset we will find, is not the real offset! Let's go back to the simulation program, and input an offset=0 on all the tweeter axis, since it is our reference. Then inform the program of the mic-baffle distance, and that the mic was on axis with the tweeter. Now the woofer: input the measured X and Y axis offset; now change the Z axis offset till the imported measured all driver working response is equal to the program calculated drivers sum response, especially in the drivers crossing region. Once you get it, you are ready to start simulating. In a 3 way speaker, you must work on couple: first woofer + midrange, with the mic on axis with the midrange (that is the reference) and regulate the woofer offsets; then midrange + tweeter with the mic on axis with the tweeter (that becomes the new reference) and regulate the midrange offsets. To test the procedure, I compared the measured and simulated responses of a speaker with a II° order filter on both the drivers, and a crossing frequency of 3000 Hz: the result is positive!

Filtered speaker freq.response: measured - simulated

Now let's go back to SW and take a look at the phase response chart:

The response is not easy to read, since we haven't detracted the mic Time of Flight, and shows the relative phase. SW allows to automatically subtract the time of flight, but it doesn't work very well, so don't do it. You could do it manually, but be aware that it is a risky job. I suggest you to take a look at the folder SW Phase.

But let's Splice together the far-field with the near-field!