Revel M106 Bookshelf Speaker Review

[jhaider]

Seriously, this cannot be good research. Think about it.

Why? You clearly haven't read said research. Your criticism boils down to "it is not consistent with my ideology, so it must be wrong."

 

[daftcombo]

Yes, yes. Please spare me your sarcasm and nitpicking which aims for your sole amusement.
Please be so kind and elaborate how Revel is extracting emotion...

Why should I endorse a marketing sentence that I didn't even quote?

 

[JustIntonation]

Why? You clearly haven't read said research. Your criticism boils down to "it is not consistent with my ideology, so it must be wrong."

I haven't other than what Toole wrote on his research on absorption vs bare sidewalls (which has to be one of the poorest research Toole has ever done, I don't hold any value to it whatsoever. And I hold other research by Toole in very high regard).
But I don't have to have read it to already know the conclusions I'm reading here are wrong. Both in logic and in my personal experience.

But common, even Amir makes statements here like with speaker A you may get better results if you absorb the floor or ceiling reflections and with speaker B this is not needed, etc.
Of course it can be the case for many reasons that speaker A will get a higher preference than speaker B in room 1 with placement X in a blind listening test, while in room 2 with placement Y speaker B will give a higher preference than speaker A in a blind listening test.
To deny the above is just absolute nonsense, no matter what research you quote that is supposed to say this.

 

[BYRTT]

Great thanks to Amir all the nice data be it objective or subjective ... it seems tons of discussions evolved lately so maybe below synthetic simulations could be of interest for some...


Here is M106 load impedance data:

Animation of simulated power to above system impedance data of M106, fullrange, or varius high pass filtered to cooperate with subs, simulation use 20 Amir RMS volts and crest factor is 3,0dB:

In relation below is animation that toggle Amirs two nice HD plots:

 

[Andreas007]

Why should I endorse a marketing sentence that I didn't even quote?

You know exactly what I want to say.
Now and then, everyone uses unscientific words do describe an experience or a feeling. Even Amir does.
So why making fun of a statement that was given in an interview we don't know the context anymore?
If you exchange "natural" with "realistic" you wouldn't have had a problem in the first place.

 

[Koeitje]

Don't compromise...just get a Rythmik. Plant optional.

Please point me towards an EU dealer so I don't have to pay crazy amounts for shipping and import costs.

 

[KaiserSoze]

Waveguides are highly variable. There are good and bad ones, so I don't think it makes sense to make a general rule like this.

Examples: the JBL's don't sound dark and they use a waveguide. The Buchardt S400 produces a huge soundstage and it uses a waveguide.

Besides: Olive's research showed that the room doesn't change what speaker we prefer. A bad room makes speakers sound worse, and a good room makes speakers sound better. It's not like it will somehow make someone's favorite speaker sound better and not any of the others under test.

As you suggest it is probably not entirely appropriate to generalize too strongly, but I will say that more often than not with waveguides I find it necessary to find a listening spot that is well off to the side for both speakers in order for the high-frequency level not to be overtly unnatural and irritating. The puzzling part of this experience is that in theory it should not be this way, not as long as the on-axis response is flat. Rather, what I should have experienced is a natural sound on-axis and a dull, treble-diminished off-axis response. On some occasions I have had this alternate experience, i.e., a speaker with a waveguide and where the high frequencies were not greatly exaggerated in the on-axis response. But this has been the minority experience for me, by far. Not all Infinity speakers I've heard since the early 90s seemed to have exaggerated treble in the on-axis response, but the majority certainly did, and many to the point that it was beyond glaring. For lack of any other explanation I assumed that either intentionally or unintentionally the power response was nearly flat, which means, considering the overall increase in directivity from low to high frequency, that in the on-axis response the high frequencies will be over-emphasized. (Essentially the same as what Audyssey did back when it first arrived - I don't know if it is still this way.) It goes without saying that in order for the on-axis response to be flat in a speaker with "controlled directivity", the power response must necessarily decline steadily from low to high frequency.

A pivotal assertion that has been made is that a room that is bad for any one speaker is bad for all speakers. Being a true scientist at heart and thus a natural-born skeptic, I am inclined not to accept this claim at face value. Partly because it is trivial for me to think of examples where it is highly unlikely to be true. Manifestly, a small room where speakers are necessarily placed close to the side walls will benefit a speaker that has comparatively wide dispersion at high frequency and an on-axis response where high frequencies are too low in level. This type of room will not provide the same benefit to a speaker with a flat on-axis response and tighter directivity. Another example: a room with a low ceiling and a smooth, hard floor will be detrimental to a speaker with a very irregular polar response in the vertical plane, but will be innocuous for a speaker with an exceptionally good polar response in the vertical plane. Conversely, a room with a high ceiling, acoustic tiles, and a thick carpet will seem to make the speaker with the poor polar response in the vertical plane sound better, but will have comparatively little perceived benefit for a speaker with an excellent polar response in the vertical plane. These examples are so obvious that I just don't see how anyone could argue to the contrary without doing so in a less than genuine manner.

It thus occurs to me that the greater generalization is not the generalization that all waveguides have a certain kind of sound, but is more likely the generalization that a room that is beneficial to one speaker will be beneficial to all speakers and that a room that is detrimental to one speaker will be detrimental to all speakers.

 

[JustIntonation]

As you suggest it is probably not entirely appropriate to generalize too strongly, but I will say that more often than not with waveguides I find it necessary to find a listening spot that is well off to the side for both speakers in order for the high-frequency level not to be overtly unnatural and irritating. The puzzling part of this experience is that in theory it should not be this way, not as long as the on-axis response is flat. Rather, what I should have experienced is a natural sound on-axis and a dull, treble-diminished off-axis response. On some occasions I have had this alternate experience, i.e., a speaker with a waveguide and where the high frequencies were not greatly exaggerated in the on-axis response. But this has been the minority experience for me, by far. Not all Infinity speakers I've heard since the early 90s seemed to have exaggerated treble in the on-axis response, but the majority certainly did, and many to the point that it was beyond glaring. For lack of any other explanation I assumed that either intentionally or unintentionally the power response was nearly flat, which means, considering the overall increase in directivity from low to high frequency, that in the on-axis response the high frequencies will be over-emphasized. (Essentially the same as what Audyssey did back when it first arrived - I don't know if it is still this way.) It goes without saying that in order for the on-axis response to be flat in a speaker with "controlled directivity", the power response must necessarily decline steadily from low to high frequency.

A pivotal assertion that has been made is that a room that is bad for any one speaker is bad for all speakers. Being a true scientist at heart and thus a natural-born skeptic, I am inclined not to accept this claim at face value. Partly because it is trivial for me to think of examples where it is highly unlikely to be untrue. Manifestly, a small room where speakers are necessarily placed close to the side walls will benefit a speaker that has comparatively wide dispersion at high frequency and an on-axis response where high frequencies are too low in level. This type of room will not provide the same benefit to a speaker with a flat on-axis response and tighter directivity. Another example: a room with a low ceiling and a smooth, hard floor will be detrimental to a speaker with a very irregular polar response in the vertical plane, but will be innocuous for a speaker with an exceptionally good polar response in the vertical plane. Conversely, a room with a high ceiling, acoustic tiles, and a thick carpet will seem to make the speaker with the poor polar response in the vertical plane sound better, but will have comparatively little perceived benefit for a speaker with an excellent polar response in the vertical plane. These examples are so obvious that I just don't see how anyone could argue to the contrary without doing so in a less than genuine manner.

It thus occurs to me that the greater generalization is not the generalization that all waveguides have a certain kind of sound, but is more likely the generalization that a room that is beneficial to one speaker will be beneficial to all speakers and that a room that is detrimental to one speaker will be detrimental to all speakers.

Agreed! And well written.
I will add that it may be easier to design a speaker with a smooth off-axis with a waveguide / horn, and many speakers designs that do not have a waveguide will have trouble achieving a smooth off-axis as can be seen in general in the measurements of various speakers by Amir.
I'm currently building a hi-end 3-way active speaker which breaks this trend and provides a very smooth off-axis (smoother than anything tested up till now) including vertical, while having very little off-axis dropoff. I'll try to get one over to Amir for testing and review in due time and before that will share anechoic chamber measurements.

 

[Chromatischism]

Good points and well stated. Have you heard the Buchardt S400? This speaker fits your first description and is ideal for a small room (in fact it was designed for that purpose).

 

[KaiserSoze]

Agreed! And well written.
I will add that it may be easier to design a speaker with a smooth off-axis with a waveguide / horn, and many speakers designs that do not have a waveguide will have trouble achieving a smooth off-axis as can be seen in general in the measurements of various speakers by Amir.
I'm currently building a hi-end 3-way active speaker which breaks this trend and provides a very smooth off-axis (smoother than anything tested up till now) including vertical, while having very little off-axis dropoff. I'll try to get one over to Amir for testing and review in due time and before that will share anechoic chamber measurements.

I agree with all you said. Many times I have bitten my tongue, so to speak, to keep from saying that waveguides are really just a sort of kludge for correcting a problem that is mostly specific to two-way speakers. Whenever I see large, expensive two-way speakers like some JBLs for one example, it always occurs to me that it would have been simpler to have made it a three-way speaker. Presumably the designers know what they are doing and have reached the conclusion that the best way to achieve the implied goals of the three-way speaker is actually by making a two-way speaker with a big waveguide like the ones JBL uses. I have to give them the benefit of the doubt, but personally I still prefer a good three-way speaker with less of the waveguide. I also strongly prefer speakers where the midrange (or woofer) and tweeter are vertically spaced as closely as possible, in order that the main lobe (in the vertical axis) will have good extension above and below the horizontal plane. The vertical spacing between the tweeter and the next-bigger driver is compromised not only by the big waveguide, but also by the size of the next-bigger driver itself. I just don't see any way that the three-way design doesn't make more sense. I'm looking forward to seeing the test results of your design. I hope it puts all these two-way bookshelf speakers with waveguides to shame both in the directivity index and in the distortion level at and below 100 Hz. I'm cheering for you!

 

[JustIntonation]

I agree with all you said. Many times I have bitten my tongue, so to speak, to keep from saying that waveguides are really just a sort of kludge for correcting a problem that is mostly specific to two-way speakers. Whenever I see large, expensive two-way speakers like some JBLs for one example, it always occurs to me that it would have been simpler to have made it a three-way speaker. Presumably the designers know what they are doing and have reached the conclusion that the best way to achieve the implied goals of the three-way speaker is actually by making a two-way speaker with a big waveguide like the ones JBL uses. I have to give them the benefit of the doubt, but personally I still prefer a good three-way speaker with less of the waveguide. I also strongly prefer speakers where the midrange (or woofer) and tweeter are vertically spaced as closely as possible, in order that the main lobe (in the vertical axis) will have good extension above and below the horizontal plane. The vertical spacing between the tweeter and the next-bigger driver is compromised not only by the big waveguide, but also by the size of the next-bigger driver itself. I just don't see any way that the three-way design doesn't make more sense. I'm looking forward to seeing the test results of your design. I hope it puts all these two-way bookshelf speakers with waveguides to shame both in the directivity index and in the distortion level at and below 100 Hz. I'm cheering for you!

Aah exactly! And thank you!
Yes, it will be with Bliesma T34A-4 tweeter (which has the off-axis response that is better than most 1" tweeters) crossed at 1300Hz with very close spaced SB Acoustics SB15CAC30-4 mid driver (CTC spacing is half wavelength of the crossover freq), and Seas L26ROY XM001-04 woofer (closed), on a wide baffle with 10" diameter roundovers. It should indeed put all two-way bookshelf speakers to shame Finally building it now, will share buildplans once finished.

 

[KaiserSoze]

Good points and well stated. Have you heard the Buchardt S400? This speaker fits your first description and is ideal for a small room (in fact it was designed for that purpose).

Can't say that I have. I just took a quick glance on the manufacturer's site, and I like the fact that they provide lots of charts and so forth. I wish more manufacturers would do the same. The horizontally off-axis response looks damn good out to maybe 60 degrees, but by the time you get to 90 degrees you have in effect a step response with a high level from low frequency up to about 1 kHz then a much lower level from there on up. But of course there aren't many speakers with a decent response curve that far off axis. What bothers me more is the off-axis response in the vertical plane, which is pretty much what is to be expected given the wide vertical separation between the tweeter and the center of the woofer. This is the price paid for achieving the nice horizontally off-axis response. It will be important with these speakers to sit at or very close to the horizontal plane in which the speaker is situated, i.e., at the same level as the speaker. A difference in sound quality may also be noticed for a room with a thick carpet and acoustic absorption treatment of the ceiling, vs. a hard wood floor and an untreated ceiling.

 

[Chromatischism]

Yep. Carpet is probably recommended. But that's probably true of most stand mount speakers. Still, it does really well in a narrow room with sidewalls nearby.

I'm intrigued by many three-way designs, but let's not pretend that's a panacea. Now you have two crossovers to contend with, potentially compounding the problem. And, the vertical space between the tweeter and the woofer is limited by having another driver in between them. The distance is probably going to be the same or even greater than a two-way.

 

[JustIntonation]

Yep. Carpet is probably recommended. But that's probably true of most stand mount speakers. Still, it does really well in a narrow room with sidewalls nearby.

I'm intrigued by many three-way designs, but let's not pretend that's a panacea. Now you have two crossovers to contend with, potentially compounding the problem. And, the vertical space between the tweeter and the woofer is limited by having another driver in between them. The distance is probably going to be the same or even greater than a two-way.

The crossover between mid and woofer is not a problematic one. They will naturally be close enough to not give any problems in the vertical off-axis.
And the distance between the tweeter and woofer in a 3-way is not important, they do not cross over to eachother and the ear isn't concerned about the woofer (unless crossed really high but why would one do that in a 3-way).
3-ways are often more expensive due to bigger box with seperate internal enclosures, twice the crossover components in passive designs or another amp in active ones, and 3 drivers instead of 2 ofcourse. But the benefit is much less intermodulation distortion, optimized CTC distance between mid and tweeter due to small mid driver and better directivity of the mid due to smaller diaphragm both resulting in better off-axis, and a woofer that's optimized to giving good bass.

 

[daftcombo]

You know exactly what I want to say.
Now and then, everyone uses unscientific words do describe an experience or a feeling. Even Amir does.
So why making fun of a statement that was given in an interview we don't know the context anymore?
If you exchange "natural" with "realistic" you wouldn't have had a problem in the first place.

I see that you don't have a lot of humor man. German perhaps?

 

[Chromatischism]

The crossover between mid and woofer is not a problematic one. They will naturally be close enough to not give any problems in the vertical off-axis.
And the distance between the tweeter and woofer in a 3-way is not important, they do not cross over to eachother and the ear isn't concerned about the woofer (unless crossed really high but why would one do that in a 3-way).
3-ways are often more expensive due to bigger box with seperate internal enclosures, twice the crossover components in passive designs or another amp in active ones, and 3 drivers instead of 2 ofcourse. But the benefit is much less intermodulation distortion, optimized CTC distance between mid and tweeter due to small mid driver and better directivity of the mid due to smaller diaphragm both resulting in better off-axis, and a woofer that's optimized to giving good bass.

My intention was to try a Philharmonic BMR. Now that it's down to DIY, it may be a bit longer, but someday I'll get around to that. Still, I wonder how good of a match that would be with its ultra wide dispersion in my small room.

Still, the downsides of a large waveguided 2-way are much smaller than the benefits, and they can be mitigated with, as discussed, carpet (and/or ceiling if you can swing it) absorption and placement. The S400 is otherwise not too picky about placement, and the sound doesn't change much with toe-in compared to other, narrower dispersion designs like KEF and Klipsch.

 

[KaiserSoze]

Aah exactly! And thank you!
Yes, it will be with Bliesma T34A-4 tweeter (which has the off-axis response that is better than most 1" tweeters) crossed at 1300Hz with very close spaced SB Acoustics SB15CAC30-4 mid driver (CTC spacing is half wavelength of the crossover freq), and Seas L26ROY XM001-04 woofer (closed), on a wide baffle with 10" diameter roundovers. It should indeed put all two-way bookshelf speakers to shame Finally building it now, will share buildplans once finished.

Sounds like you've got your ducks in a row. That tweeter is evidently something very special, although I haven't been to figure out exactly how they managed to make the directivity so much better than it is with any other dome of similar size. Vance Dickason tested it and seemed very impressed. His measurements of distortion only went as low as 2 kHz, however it was also tested by hificompass. If distortion were the only reason for not crossing it lower (i.e., you weren't using a midrange driver), you could cross it to a woofer well below 1 kHz, where the directivity of the woofer will likely be a good match to the tweeter without sticking the tweeter in a waveguide, which by the way suggests another alternative to the constant directivity / tweeter-in-waveguide approach. Troels Gravesen also tested it and wrote that he is looking forward to using it in a future design. It is a cost-no-object tweeter, but likely the high cost is better justified with this tweeter than with most all other expensive tweeters, most of which seem to offer very little to justify the higher cost.

One less-than-enthusiastic comment is that if you intend to make the big woofer sealed a.k.a. acoustic suspension, that while this is something I applaud vigorously, I do not think the Seas ROY is the most suitable large driver for an acoustic suspension application. Certainly it will give you clean bass for however low it will play, however Fs is not low enough for it to play particularly low in a sealed enclosure. Like 99% of modern off-the-shelf woofer and subwoofer drivers, it is intended for use in a ported enclosure augmented by a port or passive radiator, and will not give especially good results in a sealed enclosure. (Notwithstanding Seas's suggestions to the contrary, which are actually kind of shameful.) In recent weeks I spent some time on that online driver database, looking for a big driver suitable for sealed enclosure application. Following is the list I came up with. The numbers are all the manufacturers' numbers except the F3 number which I calculated using the manufacturers' numbers for Fs and Qts, and a nominal value for Qtc, which does not have much affect on F3 as it turns out. (From what I recall, my similar calculation of F3 for the Seas XM001-04 was around 50 or 55 Hz.)

(Sorry but the only font available that should have provided constant character spacing is Courier New, which I tried, and while it looked correct in the text entry box, it came out wrong in the preview.)

Brand/model Size Fs/Qts F3 Vas SPL Cms Bl Mms
(inch) (Hz) (Hz) (liter) (dB) (mm/N) (T-m) (gm)
Dayton Audio
DCS385-4 15 18/.39 33 311 92 .30 15 273

SB Acoustics
SB34NRX75-6 12 19/.40 34 260 90 .71 11 99

Scan-Speak
25W/8565-00 10 20/.41 35 225 88 1.5 8.2 43

Scan-Speak
32W/4878T05 13 17/.32 38 180 89 .52 13 165

Scan-Speak
30W/4558T00 12 17/.32 38 197 89 .65 11 135

Peerless XXLS-
300F50AL01-04 12 22/.41 39 113 85 .37 11 138

 

[KaiserSoze]

The crossover between mid and woofer is not a problematic one. They will naturally be close enough to not give any problems in the vertical off-axis.
And the distance between the tweeter and woofer in a 3-way is not important, they do not cross over to eachother and the ear isn't concerned about the woofer (unless crossed really high but why would one do that in a 3-way).
3-ways are often more expensive due to bigger box with seperate internal enclosures, twice the crossover components in passive designs or another amp in active ones, and 3 drivers instead of 2 ofcourse. But the benefit is much less intermodulation distortion, optimized CTC distance between mid and tweeter due to small mid driver and better directivity of the mid due to smaller diaphragm both resulting in better off-axis, and a woofer that's optimized to giving good bass.

Yes. You do have to design two good crossovers and in many cases the combined output from the woofer and the tweeter will be strong enough at the middle of the midrange driver's range that its output will need to be lower than the output of either the woofer or the tweeter. It depends on the steepness of the crossovers and the frequency distance between the two crossover points.

But the lower crossover is not generally any concern with respect to nulls in the vertical off-axis response because the wavelength for this crossover point will typically be more than two feet whereas the vertical spacing between the woofer and the midrange will be less than half that. Likewise there typically isn't much of a horizontal directivity mismatch at this crossover because the woofer still has very good dispersion at this point. The crossover between the midrange and the tweeter can potentially be more problematic because the wavelength is shorter, however in the net effect it is usually less problematic because (1.) the crossover point will likely not be much higher compared to a two-way design, (2.) the vertical spacing between the tweeter and the center of the midrange will be much less than the vertical spacing between the tweeter-in-big-waveguide and the larger woofer in a typical two-way speaker, (2.) the directivity mismatch is much less severe because the tweeter is crossing to a small midrange rather than to a big woofer. All in all the three-way design has strong inherent advantages, which is why I can hardly avoid thinking of waveguides as a sort of kludge needed to correct an inherent limitation of two-way speakers that doesn't exist with a well-designed three-way speaker.

And you also mentioned reduced intermodulation distortion, which was an oversight for me but that is certainly important.

 

[JustIntonation]

Sounds like you've got your ducks in a row. That tweeter is evidently something very special, although I haven't been to figure out exactly how they managed to make the directivity so much better than it is with any other dome of similar size. Vance Dickason tested it and seemed very impressed. His measurements of distortion only went as low as 2 kHz, however it was also tested by hificompass. If distortion were the only reason for not crossing it lower (i.e., you weren't using a midrange driver), you could cross it to a woofer well below 1 kHz, where the directivity of the woofer will likely be a good match to the tweeter without sticking the tweeter in a waveguide, which by the way suggests another alternative to the constant directivity / tweeter-in-waveguide approach. Troels Gravesen also tested it and wrote that he is looking forward to using it in a future design. It is a cost-no-object tweeter, but likely the high cost is better justified with this tweeter than with most all other expensive tweeters, most of which seem to offer very little to justify the higher cost.

One less-than-enthusiastic comment is that if you intend to make the big woofer sealed a.k.a. acoustic suspension, that while this is something I applaud vigorously, I do not think the Seas ROY is the most suitable large driver for an acoustic suspension application. Certainly it will give you clean bass for however low it will play, however Fs is not low enough for it to play particularly low in a sealed enclosure. Like 99% of modern off-the-shelf woofer and subwoofer drivers, it is intended for use in a ported enclosure augmented by a port or passive radiator, and will not give especially good results in a sealed enclosure. (Notwithstanding Seas's suggestions to the contrary, which are actually kind of shameful.) In recent weeks I spent some time on that online driver database, looking for a big driver suitable for sealed enclosure application. Following is the list I came up with. The numbers are all the manufacturers' numbers except the F3 number which I calculated using the manufacturers' numbers for Fs and Qts, and a nominal value for Qtc, which does not have much affect on F3 as it turns out. (From what I recall, my similar calculation of F3 for the Seas XM001-04 was around 50 or 55 Hz.)

(Sorry but the only font available that should have provided constant character spacing is Courier New, which I tried, and while it looked correct in the text entry box, it came out wrong in the preview.)

Brand/model Size Fs/Qts F3 Vas SPL Cms Bl Mms
(inch) (Hz) (Hz) (liter) (dB) (mm/N) (T-m) (gm)
Dayton Audio
DCS385-4 15 18/.39 33 311 92 .30 15 273

SB Acoustics
SB34NRX75-6 12 19/.40 34 260 90 .71 11 99

Scan-Speak
25W/8565-00 10 20/.41 35 225 88 1.5 8.2 43

Scan-Speak
32W/4878T05 13 17/.32 38 180 89 .52 13 165

Scan-Speak
30W/4558T00 12 17/.32 38 197 89 .65 11 135

Peerless XXLS-
300F50AL01-04 12 22/.41 39 113 85 .37 11 138

I'll be using the Hypex Fusion plate amps in my design so I'm not limited by the natural Fs in my design as the Fusion amps have built in DSP I'll be using a Linkwitz transform to lower the -3dB point. My design is primarily for a nearfield / midfield monitor, it will provide enough SPL for me even after the Linkwitz transform.
With this in mind I think the new XM001-04 L26ROY is just about the perfect woofer for closed appication. What is important in the woofer is the SD and the xmax ofcourse but also especially how linear the motor force is under excursion (so how clean it stays under excursion). The new L26ROY excells in this. It will really give clean bass up till right about its limit also under Linkwitz transform in a closed box.
I have the woofers here already and tested them and they are amazing, truly clean and deep and in control. Had looked into other woofers before that but they dissapointed.
Btw here are Klippel test results of the new L26ROY: https://audioxpress.com/article/test-bench-the-l26roy-10-subwoofer-from-seas They are truly good.
Also the L26ROY is not just a sub, it can be used as a woofer as it stays clean enough and doesn't have fr problems >100Hz or too high a inductance (which modulates with excursion) like many subwoofers. I plan on crossing mine at about 180Hz still soubting between LR4 or LR6 (will determine final crossover freq and steepness from testing).
If the driver you're looking for is for sub duties only then many more drivers become available, but for my application I don't think there are many other options which are as good as the new L26ROY.

About the T34A-4, the way he got the off-axis so amazingly good is by the geometry of the tweeter, it is extremely curved / round. Though I think the real magic was how can such a large tweeter behave so well up to 20kHz, this was achieved by making it very light and unequal thickness across the membrane this is really a tweeter from the future It is a bit expensive (I paid about €265 per tweeter inc shipping) but indeed as you say not crazy for the things it does (I think it's even cheap for that actually haha).
Btw, don't mind Dickason's measurements of the T34A-4 he did them badly I think under not good circumstances and his measurements show that, Troels and Hificompass made much more clean measurements.
I also have already tested the tweeters myself and they sound jaw dropping amazing The transparancy and dynamics are the best I've personally heard and that was only on a test baffle yet which still had diffraction.
One other thing which is not often looked at in speaker design is that our ears are so much more sensitive between 2-5kHz and I've found that any crossover effects in this range sound the worst, they make for harsh sounds and destroy transparancy for me and just totally mess up the whole smoothness and balance of the treble to me. With the T34A-4 one can cross low enough that all crossover effect are not just very far off-axis but also stay below 2kHz. Our ears have a sensitivity dip between 1-2kHz and I've tested the effects of this and the crossover dip is just so much less noticeable between 1-2kHz and doesn't make the treble sound bad (I find it mostly makes things sound a bit further away which isn't that bad a thing). In my personal view of things the treble "sits" on the very important 2-3kHz range, I personally call that the lower trebble. I make sure that part is completely flat and smooth also off-axis. Completely unaffected by the crossover or mid-driver dropoff or by diffraction. I have heard the effect of this only in simulations on the computer (where I run audio through it and listen with headphones to the sim), to hear it for real I have to finish my speakers

Edit: oh btw, about intermodulation distortion. Since there are almost no measurements of this one must go by the standard HD measurements. There are two general rules here. If the driver has a rise in its fr this will also amplify IMD products, and one can deduce IMD amount by the relation between the 2nd harmonic and 3rd and 5th harmonics. It is especially important that the 3rd and 5th harmonics are very low. not so much for the 2nd harmonic. So the T34A-4 may not look to be too impressive with its 2nd harmonic level, but that is inaudible in reality what's important is that its 3rd and 5th harmonics are the lowest there are, and that it's cone breakup / rise is above the audible range.
And for the mid driver this is where it most often goes wrong. Most mid drivers will have levels of 3rd and 5th harmonic which are not particularly low and rising a bit till the crossover point and then they often have a beakup mode or rise close to 5kHz. Even though you cross the driver below 5kHz this does not matter at all for the HD and IMD products falling around 5kHz they get amplified all the same. The SB15CAC30-4 is simply amazing in this, it has extremely low 3rd and 5th harmonic, and a small drop in sensitivity around 2-5kHz no rise at all. The result is the cleanest mid driver I've ever heard, also extremely linear. On a wide baffle where the baffle step occurs below where you cross the mid driver it can go plenty loud too / plenty of dynamic range. I think it's a steal for the price when used this way I'm so happy with it. (for a higher more normal crossover point it's perhaps not the best, it has a bit of a dip at around 2.6kHz in the off-axis for instance, etc. but that's all not a problem when crossing it at 1.3kHz with the T34A-4)
Btw, the SB15CAC30-4 has it's main cone breakup / rise in the 9-10kHz range. This is where we find another dip in the ears sensitivity / equal loudness curve, and it's far enough away from the crossover point so that hardly any distortion product fall on that anyway so this is really the most perfect metal mid-driver there is in my opinion It's like it's psychoacoustically perfectly made for the human ear.

 

[Chromatischism]

One other thing which is not often looked at in speaker design is that our ears are so much more sensitive between 2-5kHz and I've found that any crossover effects in this range sound the worst, they make for harsh sounds and destroy transparancy for me and just totally mess up the whole smoothness and balance of the treble to me. With the T34A-4 one can cross low enough that all crossover effect are not just very far off-axis but also stay below 2kHz. Our ears have a sensitivity dip between 1-2kHz and I've tested the effects of this and the crossover dip is just so much less noticeable between 1-2kHz and doesn't make the treble sound bad (I find it mostly makes things sound a bit further away which isn't that bad a thing). In my personal view of things the treble "sits" on the very important 2-3kHz range, I personally call that the lower trebble. I make sure that part is completely flat and smooth also off-axis. Completely unaffected by the crossover or mid-driver dropoff or by diffraction. I have heard the effect of this only in simulations on the computer, to hear it for real I have to finish my speakers

Edit: oh btw, about intermodulation distortion. Since there are almost no measurements of this one must go by the standard HD measurements. There are two general rules here. If the driver has a rise in its fr this will also amplify IMD products, and one can deduce IMD amount by the relation between the 2nd harmonic and 3rd and 5th harmonics. It is especially important that the 3rd and 5th harmonics are very low. not so much for the 2nd harmonic. So the T34A-4 may not look to be too impressive with its 2nd harmonic level, but that is inaudible in reality what's important is that its 3rd and 5th harmonics are the lowest there are, and that it's cone breakup / rise is above the audible range.
And for the mid driver this is where it most often goes wrong. Most mid drivers will have levels of 3rd and 5th harmonic which are not particularly low and rising a bit till the crossover point and then they often have a beakup mode or rise close to 5kHz. Even though you cross the driver below 5kHz this does not matter at all for the HD and IMD products falling around 5kHz they get amplified all the same. The SB15CAC30-4 is simply amazing in this, it has extremely low 3rd and 5th harmonic, and a small drop in sensitivity around 2-5kHz no rise at all. The result is the cleanest mid driver I've ever heard, also extremely linear. On a wide baffle where the baffle step occurs below where you cross the mid driver it can go plenty loud too / plenty of dynamic range. I think it's a steal for the price when used this way I'm so happy with it. (for a higher more normal crossover point it's perhaps not the best, it has a bit of a dip at around 2.6kHz in the off-axis for instance, etc. but that's all not a problem when crossing it at 1.3kHz with the T34A-4)

I actually think speaker designers think about this a lot, and is why a dip at the crossover around 2-3 kHz off-axis isn't terrible, as in many rooms it won't be very audible.