Does A beyma woofer and a Tang band tweeter make an excellent 2 way system?

Where is the woofer down -6dB at 30 degrees off-axis compared to axial? That's the highest frequency where you'd want to cross over to a tweeter.

This tweeter's been successfully used with low (under 1.5 khz) crossovers many times. How well it would keep up with the Beyma is another question.

http://www.parts-express.com/pe/showdetl.cfm?partnumber=275-140

I'm confused. If you already built this, why are you asking?

I believe a 2nd order crossover at 2300 Hz used with these two drivers will result in sound with a big hole in the mid range performance.

That 10" woofer might be going into breakup at that frequency. With a 2nd order crossover, you are asking it to produce smooth sound at least one octave higher or as high as 4600 Hz.

This calls for a much smaller diameter woofer, or a 3-way design.


You can't realistically expect any 10" to have wide off-axis performance at that high a frequency. Bill Fitzmaurice is right. The frequency at which that driver starts producing sound in a narrow pattern heard best only directly in front of it (beaming) is likely to be well below 1000 Hz. That will be well below that Tang Band tweeter's ability to make undistorted sound. Tang Band has been known to be rather optimistic in its driver spec. sheets, but you shouldn't buy their info without testing it yourself.

How wide is your cabinet? Did you measure that tweeter's response while mounted in that cabinet. The resulting frequency response curve will look significantly different than the manufacturer's curve because of cancellations caused by diffracted sound coming from the cabinet edges.
Edited by R Swerdlow - 3/30/13 at 5:14pm

I didn't pay attention to the date of your original post. My fault.

I see your enthusiasm for DIY speaker building, and I want to encourage it, not pour cold water on it. Please understand my comments with that in mind.


Yes, I looked a the Beyma spec sheet. It only showed an on-axis frequency response curve. I suspect (but don't know for certain) that this driver was made for PA or guitar amplifier purposes. It looks fairly smooth for a 10" driver at high frequencies, but only on-axis.

For PA or instrument amp uses, beaming sound in the mid range is not such a problem as it is for hifi audio, where it will be crossed over to smaller high frequency drivers. The problem comes in the disconnect you'll hear in sound as it goes from a beaming woofer to widely dispersed sound from a smaller diameter driver. You may hear the mid range sound directly on-axis, but off-axis the sound will drop significantly. That will significantly impact the detail of the critical mid range sound, and the lack of dispersion will reduce the ability of speakers in creating a realistic stereo image.


I doubt it. Here are 3 very good quality 10" drivers made by Seas. I use these as examples because I know that Seas publishes reliable frequency response curves measured at 0°, 30°, and 60° off-axis.

http://www.madisoundspeakerstore.com/approx-10-woofers/seas-prestige-ca26rfx-h1305-10-coated-paper-cone-woofer/

http://www.madisoundspeakerstore.com/approx-10-woofers/seas-prestige-ca26re4x-h1316-10-paper-cone-woofer/

http://www.madisoundspeakerstore.com/approx-10-woofers/seas-prestige-l26rfx/p-h1209-10-aluminum-cone-woofer/

With those curves, you can estimate by eyeball what Bill Fitzmaurice described when he said look for the frequency where the sound is (no lower than) 6 dB compared to the on-axis curve. Above that frequency range, these 10" drivers are beaming too much. It's a simple matter of physics, no 10" diameter driver can do much different.

You can also clearly see the frequency where these hifi 10" drivers are going into break-up mode. That kind of noise is ugly and should be suppressed by the crossover. You can also see how the cone material effects how big the break-up peaks are, and at what frequency they begin.


A rough rule of thumb says that you can safely crossover a tweeter at double the Fs if the crossover slope is 4th order, and roughly triple at 2nd order. But that is a general rule of thumb only. You really need to know how the tweeter's frequency response curve looks when mounted in the cabinet.

This short article is an example of what I mean http://www.salksound.com/wp/?p=135#more-135. (I hope you read the whole article ).

Look at the raw frequency response of the tweeter in the 2nd graph in the article. See the peak at roughly 1700 Hz, followed by the dip at roughly 2700 Hz. This is caused by the interaction of the tweeter and the edges of an 8" wide rectangular cabinet. All tweeters will have a similar profile in that cabinet. But none of this shows up in the manufacturers spec sheet, because it is usually measured with the tweeter mounted on a very wide baffle.

If I understand what you said, where your tweeter is, your cabinet is about 1.5" wider than the tweeter plate? I couldn't predict how your cabinet edges will impact the tweeter, but I hope that article makes it clear why you should know that as you design a crossover.
Edited by R Swerdlow - 3/30/13 at 7:17pm

If you've already built this 2-way, I wonder if you'll really want to change things, but we can always discuss ideas.

The biggest problem I see is trying to get both more bass output and extended high frequency range from a 10" driver. You can't have it both ways without something suffering. You may have measured a frequency response (FR) as high as 4000 Hz, but a FR curve only tells you how loud the sound is at a given frequency. It won't tell you if the speaker is in break-up mode making unwanted noise.



In this FR curve, look in the 2,000 to 5,000 Hz range. There is a sudden drop of nearly 10 dB in response just below 2,300 Hz, followed by a rapid climb by nearly 20 dB by 3,300 Hz. Above that frequency, there is a steady drop off in response. This is characteristic of a paper coned driver in break-up mode, and suggests (but does not conclusively demonstrate) that's the Beyma 10" is also doing that. The rapid up and down swings seen the harmonic distortion curves between 800 and 4,000 Hz also suggest something unwanted may be going on.

The sound may be as loud in the 4,000-5,000 Hz range as it is below 1,000 Hz, but it will not be smooth sounding. At lower volumes this may contribute to an apparent increased detail over and above what's in the recording. Do brass trumpets or horns sound more detailed or more forward? Some people may at first like this increased edge in the upper mid range/lower treble. If you increase the volume, do the brass horns or trumpets sound too forward or become fatiguing? This will vary with the type of music or sound. You have to have a signal that produces those frequencies to stimulate that response.

I wonder if a 2nd order low-pass filter at 2,300 Hz will effectively suppress the break up noise. You can try using a lower crossover frequency, using steeper roll-off slopes, or adding an LCR trap filter to suppress the noisy frequencies.

And then there is the problem of beaming, but I already mentioned that.

Dennis Murphy is a self taught crossover designer with over 25 years experience. He has designed most of the Salk speakers, all the DIY designs at Murphy Blaster Productions, and more recently all of the Philharmonic Audio speakers. Over the years, I have become familiar with his speaker designs, and I've learned that I like his design taste and philosophy. In particular, he favors a flat FR curve over as wide a range as possible. He also depends on his own accurate measurements as a necessary step in crossover design. He has successfully used a variety of 2nd, 3rd, and 4th order crossovers, and is not wedded to a single type. He also makes the distinction between a crossover's electrical performance and roll-off curves, and the acoustic roll-off of a driver combined with the crossover. What matters is the acoustic performance. He has very rarely (or never) seen a successful speaker with a simple 1st order (electrical) crossover. There are some good commercial designs such as Vandersteen, that tout 1st order crossovers, but a quick look at them reveals that they are far from simple minimalist crossovers.

Baffle step compensation (BSC) is not so much a matter of preference as it is based on knowledge of how close a speaker cabinet will be placed to the wall behind it. The further away from the wall, the more BSC is needed to maintain a flat FR curve. In your 2-way speaker with a 6½" woofer, did you use BSC or avoid it to keep the sensitivity higher?

Others may have different opinions than mine .

Zaph Audio's driver tests are a very good source of info and his measurements are widely considered as reliable.

The other web site you mentioned (http://www.mh-audio.nl/index.asp) has "textbook" calculators for crossovers, but you mentioned how you've already observed how small changes in components have unexpected effects on the the rest of the crossover, such as in the notch filters. The trouble with the crossover calculators based on "textbook" values, is that they assume driver impedance is constant across the bandwidth.

Yes, I agree, there is much to learn in this hobby .
Edited by R Swerdlow - 3/31/13 at 10:45am

Interesting. So the 10" woofer got a 4th order low-pass at 2300 with a zobel and an additional notch filter to quiet the worst break up. It seems like you tried nearly every reasonable approach. I had thought you used a 2nd order filter – I must have misread something.

In my previous post when I speculated about the woofer's break up, I forgot to point out another sign of something amiss. In the impedance vs. frequency curve, there is a small blip between 800 and 900 Hz. It may look tiny and ignorable, but in this range I'd hope to see a very smooth impedance curve. Because it occurs at the same frequency as the onset of the harmonic distortion swings, its another warning sign that something you don't want to hear, like break-up, is beginning.



The tweeter has gotten little mention. 2nd order high-pass at 2300 Hz shouldn't be a stretch for it. I'll guess it sounds good. Did you have to pad it down to get it at an even loudness level with the woofer, or were they pretty close?

Yes, your long term listening sessions will tell whether or not you like the dispersion and imaging these speakers can create. In my thinking the Beyma 10" would be a very worthy bass driver in a 3-way speaker. If you cross over the bass to mid range at roughly 400 to 600 Hz, you might not need the zobel filter for it. But a 3-way crossover will also introduce other cans of electrical worms for you to wrestle with. That's for you to decide.

At least you did deal with the big problem of filtering out the woofer's break up noise. To my thinking hearing that is much worse than not hearing enough mid range dispersion.

LinearX, are you taking measurements and simulating this crossover? Would you mind posting your results?

Generally speaking you will rarely see a 10" woofer used higher than about 2khz. Even if you find a 10" woofer that is flat to 10khz it will be a mess off-axis. As a rule of thumb, I try to minimize waistbanding of the horizontal directivity. Waistbanding is where the directivity narrows and then widens as frequency increases. Instead you should shoot for a gradual transition from wider directivity to narrower directivity with minimal fluctuation. I'll explain below why you are likely not achieving this with your present design.

You can generally estimate the directivity of a woofer based on its diameter from surround to surround. Your 10" woofer is likely about 8" across between the surrounds. This corresponds with an ~1700hz wavelength. This is important because around 1 wavelength the woofer's directivity should be in the ballpark of 90deg. Below ~1700hz it will approach omnidirectional and become a function of the baffle and boundary spacing (aka BSC). Above ~1700hz it will narrow in directivity. This is referred to as beaming.

The tweeter, due to its 1" size will be radiating omnidirectionall below about 3khz...well except that the baffle itself is likely acoustically large. Assuming a 12" baffle, the tweeter should have a directivity of about 180deg from about 1200hz up to about 13khz.

(Caveat: These directivity numbers are rough estimates based on driver sizes. It is is better to measure these for your actual drivers and baffles.)

So with a 10" woofer crossed to a 1" dome around 2300hz I would roughly estimate your horizontal beamwidth as follows:

1000hz - 140deg
1300hz - 110deg
1700hz - 90deg
2000hz - 75deg
2300hz - 75deg
3000hz - 130deg
4000hz - 180deg

To compare to a more common standard like a 6" woofer crossed around the same spot:

1000hz - 180deg
1300hz - 180deg
1700hz - 150deg
2000hz - 120deg
2300hz - 110deg
3000hz - 150deg
4000hz - 180deg

Neither is ideal, but you can see how narrow the 10" woofer version gets. If you take the 6" version and cross it around 1800hz the directivity would be great (for a 1" dome 2-way that is). Here is an example of a very nice 2-way dome commercial speaker that exhibits good directivity with minimal waistbanding: http://redspade-audio.blogspot.com/2013/02/sonus-faber-cremona-auditor-stand-mount.html

You could also cross around 1500hz to the Beyma 10 (someone mentioned this tweeter is comfortable that low):

1000hz - 140deg
1300hz - 130deg
1700hz - 140deg
2000hz - 170deg
2300hz - 180deg
3000hz - 180deg
4000hz - 180deg

So you might be wondering, "why do I want to avoid waistbanding? I can sit with my head in a vice". The reason is that you don't only hear direct, on-axis sound...especially in a home environment. There will be an "unevenness" to the sound because in the range that is rather narrow (from 1700hz - 2500hz) the total sound energy will be far less than the total sound radiated into the room above and below that. I'd go further in depth but if you search on constant directivity or uniform directivity there is enough discussion already out there.

Of course, some people completely ignore directivity. IMO, a uniform off-axis response is the second most important metric when designing a speaker (behind on-axis FR).

My suggestion would be to measure for directivity and try a cross in the 1500hz range (maybe even a bit lower). If you want to quickly hear these differences use an active DSP and switch between two settings where it is flat on-axis with a cross around 1500hz vs the same around 2300hz. The key is to do this in room with both speakers running.

Thanks cocostan for helping to make this important point clear. I thought I had talked enough about it earlier, but I guess it never hurts to repeat this point .

Think about a bare light bulb, and the same bulb with a bowl reflector and lens, a spotlight. Light travels from the bare bulb in all directions, up, down and sideways. Nearly all the light from the spotlight is contained in a narrow beam. The narrow beam from the spotlight carries farther away than light from the bare bulb, but light from the bare bulb spreads over a wider area than the spotlight. Which is better? It depends on how far away you are.

The same goes for sound from speakers, but with one important difference. Sound waves are significantly longer than light waves. The wavelengths of sounds in the audio range are:

20 Hz = 56.5 feet
200 Hz = 5.65 feet
2,000 Hz = 0.565 feet = 6.8"
20,000 Hz = 0.0565 feet = 0.68"

Because sound wavelengths can be similar to the diameters of speaker cones, something important happens. Sound with a frequency of about 1360 Hz has a 10" wavelength, the same as the diameter of the 10" woofer we’re talking about. At lower frequencies, the sound waves are much larger than the cone, and the sound radiates widely (or disperses well), similar to light from a bare bulb. As the sound wavelength decreases, approaching 10", the sound disperses less and less widely. At wavelengths shorter than 10", the sound beams instead of disperses, similar to light from a spotlight. This transition is gradual.

If sound from a speaker (in normal sized rooms) is widely dispersed, we hear it as spacious, and if two speakers produce widely dispersed sound in stereo, we hear it as a well-defined stereo image. The wider that speakers disperse their sound, the better we perceive the stereo image.

Speaker beaming, or waistbanding (I like that term ), is an important aspect of speaker design. A 10" woofer can move more air and dig deeper than a 6½" woofer, and it projects acoustic power over a greater distance, like a spotlight. But the 6½" woofer can disperse that acoustic power better (up to about 2,000 Hz), like a bare light bulb. Because rooms in our homes have walls that reflect sound, and because we tend to sit about 10 to 15 feet away from the speakers, this dispersed acoustic power sounds different and better than acoustic power projected from a beaming driver. In a large auditorium, the large woofer might still be beaming, but the listeners might be sitting 100 to 150 feet away. The speaker's beamed sound will be spread wider at that longer distance.

So in speaker designing, it's worthwhile to use a woofer at frequencies where is disperses sound widely, and cross it over to a smaller tweeter at frequencies where it begins to beam. That way, you avoid waistbanded sound.

power response may even be more important than on-axis response because on-axis response can be fixed with equalization, power response cannot.

Can you explain what is "power responce" and how is it adjusted in a crossover? Also how does one maximize on-axis performance within the actual crossover?

Power response is the average of axial and off-axis response. In the omni-directional lows power and axial frequency response are the same, as when you look at a polar plot there's equal SPL in all directions. As the off-axis sensitivity drops, seen in a polar plot as a shift from a circle to a lobe to a narrowing lobe, power response goes down. Where it enters the crossover equation is that you want to crossover from the lower frequency to the higher frequency element before the lower frequency element has too much directivity, typically no higher than where the 30 degree off-axis response is down 6dB from axial. If you crossover too high there's a tightening of the polar response at the upper end of the lower frequency element, which balloons out to a wide polar response at the low end of the higher frequency element. You won't see the implications of that on an axial chart, where response will appear flat. You wouldn't hear it on-axis in a very dead room. But you'll hear it off-axis, and in a normal room where you hear both the axial output and the off-axis reflecting off the walls you'll hear it on-axis as well, as a response dip in the combined response of the direct and reflected waves.

So let me see if I got this straight, Power responce is the sum or average of both the on-axis and off-axis frequency responce?
What makes the off-axis sensitivity drop? Do you mean the same sensitivity as the impedence? Or is this a seperate term all-together?

I understand that its optimal to have the low frequency driver cross to the high frequency driver before the low frequency driver becomes
narrower in its responce or directivity. I am also curious as to how you determine the 30 degree off-axis responce? Is this something that
is measured with something like REW?

Is this still the case when designing a crossover in a 3-way design? (ie low frequency woofer to mid-frequency driver to high frequency driver-tweeter)

Sorry for what may seem like dumb questions to you guys, but I am just starting to learn about crossovers and acoustics. LoL, I am trying
really hard!

It may seem counter intuitive, but it is important. That's why some of us have written so many words trying to explain it. Remember that the Beyma woofer was designed for pro audio, where the listening distance may be much greater than in home audio.


If you're going to try using that tweeter at 1500 Hz, it may be safer with steeper 4th order slopes. Remember that with a 2nd order crossover, the tweeter will still be trying to make sound (12 dB quieter) at one octave lower than the crossover point, or at 750 Hz. A few more crossover parts will cost less than replacing the tweeter.


You have to measure it to find out what your woofer does in your cabinet. Measure frequency response with the test microphone 1 meter away on-axis, and again at 30° and at 60° off-axis. You can move the microphone or move the speaker to get those angles. As Bill Fitzmaurice said, make the crossover frequency "no higher than where the 30° off-axis response is down 6dB from axial". At a practical level, it's easier to look at it that way.

Here is an example of a Seas 10" woofer (CA26RFX) where the manufacturer's spec sheet shows 3 frequency response curves in one graph, the dark line is measured on-axis, and the fainter lines beginning at 500 Hz are measured at 30° and 60° off-axis.



Wait to see what Tang Band says. Conventional practice uses a tweeter in a closed chamber separated from that of the woofer. Many tweeters come with a closed back, just for that reason.


Yes, power response, off-axis response, dispersion, lack of beaming, what ever you call it, is important for any crossover between any two drivers.
Edited by R Swerdlow - 4/3/13 at 8:25am

No problem. IMO, one of the differences between a decent home speaker and a great home speaker is how off-axis response is handled. You can achieve whatever on-axis response you desire through the crossover. The off-axis response is first dictated by the physical characteristics of the speaker and no crossover can alter this just like a crossover can't make a 10" woofer play nicely at 20khz.

All dome 1" drivers will collapse directivity similar to that Sonus Faber speaker. The ring radiator style alters it some but not in a large way. You will also see that the baffle is not completely flat in an effort to minimize baffle diffraction. This affects both on-axis and off-axis response.

The most well-known designers (at least that I know of) that emphasize off-axis response are Wayne Parham and Earl Geddes. They go to the extent of using horns to control the upper frequency directivity. You can read some of their papers and web posts.

I'm not sure I agree with the comment about widely dispersed sound being more spacious or giving a more well-defined image. I would argue that the opposite is true. A wide-directivity speaker will simply produce more early reflections which worsen the image. Spaciousness is more a function of late reflections and those can be achieved with narrower directivity speakers.

If your only or primary criteria were FR and distortion, then yes you would choose the bandpass of your drivers accordingly. I'm suggesting this is far from optimal. I would also suggest that listening to drivers individually is not useful. How a driver sounds individually (crossover or not) doesn't hold much weight to be honest. In fact it can lead you down the wrong path. For instance a low-passed woofer will typically sound better than a high-passed tweeter when both are played individually and might skew you towards a higher cross.

Ulimately, using your ear to design a speaker is not a good practice. IMO this should be done only with final tweaking and evaluation of different design hypotheses (such as different on-axis and off-axis targets).

There is no directivity sweet spot (at least not one that is widely held). At a higher level I would suggest targeting uniformity where you don't have significant fluctuations in off-axis response. This creates a more even "total sound" which is what the listener hears in a home setting.

With a 1" dome on a flat baffle you will have collapsing directivity. The question is how you handle the difference in directivity between the woofer and tweeter around the crossover region. You will have to work some compromise of off-axis response and how low that tweeter can play cleanly. The good thing is that you chose a tweeter that appears to play fairly low for a 1" dome.

http://techtalk.parts-express.com/showthread.php?225005-Uniform-power-response-best-way-to-get-it

First google result for "power response loudspeaker". It is a pretty good discussion.

The best way to determine the off-axis response of a speaker is to measure it. You simply point a mic at it X deg off its axis and take a sweep. You can also get a rough estimate of based on the size of the diaphragm. That is what I did when estimating where a 10" woofer would be -6db at 90deg off axis. This is only a rough estimate though.

This applies to 10-way speakers. There is nothing fundamentally different about a 3-way. Retail speaker companies have been making a big deal of it for decades because it is "more". That is not to say there is anything wrong with 3-way's but like with any design exercise, "right tool for the job" rules.

---

Ultimately, it is no coincidence that most 1" dome tweeters are used with something like a 6" woofer. As seen with that Sonus Faber speaker, you can get a very nice uniform off-axis response with a 1" dome that can handle a 1st order (guessing that is electrical not acoustic) crossover at 1800hz. There is a little waistbanding, but nothing terrible and about as good as it gets with a 2-way dome speaker. This is absolutely not a coincidence. I'd bet that the woofer they use could easily play up to 3khz nicely on-axis. Given that this is a very expensive speaker, they are also not using the simple 1st order high pass for cost savings or a better on-axis response. The reason they do this is for optimal off-axis response.

Now many cheaper 2-way dome speakers would use a tweeter that might play down to 4khz since it is cheaper to find a woofer that plays that high than a 1" dome that plays down to 1.8khz. This speaker could have a ruler flat on-axis response, but it will sound far worse because there will be a suckout in reflected sound from about 1800hz to 4khz.

The Sonus Faber will sound better than this typical but similar cheaper 2-way dome with 6" woofer. The primary reason (assuming on-axis response is flat in the cheaper speaker) is not because the Sonus Faber will show less distortion (we can assume it will). The primary reason is that it will have a far more uniform off-axis response. Certainly, the lower distortion, etc add to its superiority, but my experience, along with the research I've read indicates the off-axis uniformity is what separates the decent from the good and great.

One other thing:

There are different levels of refinement. Avoiding beaming is akin to making sure it is not a bad design. Full on beaming, where the -6db beamwidth encroaches upon the direct listening area is simply a poor design IMO. That is the classic "head in a vise" requirement where it will sound bad in the next chair over or if you slouch 12" lower.

What I call waistbanding won't become an issue to direct sound for someone sitting a bit off-axis in most rooms. Waistbanding will have a greater effect on how the speaker interacts with the room and the reflected sound which is also heard by the listeners. The speaker that exhibits waistbanding won't be nearly as bad as the prior severely beaming speaker.

The final level of refinements are how the off-axis response tapers, how low the directivity narrows and the actual beamwidth. What is optimal for all this is up for debate and depends on your constraints (budget, speaker size, etc).

You are now getting to see that in DIY speaker building, there are few simple yes-or-no decisions, and lots of decisions based on compromises. With compromises, it's always better to understand the choices.

Your tweeter does have a closed chamber behind it. In your cabinet, it sits in its own chamber sealed off from the rest of the speaker. The written description at Parts Express clearly says that, but there was no photo of the back of the tweeter.

Tang Band recommends a crossover frequency no lower than 1,800 Hz, double the Fs of 900 Hz. They didn't say what crossover slopes apply for that minimum frequency, but I'll guess it might be 4th order. At that frequency, the tweeter will produce about half the total SPL (equal to the woofer output), and one octave lower, at 900 Hz, the tweeter will produce sound about 24 dB lower SPL than at 1,800 Hz. With a 2nd order crossover, the tweeter will make sound about 12 dB lower at 900 Hz, possibly producing audible distortion, or worse.

Because you want to use that 10" woofer in a 2-way design, your choices are go with a different dome tweeter, such as the Dayton RS28 silk dome, or a tweeter with a wave guide or horn. The RS28 dome has an Fs under 600 Hz, and has been crossed over between 1200 and 1500 Hz with 4th order slopes. It's an excellent sounding low-distortion tweeter. There are other dome tweeters that are robust enough for such a low crossover point, but they are all much more expensive.

A wave guide mounted tweeter does allow lower crossover points, but like anything else it comes with some accompanying downsides. I have no experience building or designing speakers with these tweeters, so I can't comment intelligently about them. I do remember once reading about an audible resonance associated with the size of the perimeter of a wave guide, but I don't remember any details about that. Maybe someone else can talk about that. There are other arguable advantages and disadvantages with wave guide mounted tweeter designs. So you should remember that speaker building is a series of decisions based on compromises. With DIY, as you are learning, it really is Compromise It Yourself.

All designs are an exercise in compromise. Whether you are designing a bridge, a plane or a simple speaker.

Some compromises were made simply by the drivers chosen. It can still work to provide a nice speaker but ultimately I think that is just part of the learning experience.

Working with what you have right now, I would suggest experimenting with playing that TB tweeter lower. From the measurements I've seen it looks like it could go a bit lower. With that said, it has been years since I've used a non-compression driver in a design. The old rule of thumb was always 2x the Fs for domes but I've seen that rule broken successfully. The guys over at Parts Express Tech Talk or HTGuide are more familiar with these design types so you could post over there.

I'm personally biased towards compression drivers on horns. The biggest downside is the price. Before Erich and his efforts it was hard to find a good CD for less than $100 and you still needed to find a quality horn. Now you can get good horn/CD combos for under $100 which is still a good amount more than a $30 dome. There just isn't anyway a dome will give the uncompressed dynamics and directivity control of a horn. Of course you want to use a quality horn and CD.

I think you can make a nice speaker if you can get the tweeter down to around 1800hz (and maybe give 1500hz a shot). You could also buy some inexpensive horn/CD combos and try those. The other option would be to find a nice 6 or 8" option and cross in the 1800hz range.

I don't remember if you mentioned, but do you have something like a Dayton USB mic and REW or HolmImpulse or the OmniMic setup? You really need to be able to take good measurements. From there you can model just about anything.