Differences between tweeter designs?

[JustIntonation]

@JustIntonation do you think you could recommend some recorded music that uses just intonation? Thanks

No not any polyphonic music specifically made for this. There's a lot of music made in wrong "just intonation" though.
But of course a lot of the music you already have will be close to correct just intonation. Even a whole orchestra does not play in 12-tone equal temperament. Most instruments can influence their pitch to some degree and some instruments can do it freely so as good as the performer can do it. Then there are string quartets, trombone quartets, choirs, etc who are all free in their pitch. There are differences in how they're schooled though, I believe for instance for string players the Russian schools teach correctly in my opinion and the Viennese(?) do it wrongly but I'm not an expert on this.
So in reality there are various degrees of correctness in this in practice. I analyzed a lot of string and trombone quartets and choirs with Melodyne DNA when it came out with varying degrees of success in analyzing the pitches of individual voices in polyphonic music. And basically each one made their own unique mess though some worse than others (though none as bad as the choirs! luckily choirs have the most forgiving pitch accuracy for our ears). I can now recognize with string quartets the sound when they're going for 5-limit chords and I don't like it. But hard to describe the sound. And can't really give a recomendation as to which string quartets have the best intonation out of the top of my head as I rarely listen to them (Do like string quartets when they're playing along with the piano in pieces by Faure and Vierne, but then the piano is 12-tet ofcourse and the strings are mostly going along with this)

 

[JustIntonation]

Finished listening to the 2kHz crossover files.
Did it with ABX comparator for foobar2000.
https://www.foobar2000.org/components/view/foo_abx
Did the listening with headphones (DT1990pro) and the LSR305P mkII (which I still haven't sent back, pushing it till tomorrow now )
So the listening is not really high resolution / quality..

I can tell that with these files I can hear the effect of the crossover better with the 12dB/oct slope of LR2. So that's the worst of the bunch.
I could reliably tell it with the "A18" file. (I did it quickly and manage to miss 2 out of 8 the first time around which surprised me and I then did it more attentively and got 8 out of 8 right.) But the LR2 A18 is basically the lower limit I could reliably tell on my current listening systems.
With the high 96dB/oct slope of the LR16 I thought that the A27 was at the limit of my hearing but I took that one twice as well and scored 4/4 first time and 5/8 second time so that's why ABX is handy, I was basically guessing and convincing myself it was at the limit of my hearing. I was able to clearly hear the effect of the crossover at LR16 A54 though which would not be acceptable to me in real use, so somewhere only slightly wider than A27 is the limit for me for LR16.

So for 2kHz crossover, the limit is something like inbetween A9 and A18 for LR2 and slightly above A27 for LR16.
If the cancellation angle for 2kHz would be at 90 degrees from on-axis, this would mean about 90/100*13*2 listening angle for LR2 and 90/100*32*2 listening angle for LR16. Which for a listening distance of 1m would be 2*Pi/360*23.4 = 0.4 meter of vertical listening distance before hearing the LR2 crossover, and 2Pi/360*57.6 = 1.30 meter vertical listening distance before hearing the LR16 crossover.
Well that sounds like a bit of overkill for my future dead room especially in the case of the LR16 crossover

I'll do some more testing with different crossover frequencies, 1600, 2500 and 3100 or so. But for now the ringing of the LR16 is not as bad as I thought it would be and certainly looks like a less audible thing than the bigger phasing bandwidth of the lower slope filters!

Here some of my ABX files for fun:
foo_abx 2.0.5 report
foobar2000 v1.4
2018-09-24 08:14:27

File A: co-sim LR16 A0.wav
SHA1: b7c096e9f0df8b0581964f56337216e50a959f7b
File B: co-sim LR16 A54.wav
SHA1: 46bdfa62fff3e4195ccee982015e021de96be5c6

Output:
ASIO : AUSBAudio ASIO Driver
Crossfading: NO

08:14:27 : Test started.
08:15:17 : 01/01
08:15:27 : 02/02
08:15:37 : 03/03
08:15:47 : 04/04
08:15:55 : 05/05
08:16:03 : 06/06
08:16:11 : 07/07
08:16:20 : 08/08
08:16:20 : Test finished.

----------
Total: 8/8
Probability that you were guessing: 0.4%

-- signature --
449e2eeb044d3c53dfa6f5aebfbf5616cdb71685

foo_abx 2.0.5 report
foobar2000 v1.4
2018-09-24 08:17:27

File A: co-sim LR16 A0.wav
SHA1: b7c096e9f0df8b0581964f56337216e50a959f7b
File B: co-sim LR16 A27.wav
SHA1: aa68af4cdc0fc0fa9de098fd332a7bd06b506727

Output:
ASIO : AUSBAudio ASIO Driver
Crossfading: NO

08:17:27 : Test started.
08:18:00 : 01/01
08:18:29 : 02/02
08:19:16 : 03/03
08:20:05 : 04/04
08:20:55 : 04/05
08:21:04 : 04/06
08:21:21 : 04/07
08:22:16 : 04/08
08:22:16 : Test finished.

----------
Total: 4/8
Probability that you were guessing: 63.7%

-- signature --
5c5bfa2f29addc056601449cc0503d57a51f2d68

foo_abx 2.0.5 report
foobar2000 v1.4
2018-09-24 07:41:58

File A: co-sim LR2 A0.wav
SHA1: 1dddb6225df1bf4f05526facd7c6e0ec1a3a3db3
File B: co-sim LR2 A27.wav
SHA1: 1b6ad117984817d264da4fae82cebbc027d89007

Output:
ASIO : AUSBAudio ASIO Driver
Crossfading: NO

07:41:58 : Test started.
07:42:43 : 01/01
07:45:10 : 02/02
07:45:40 : 03/03
07:46:07 : 04/04
07:46:23 : 05/05
07:46:41 : 06/06
07:47:01 : 07/07
07:47:27 : 08/08
07:47:27 : Test finished.

----------
Total: 8/8
Probability that you were guessing: 0.4%

-- signature --
858f74f8570eb2d9ab9968c92b34d61b0c03c5af


foo_abx 2.0.5 report
foobar2000 v1.4
2018-09-24 08:37:56

File A: co-sim LR2 A0.wav
SHA1: 1dddb6225df1bf4f05526facd7c6e0ec1a3a3db3
File B: co-sim LR2 A18.wav
SHA1: fe02cacaad517041a2d1cc2c4a2e298ab5b0f4e2

Output:
ASIO : AUSBAudio ASIO Driver
Crossfading: NO

08:37:56 : Test started.
08:38:30 : 01/01
08:38:57 : 02/02
08:39:26 : 03/03
08:39:49 : 04/04
08:40:09 : 05/05
08:40:35 : 06/06
08:40:53 : 07/07
08:41:14 : 08/08
08:41:14 : Test finished.

----------
Total: 8/8
Probability that you were guessing: 0.4%

-- signature --
afdaca816ab5d9e340a6de743ae8e5a639dc4b6d

 

[JustIntonation]

Oh the idea is once this is finished I can start looking at drivers and distances that match the minimal crossover requirements. And then model mid-driver droppoff at the crossover freq and do some more listening tests and then finally able to select a middriver and tweeter based on these specs and online distortion measurements etc.
Sounds like a good plan right? All from the comfort of sitting at home lol.

 

[JustIntonation]

Oh and one more thing. To put this in perspective how different speakers can behave at this.
My previous K+H O300 had a crossover point of 3300Hz and I'm guessing around 12-13cm? distance between mid and tweeter.
It would have had a canclellation axis at roughly 38 degrees from on-axis. So the vertical (total up and down) listening window would be about 17 cm at LR2. But it had LR4 filters so that adds a bit of extra listening window, let say 24cm rough guess. So moving ones head up 12cm up or down at 1m listening distance would make the crossover borderline audible.. That is for a Linkwitz-Riley crossover, for a Butterworth crossover it would be far less (like only a few cm total window).
Though if it had a Linkwitz-Riley that window actually isn't that bad I'm realizing as I'm typing this.. Am I wrong in thinking it was the crossover phasing that bothered me and it was a floor reflection or something else all together? Hmm.. likely. Maybe I'm testing this crossover thing for no good reason.. Other than to clear myself from wrong guesses.
Hmm looking at the vertical directivity plot:
http://www.neumann-kh-line.com/klei..._dir_510.gif/$File/o300d_vertical_dir_510.gif
It seems like I'm spot on with 38 degrees for the cancellation degree above, the one below is less clear. I though see though that only 5 degrees or so below on axis there is already a 3db dip at around 4k which is quickly 3dB up again higher in the freq range.
I wasn't that careful yet back then, maybe I didn't have them set up precisely enough and was listening 20 degrees too low or so, which looks very rough on that plot. Though can't imagine I set them up that much off.. So hmm it looks indeed like it was something else and not the crossover. So I'm probably being afraid of doing the wrong thing wrong which isn't much of an issue within reasonable limits..

 

[andreasmaaan]

FWIW, I measured a Neumann KH120 at one point and found the crossover to be lobing at different angles to those shown by Neumann in their data (and not for the better). I might have been unlucky, but it’s at least possible that the published graph you’re looking at there doesn’t reflect how that particular speaker actually measured...

 

[JustIntonation]

FWIW, I measured a Neumann KH120 at one point and found the crossover to be lobing at different angles to those shown by Neumann in their data (and not for the better). I might have been unlucky, but it’s at least possible that the published graph you’re looking at there doesn’t reflect how that particular speaker actually measured...

That is indeed yet another posibility..
Ah well, there was enough wrong with that speaker (much of it improved in the O310 though from what I read) and with my experience in setting it up properly. I never once measured it nor my room. Didn't start measuring untill I built my own single driver speaker (which was another bad decision for the purpose I had in mind but learned some again).

Anyhow, giving myself this week to figure out drivers and then it's time to start building both the studio and speakers
Will probably be nothing exotic, treated paper cone drivers from Seas and or Scanspeak. Hoping to build something that approaches detail and neutrality from the "musical"/good sounding side. ATC manages to do this (reportedly even in their smaller models without their midrange dome) and I'll have Hypex and DSP Should all work out fine can tweak to my desire once built..

 

[Deleted member 65]

Stereophile reviewing Legacy speaker with AMT tweeters, description of AMT's qualities reflecting mine. Go listen before deciding would be my advice.

" When I switched from my Harbeth M30.2s (domes) to the Legacy Studio HDs (folded ribbons), I noticed right away how much the Legacy's AMT tweeters affected their overall sound character, even in the bass. The Studio HDs played music with a supple dynamic quality. High frequencies were clear, focused, and airy. "

" In its anechoic flat position, the Legacys generated grainless, breathy top octaves whose benign quality was the opposite of noticeable, annoying, or fatiguing. I have rarely heard massed strings, cymbals, or the top 20 keys of a piano sound more natural. "

Read more at https://www.stereophile.com/content/legacy-audio-studio-hd-loudspeaker#BmSmqf3ZSpyAs7Qr.99

 

[JustIntonation]

Stereophile reviewing Legacy speaker with AMT tweeters, description of AMT's qualities reflecting mine. Go listen before deciding would be my advice.

" When I switched from my Harbeth M30.2s (domes) to the Legacy Studio HDs (folded ribbons), I noticed right away how much the Legacy's AMT tweeters affected their overall sound character, even in the bass. The Studio HDs played music with a supple dynamic quality. High frequencies were clear, focused, and airy. "

" In its anechoic flat position, the Legacys generated grainless, breathy top octaves whose benign quality was the opposite of noticeable, annoying, or fatiguing. I have rarely heard massed strings, cymbals, or the top 20 keys of a piano sound more natural. "

Read more at https://www.stereophile.com/content/legacy-audio-studio-hd-loudspeaker#BmSmqf3ZSpyAs7Qr.99

Thanks! Still looking at tweeters and indeed including AMTs, especially the Mundorf AMT21. Though not puting much weight on user reviews as surely most of the difference with domes in sound will be due to different vertical dispersion pattern where the AMTs are very narrow especially the bigger ones significantly reducing ceiling and floor reflections in normal rooms (which my room won't be).
Also looking at mid drivers and woofers but it seems that maybe this is somewhat easier and Volt drivers will be the go to for me.
Adding up though.. going to be expensive speakers. Though should give me pleasure for many many years.

 

[March Audio]

I don't think the 23 is bad at all in terms of dispersion. You frequently see things like the Scanspeak below. Note the different angles in the measurements, you have to guesstimate I'm afraid. Mo
St domes are falling off below 10khz, the 23 doesnt

23

Scanspeak dome

Very expensive Scanspeak revelator

Seas millenium

 

[andreasmaaan]

Which Volt drivers are you looking at?

 

[dc655321]

I don't think the 23 is bad at all in terms of dispersion. You frequently see things like the Scanspeak below. Note the different angles in the measurements, you have to guesstimate I'm afraid. Mo
St domes are falling off below 10khz, the 23 doesnt

23
View attachment 15964
View attachment 15965

Scanspeak dome
View attachment 15966

Very expensive Scanspeak revelatorView attachment 15967

Seas millenium
View attachment 15968

I would be interested to see the AMT plots produced over the same width as the Seas (i.e. up to 60 deg, not 30 deg). Hopefully the response doesn't fall off a cliff at 31 deg

The manufacturer provided plots here (AMT) have a bit of a over-processed appearance, so it would also be great to see a real-world measurement. Anyone recommend any pointers to full(er)-range directivity info for the AMTs?

 

[andreasmaaan]

The other thing to keep in mind looking at those graphs is that the AMTs are plotted at 10dB vertical increments vs 5dB for the domes.

Still, their off-axis performance is clearly quite impressive.

 

[Deleted member 65]

Well, my real world experience of AMT's in my two home setups and the one in our summerhouse setup says vertical dispersion is not a problem,
but in order to comply with the intent of ASR here's what I'll do in the next few days (when I'm not occupied with mountainbiking).

1. Pick out my Umik & install it on a camerapod and measure with REW on one of my home setups, say from 2,5 - 20 KHz.
2. Move Umik up & down and take three/four measurements, not sure I can give the degree, +-mm from on axis is easier for me.

 

[maverickronin]

The other thing to keep in mind looking at those graphs is that the AMTs are plotted at 10dB vertical increments vs 5dB for the domes.

Also the AMTs are only at 0, 5, and 15 degrees. The domes are unstated, but isn't the usual 0, 30, and 60?

 

[March Audio]

I would be interested to see the AMT plots produced over the same width as the Seas (i.e. up to 60 deg, not 30 deg). Hopefully the response doesn't fall off a cliff at 31 deg

The manufacturer provided plots here (AMT) have a bit of a over-processed appearance, so it would also be great to see a real-world measurement. Anyone recommend any pointers to full(er)-range directivity info for the AMTs?

I haven't performed proper off axis measurements of my mats but anecdotally I don't find any sense of beaming, you can move around without significant change in high frequency sound.

 

[March Audio]

Well, my real world experience of AMT's in my two home setups and the one in our summerhouse setup says vertical dispersion is not a problem,
but in order to comply with the intent of ASR here's what I'll do in the next few days (when I'm not occupied with mountainbiking).

1. Pick out my Umik & install it on a camerapod and measure with REW on one of my home setups, say from 2,5 - 20 KHz.
2. Move Umik up & down and take three/four measurements, not sure I can give the degree, +-mm from on axis is easier for me.

I will try and do the same

 

[March Audio]

The other thing to keep in mind looking at those graphs is that the AMTs are plotted at 10dB vertical increments vs 5dB for the domes.

Still, their off-axis performance is clearly quite impressive.

It was one of the intents in my speaker design, wide dispersion. Looked for the same in the mid.
Morel 1308 to 4.6khz where the AMT takes over

 

[Sal1950]

" When I switched from my Harbeth M30.2s (domes) to the Legacy Studio HDs (folded ribbons), I noticed right away how much the Legacy's AMT tweeters affected their overall sound character, even in the bass.

I found it interesting how he could comment on how the tweeters sound could effect the bass reproduction when comparing two different speakers, if you follow my thinking ???

 

[JustIntonation]

Yes the AMT23 is quite bad (or good, depending on your room) / narrow in vertical dispersion.
See how for the AMT23 the dispersion given is for 5 and 15 degrees and for the dome tweeters the dispersion is given for 30 and 60 degrees.
And you can't look at the 15 degree dispersion and add two of them to guess for the 30 degree dispersion as usually the 30 degree dispersion will be far worse than adding twice the 15 degree number.
Similarly you can't guess the 15 degree dispersion from the 30 degree dispersion number of the domes as the 15 degree dispersion of the domes will be much better than half the 30 degree dispersion.
So in my guess the AMT23 will be at least 4 times as narrow vertically than the already "bad" measuring domes in this respect you listed below.

I don't think the 23 is bad at all in terms of dispersion. You frequently see things like the Scanspeak below. Note the different angles in the measurements, you have to guesstimate I'm afraid. Mo
St domes are falling off below 10khz, the 23 doesnt

23
View attachment 15964
View attachment 15965

Scanspeak dome
View attachment 15966

Very expensive Scanspeak revelatorView attachment 15967

Seas millenium
View attachment 15968

 

[JustIntonation]

I've found my tweeter!
The BlieSMa T34A-4
http://www.bliesma.de/product.html
http://hificompass.com/en/speakers/measurements/bliesma/bliesma-t34a-4
http://www.audioxpress.com/article/test-bench-the-t34a-4-tweeter-from-bliesma
http://www.troelsgravesen.dk/BlieSMa_T34A-4.htm

I was looking in a very different direction, using an ATC or Volt middome and crossing this over at 3 to 3.5kHz to the tweeter, perhaps Mundorf AMT21.
But then I was a bit hesitant about crossing over around 3 to 3.5kHz. I was thinking why am I hesitant about this.. Is it past speaker experience? I've had a bad experience around this region with the K+H O300 and to some degree several other speakers, it is the presence / harshness region. And speakers that cross lower haven't sounded harsh to me in this region. Also when looking at reviews of quality speakers which use a very low distortion ATC dome but again cross at ~3kHz they seem to give a magnifying glass to the mids but are particularly hard on songs and are not particularly good at giving a whole picture of the music. In contrast, for instance the Amphion speakers get rave reviews at how great they are at giving the whole picture and fantastic holographic soundstage and these cross over at 1.6kHz (with a waveguide and measurements still show a distortion bump around the crossover due to the tweeter being stressed). And these don't even measure as good as an ATC for instance. How can this be, while the crossover is smack in the middle of the midrange and there are so many testaments all over the internet on how not to cross over in the mids as this is where our ears are most sensitive.
And then it occured to me.. The equal loudness curve, doesn't this have a little dip in the middle?
I looked it up and it sure does!

And just look at that..
How can you get a more clear picture of where to cross over?
You cross over at ~1300-1400Hz!
And certainly not at 3-3.5kHz where the ear is ~6 to 9dB more sensitive.
And this is not just because of crossover slight off-axis phasing and ringing. There are many other things to consider around the crossover. Mid/woofer dropoff, change of diffraction due to different location of the drivers, shift in harmonic and non linear distortion signatures from the different drivers, even the effect on the ear/pinna of the different vertical placement of the drivers which is more sensitive in higher frequencies. All of these are ~6-9dB less audible when placing the crossover this low. And you get the most sensitive and important region of about 2-5kHz handled by a single driver. Coincidentally you also get easy near coaxial vertical phase behavior as discussed earlier in this thread
Perhaps an alternative could have been a mid driver handling 400Hz to 5-6kHz or so, but it feels less natural to me to cross over to a supertweeter at 5-6kHz or higher (hard to get a good in phase angle at the crossover due to the very small distance required to the mid driver) and I'm not sure if there is a mid driver that can do the ~400-6000Hz range well enough with room to spare around the crossover points.

Anyhow, this is how I found the BlieSMa driver. It's clear a 1" tweeter is not up to the task of this low a crossover even in a waveguide (compression driver in a horn is a different story but not suitable for my nearfield use) and the BlieSMa is the only one that can do it this well. Lucky I didn't want to design this speaker a year ago as there would have been no good tweeter yet for the task, the BlieSMa was only released this year
Btw, there is a beryllium version as well, the T34B-4, but I don't see any benefits other than for bats and a slightly enhanced sensitivity (and it does slightly worse in fact for the third harmonic around the crossover).

Now looking for a great woofer to go with the BlieSMa. Perhaps a Seas W18EX001 would be a natural match.. Could add a true sub later as well.