This is another topic that I wrote about and posted in one of the Facebook pages for audio stuff.
It's a long post, but follow through and study the measurements and it will all make sense.
I see lots of guys here running wide band (or full range) drivers on their open baffle speakers. Most go for large diameter drivers as well. Unfortunately most large diameter drivers struggle to play the top octave. They also have very poor off axis response because of the size of the diaphragm. So the rooms are loaded un-evenly. Lower frequencies and even mid-range areas are somewhat omni direction minus the 90 degrees off axis cancellation of the open baffle design. But upper mid-range areas and all the higher frequencies beam the signal forward only, and as frequency increases the coverage area gets smaller and smaller. You hear the upper ranges only if your ear is directly in line with the drivers.
Off topic but another issue is that most of the wide band drivers are run without a filter at all. I have yet to measure one that did not need some sort of filter to control a rising response, break up, ringing, and in some cases a baffle step loss. Don't worry guys, high quality parts in a filter won't degrade the signal, but they can solve or help control amplitude and break up issues. And those issues are much worse and much harder to listen to.
Okay back to adding a super tweeter. Most guys think that going for some type of ribbon or air motion device is the way to go. It's not. Most are too tall and limit dispersion. You are trying to add the lost upper ranges in a universal pattern and not a limited one.
Another issue is that often the acoustic centers are way too far apart. So the two drivers will always sound like too different drivers playing from two different sources. To accurately blend to drivers crossed at 10kHz then the acoustic centers of the two drivers need to be within 1.25" apart. This is because the wavelengths are so short. And that is not possible.
You are much better off allowing the tweeter to cover as wide of a range as possible and cross as low as the tweeter will allow. And it doesn't matter how good you think your full range driver is. Any tweeter from 2kHz and up will outperform it in every way. It is just flat out going to have better dispersion, speed, and resolution due to its smaller size and much less moving mass.
But many insist on running the wide band driver full range and dropping in the tweeter in the top octave. This doesn't work like most think it does. It really doesn't fill in the upper range as much as it causes huge suck outs over a really wide range. This is especially the case when a single capacitor is used on the tweeter as a filter, and it is especially worse the farther the acoustic centers are from one other. I will illustrate exactly what happens in the measured responses that I will follow this with.
So to illustrate what happens. I used a FR-125 full range driver. It has an exceptionally smooth response compared to most drivers. And it sounds really good too. It is a 4" diameter driver. The cabinet is a small ported box. The front baffle is 9" tall by 6" wide and the driver is mounted in one end of it. To show the effects of the distance differences I have taken measurements with the woofer farther from and closer to the added tweeter that I will sit on top of the box.
First lets try what everyone does. I have oriented the woofer to near the bottom of the box (flipped it over) and added an Aurum Cantus G-2 ribbon tweeter to the top of it. To minimize cancellation effects I used only a 1.0uF capacitor on it. The green line is the FR-125 driver alone and the purple line is the ribbon tweeter alone. Note how low in range the ribbon still plays even with a tinny 1.0uF cap on it. Since the full range driver is the primary driver I took the measurements on the full range driver axis (in line with it).
So we should get green plus purple and equal additional output right? Okay, here is the sum of them together in red. What happened? This isn't a better response. There is almost a 20db swing in response. Well what's going on is that the short wavelengths arrive out of phase with each other from the two sources due to distance differences. This is the affect most people get when doing this.
Now lets move up to the tweeter axis. Note how the top end picked up on the tweeter. That is because of the height of the tweeter. It will have a limited vertical dispersion due to its length that is about two and a half inches tall. So getting in line with it will bring the top end up and it changes the time delay to the lower full range driver.
Now let's combine them again. We still have multiple cancellation points where the output of both drivers is less than the output of one driver. It's still a big mess. And this is just what is happening at one point in space. In the off axis it gets much worse.
Now lets trying varying the offset. So for this measurement I pushed the tweeter forward by 3/8". Yes, three eighths of an inch! The new measurement in the orange line. It moved a lot didn't it, and it is still a mess.
Okay, now I am going to push the tweeter forward to 1/2". Yes just one half of an inch forward. It is now the yellow line. Keep in mind that it is just pushed forward only one eighth of an inch further than the measurement that is the orange line. That's what happens in those shorter wavelength ranges. It requires very close center to center spacing and steep filters to avoid these effects.
Now let's try flipping the polarity of the tweeter to see what happens. Now we have coupling above and below the crossing frequency and cancellation at the crossing frequency.
This time I moved them around to find a distance to where they coupled the most. The red line is the combined output again.
Now pay attention. Watch what happens when I move the microphone up just 4". Keep in mind also that the microphone is 1 meter away. The response moved a huge amount. You see, even if you use a digital filter to correct the response in one spot, moving just out of that spot nullifies the whole thing.
Now I am moving the microphone up another 4". It made the yellow line. Now it is coming back up again, but still showing cancellation effects.
Let's try something else. Let's use a 1" dome tweeter. It will have a more even off axis response in all directions. Here the full range driver is in green again. The purple response is the dome tweeter with a 1.0uF cap on it. The level was a little high. Most people think they can just add an inline resistor to balance the output. So I added an inline resistor to bring the level down. Note though, that the level did not come down evenly. It brought down the top, but left the bottom end. It is in light blue. That didn't help at all. These were measured on the tweeter axis.
Now here is the measured response of both drivers on the full range driver axis.
Now here in red is the sum. Not too bad for a first shot, but it didn't help the overall response any.
Let's flip the polarity of the tweeter and see what happens. That is it in red. This made them in phase across the board for this point in space. This is not an easy spot to find either. The response can easily fall apart anywhere else that you can measure from.
Now here is just the summed response in red. Let's see how fragile this is now.
So I pushed the dome tweeter forward three eighths of an inch and this happened (in orange).
Now the tweeter has been pushed forward a total of just over one half of an inch. See it in yellow. Moving the microphone up or down also changes the time delay and has the same effect. Quite the mess huh? But a lot of people think they can just mount a tweeter on the baffle somewhere and it will fill in just fine. The really bad part is that many people really don't hear the upper octaves very well and don't know any different.
Another really bad part is that this negatively affects the imaging and sound stage layering by the cancellation effects and holes in the response. This creates a very unbalanced room response.
Now let's look at a solution. Some people resort to rear firing the tweeter. This helps in that it doesn't effect the on axis response but in an open baffle design it does affect the rear wave output just like in the illustrations that you saw already. So the in room response is still negatively affected.
The real way to implement a super tweeter without any of the negative effects that you've seen here is to face the tweeter upward. And ideally line the tweeter up (tweeter voice coil) so that it is inline with (or centered on) the voice coil of the full range driver. This will keep the drivers time aligned facing forward and backward. So the full range drivers limited off axis response won't allow it to overlap to the limited (90 degrees off axis) off axis response of the upward facing tweeter. No overlapping.
The great part is that this adds the low output, upper octaves, back to the room response, and balances out the room load. This more balanced sound and upper range extension greatly improves the imaging and sound stage layering that is present in those upper octave harmonics. All of those spatial cues in the music are up there in those upper ranges. Bringing them back can be a revelation.
So let's see. Here is the on axis measured response of both drivers. Note the tweeter is very down in output over 90 degrees off axis.
The red line is the sum of the two. Looks pretty good huh?
Now let's improve them further by flipping the box over to get the drivers acoustic centers closer together. This will also help make the driver blending better.
So here is the new on axis response. And yes a little filter on the full range driver to compensate for baffle step loss would bring the level above 1kHz down to match the level below that, and balance out the sound. See I told you guys they always need a filter.
Now let's test the concept and measure in the vertical off axis again. Here is a new measurement (in orange) of the response with the microphone moved up 4".
Now up 8" (yellow line).
And up 12" (green line). So even in the vertical off axis the response remains clean and free of coupling and cancellation errors.
For you guys running a full range driver try adding an upward facing tweeter and hear what happens.