What is the best tweeter

[Kerry Armes]

How about a good CD (18 sound 1.4” model, HF garbage is largely dictated by the horn or waveguide and may look different on a better one):
View attachment 11686

View attachment 11688
As you can see it has a good flat response with very good directivity control. Better than any other design so far. It’s distortion is silly low. More amazing, look carefully at the drive level. Pro drivers are tested at far higher levels than consumer drivers. This tweeter can play far louder than any consumer tweeter can dream of with no distortion. Obviously the really Hf stuff is a little messier than the ribbon, but again, this is completely waveguide dependent. Even the measuring distance matters here. It’s caused by reflections in the waveguide. Other designs don’t have this problem and it’s not as audible a problem as it looks.

Which 18sound model is this? I've never seen a compression driver measure like that before. Even the very good ones usually have high second order (above 1%) and low third order but not that low.

 

[Matthew J Poes]

Which 18sound model is this? I've never seen a compression driver measure like that before. Even the very good ones usually have high second order (above 1%) and low third order but not that low.

https://www.audioxpress.com/article/test-bench-eighteen-sound-nd3st-1-4-compression-driver

ND3ST. It’s excellent.

When you say this, are you referring to your own measurements or those from Vance? He and I had a chat about this and tested a Dome tweeter at the same level as a CD is tested at. The Domes distortion was much higher.

 

[Kerry Armes]

Measurements from both. If you look at most other CDs he tests they have much higher second order distortion.

 

[Matthew J Poes]

Measurements from both. If you look at most other CDs he tests they have much higher second order distortion.

They are being played back at 10dbs higher and it’s based on a full band pink noise level. A dome won’t be any better at 104dB (and many of those CD’s are actually 120dB in their operating bandwidth. I’ve actually asked Vance to explain why. I think I know but want confirmation.

 

[Matthew J Poes]

I currently have speakers with domes, ribbons, and CD. I can test later at different safe volume levels and see what I get. That will make direct comparison possible.

I can’t test Planars or such as I don’t have any to test at the moment. Maybe we can get some sent to me later to add to the comparison data.

I was thinking of testing at 75dB, 85dB, and 90dB. The 90 is probably the highest level I’d be comfortable testing a speaker I don’t own with $500+ worth of tweeter.

 

[tesseract]

Matt, the Chane 5.0 A2rx-c set I owned prompted me to add a 3 channel amp to my AVR. Those planars love power.

 

[Matthew J Poes]

Matt, the Chane 5.0 A2rx-c set I owned prompted me to add a 3 channel amp to my AVR. Those planars love power.

I have very little experience with finished product Planar speakers. I wouldn’t know. I’ve tested raw B&G drivers or cheap crap speakers with leaf tweeters (as they call it). I may have had one or those Dayton planars once, can’t recall for sure anymore.

I’ve mostly played with domes, ribbons, and CD’s. Pretty much only CD’s the last 5-6 years.

 

[tesseract]

Planars exhibit the same dispersion properties as ribbons, wide to the sides, with vertical nulls. The main difference is that the voice coil is also the driver with ribbons, necessitating a transformer to give the amp something other than a dead short to see. Lighter, quicker is the gain.

 

[Kerry Armes]

They are being played back at 10dbs higher and it’s based on a full band pink noise level. A dome won’t be any better at 104dB (and many of those CD’s are actually 120dB in their operating bandwidth. I’ve actually asked Vance to explain why. I think I know but want confirmation.

Hi Matt,

I just want to clarify a couple of things. First, you cannot test distortion with pink noise because all of the frequency band is being played at the same time and therefore you would never be able to accurately measure the harmonics. Vance uses pink noise to set the SPL level and then tests distortion in the same way as everyone else would, which is a long wave sign sweep.

Second, one of the issues I have with the way Vance measures is that he is not consistent from driver to driver, which makes them impossible to legitimately compare. There are two ways to test distortion that provides consistency and both have benefits and downsides. You could either match SPL and distance between all measurements or you could test everything at 2.83V (or some other consistent voltage) and keep distance the same. If distance changes, the measurements are not directly comparable. If input level or SPL are not kept constant, measurements are not directly comparable. Vance even sometimes changes the scale of the graph to show a different frequency band.

To me, this is a little frustrating because he does a great job of getting accurate measurements, it's just that they aren't necessarily comparable. Now, Vance does have to be somewhat careful because he is working for a magazine that is trying to sell publications and advertising to the people who are getting their drivers tested, so it doesn't surprise me that this might occur.

But all I was really referring to is that every other compression driver I've seen him test has much higher second order distortion. I wasn't comparing to a dome but to those other compression drivers he tests, which are usually in a somewhat similar manner.

Here are some links to show you what I am talking about with differences in testing format. All of these were tested differently.

Two compression driver:
Measured at 2.83V/1M equivalent
https://www.audioxpress.com/article/test-bench-b-c-speakers-de780tn-8-compression-driver

Measured at 2.37V/1M equivalent
https://www.audioxpress.com/article...pression-driver-from-radian-audio-engineering

Two dome tweeters:
Measured at 4.82V/1M equivalent
https://www.audioxpress.com/article...s-from-dayton-audio-rst28a-4-and-the-rst28f-4

Measured at 4.82V/1M equivalent but graph only goes to 3 kHz as opposed to 2 kHz like the previous
https://www.audioxpress.com/article...on-1-inch-aluminum-and-magnesium-dome-tweeter

 

[Matthew J Poes]

Hi Matt,

I just want to clarify a couple of things. First, you cannot test distortion with pink noise because all of the frequency band is being played at the same time and therefore you would never be able to accurately measure the harmonics. Vance uses pink noise to set the SPL level and then tests distortion in the same way as everyone else would, which is a long wave sign sweep.

Second, one of the issues I have with the way Vance measures is that he is not consistent from driver to driver, which makes them impossible to legitimately compare. There are two ways to test distortion that provides consistency and both have benefits and downsides. You could either match SPL and distance between all measurements or you could test everything at 2.83V (or some other consistent voltage) and keep distance the same. If distance changes, the measurements are not directly comparable. If input level or SPL are not kept constant, measurements are not directly comparable. Vance even sometimes changes the scale of the graph to show a different frequency band.

To me, this is a little frustrating because he does a great job of getting accurate measurements, it's just that they aren't necessarily comparable. Now, Vance does have to be somewhat careful because he is working for a magazine that is trying to sell publications and advertising to the people who are getting their drivers tested, so it doesn't surprise me that this might occur.

But all I was really referring to is that every other compression driver I've seen him test has much higher second order distortion. I wasn't comparing to a dome but to those other compression drivers he tests, which are usually in a somewhat similar manner.

Here are some links to show you what I am talking about with differences in testing format. All of these were tested differently.

Two compression driver:
Measured at 2.83V/1M equivalent
https://www.audioxpress.com/article/test-bench-b-c-speakers-de780tn-8-compression-driver

Measured at 2.37V/1M equivalent
https://www.audioxpress.com/article...pression-driver-from-radian-audio-engineering

Two dome tweeters:
Measured at 4.82V/1M equivalent
https://www.audioxpress.com/article...s-from-dayton-audio-rst28a-4-and-the-rst28f-4

Measured at 4.82V/1M equivalent but graph only goes to 3 kHz as opposed to 2 kHz like the previous
https://www.audioxpress.com/article...on-1-inch-aluminum-and-magnesium-dome-tweeter

Thanks Kerry.

I think you misunderstood my post. I didn’t mean to imply he tested distortion with punk noise, it was that he set the levels that way.

I emailed him to get specifics. He measures with bandwidth limited pink noise at 1 meter but measures distortion at 10cm. He indicates that it is appropriate and accepted that pro drivers are measured at a higher level. They aren’t meant to be compared. That the higher levels and variation in the fundamentals isndue to difference in the response and sensitivity. I’m not totally sure why sensitivity matters but that’s what he said.

You are absolitely right about that CD being unusually low in distortion. I’m not totally sure why, it’s Lowe than others he’s tested.

But some things I’ve cobsidered. First, it is an unusual Cd so that’s may be part of it. second, it’s a large optimally designed waveguide that was developed recently and does not use diffraction. All things considered, it is possible that it contributes less distortion than other waveguides.

You may already know this but a waveguide itself develops specifically 2nd harmonic distortion. How much and where depends on the design and size of the waveguide. Nearly every CD Vance tests is in a different waveguide, as such it’s unclear how much the waveguide contributes. Vance indicates that the CD is going to be the dominant source of distortion. However Geddes disagrees and a review of some articles examining this also seem to suggest that isn’t true. It seems that the waveguide is often either a dominant or equal source of 2nd harmonic distortion.

I decided to test this myself last night. I was curious how a CD would compare to a dome tweeter under otherwise similar conditions. I have a ton of data from the experiment to go through still, including compression data. Basically the test involved raising the level of the speaker in 3dB increments until such time that compression was reached (level wouldn’t increase at 3dB anymore).

This goes from 88dB to 105dB. It was labeled but the label is cut off.

The CD is the one with higher distortion above 3khz. That is a DE250 on a Geddes OS waveguide. Dome was a Dayton silk dome. The spike in distortion was caused by the location of the mic causing a cancelation in the response of the waveguide. Just ignore it.

 

[Kerry Armes]

Thanks Kerry.

I think you misunderstood my post. I didn’t mean to imply he tested distortion with punk noise, it was that he set the levels that way.

I emailed him to get specifics. He measures with bandwidth limited pink noise at 1 meter but measures distortion at 10cm. He indicates that it is appropriate and accepted that pro drivers are measured at a higher level. They aren’t meant to be compared. That the higher levels and variation in the fundamentals isndue to difference in the response and sensitivity. I’m not totally sure why sensitivity matters but that’s what he said.

You are absolitely right about that CD being unusually low in distortion. I’m not totally sure why, it’s Lowe than others he’s tested.

But some things I’ve cobsidered. First, it is an unusual Cd so that’s may be part of it. second, it’s a large optimally designed waveguide that was developed recently and does not use diffraction. All things considered, it is possible that it contributes less distortion than other waveguides.

You may already know this but a waveguide itself develops specifically 2nd harmonic distortion. How much and where depends on the design and size of the waveguide. Nearly every CD Vance tests is in a different waveguide, as such it’s unclear how much the waveguide contributes. Vance indicates that the CD is going to be the dominant source of distortion. However Geddes disagrees and a review of some articles examining this also seem to suggest that isn’t true. It seems that the waveguide is often either a dominant or equal source of 2nd harmonic distortion.

I decided to test this myself last night. I was curious how a CD would compare to a dome tweeter under otherwise similar conditions. I have a ton of data from the experiment to go through still, including compression data. Basically the test involved raising the level of the speaker in 3dB increments until such time that compression was reached (level wouldn’t increase at 3dB anymore).

View attachment 11761

This goes from 88dB to 105dB. It was labeled but the label is cut off.

The CD is the one with higher distortion above 3khz. That is a DE250 on a Geddes OS waveguide. Dome was a Dayton silk dome. The spike in distortion was caused by the location of the mic causing a cancelation in the response of the waveguide. Just ignore it.

Oh, ok. Got ya. I thought you were saying he tested at with pink noise. I think we are on the same page then.

I would agree with Vance that the compression driver is going to be the dominant source. This should be testable by taking a distortion measurement with and without the horn. The horn does reduce relative distortion at the low end of the driver though, so you'd probably have to EQ to flat to make a comparison.

Thanks for posting the data. Were these both tested at the same initial SPL and was there any crossover in place on the low end?

 

[Matthew J Poes]

Oops I meant he measures the spl level with bandwidth limited pink noise. Did it again.

 

[Matthew J Poes]

Oh, ok. Got ya. I thought you were saying he tested at with pink noise. I think we are on the same page then.

I would agree with Vance that the compression driver is going to be the dominant source. This should be testable by taking a distortion measurement with and without the horn. The horn does reduce relative distortion at the low end of the driver though, so you'd probably have to EQ to flat to make a comparison.

Thanks for posting the data. Were these both tested at the same initial SPL and was there any crossover in place on the low end?

Both tested at the same levels and yes there is a highpass crossover on both. They both had a roughly flat response. I was trying to compare on similar grounds.

 

[AudiocRaver]

Great discussion!

I agree, Matt, that a good CD design can tend to be a "room eraser," making early reflections much less of an issue than with other designs. A good CD really pulls you into the recording and the ambient and reflected qualities that ended up in the mix. For instance, ever heard a recording where you could almost hear the size of the vocal booth the vocals were recorded in? That kind of detail stands out with a good CD design, and the soundstage and imaging (SS&I) end up dependent upon mainly the recording and the speaker, with minimum contribution from the room. With most other speaker/tweeter types, the SS&I are dependent on a mixture of the speaker's characteristics and the room. Early reflections become a critical part of that equation (topic for another thread).

Each of these two SS&I approaches has its own characteristics to love or hate. With a CD tweeter, the room matters less. With a dipole (at the extreme other end of the "how much the room matters" spectrum), like an electrostatic design, the room matters A LOT! I have reviewed a few CD designs that I really liked a lot for the reasons Matt has given (I really need a decent CD speaker pair for reference one of these days). But I adore the soundstage from a pair of electrostatics properly set up as well!

So here is a fun question: Is the SS&I resulting from a good CD tweeter design a more accurate representation of the recording, and therefore the ideal design approach? I suppose one could argue yes. However, the SS&I from a dipole with the right setup is not to be taken lightly. It can be absolutely yummy!

Once again, we get down to personal preference.

 

[Matthew J Poes]

Great discussion!

I agree, Matt, that a good CD design can tend to be a "room eraser," making early reflections much less of an issue than with other designs. A good CD really pulls you into the recording and the ambient and reflected qualities that ended up in the mix. For instance, ever heard a recording where you could almost hear the size of the vocal booth the vocals were recorded in? That kind of detail stands out with a good CD design, and the soundstage and imaging (SS&I) end up dependent upon mainly the recording and the speaker, with minimum contribution from the room. With most other speaker/tweeter types, the SS&I are dependent on a mixture of the speaker's characteristics and the room. Early reflections become a critical part of that equation (topic for another thread).

Each of these two SS&I approaches has its own characteristics to love or hate. With a CD tweeter, the room matters less. With a dipole (at the extreme other end of the "how much the room matters" spectrum), like an electrostatic design, the room matters A LOT! I have reviewed a few CD designs that I really liked a lot for the reasons Matt has given (I really need a decent CD speaker pair for reference one of these days). But I adore the soundstage from a pair of electrostatics properly set up as well!

So here is a fun question: Is the SS&I resulting from a good CD tweeter design a more accurate representation of the recording, and therefore the ideal design approach? I suppose one could argue yes. However, the SS&I from a dipole with the right setup is not to be taken lightly. It can be absolutely yummy!

Once again, we get down to personal preference.

Wayne you raise a great point. I think arguing the technical merit of a speaker is great fun, I do it all the time. But my dirty little secret is that I've heard speakers that I think sound absolutely amazing that do not, in any way, meet my criteria of an ideal speaker. That shouldn't be if one design is right and others are wrong. Yet that is my lived experience.

Since the embrago is now over I can finaly share that I am currently lucky enough to get to use the Philharmonic BMR Philharmonitor speakers. This amazing 3-way speaker is an ideal example of what I think is a "wrong" speaker. If my view that a controlled dispersion speaker with an elevated DI is the right speaker, then this speaker, one with an extremly wide dispersion, a totally flat DI that is very low, must be a "wong" speaker. Yet, I have to say, its a glorious sounding speaker. The soundstage isn't wrong, it isn't bad, its amazing, it's just different than what my Gedlee speakers produce.

It is a far more room dependent speaker, it's placement within the room isn't hard to get right, but it doesn't sound good against a wall. Moving acoustic panels in and out of the room's sidewalls make obvious differences. That isn't so with my Gedlee speakers. I really wanted to not like these speakers. I have huge respect for their designer, Dennis Murphy, and I think what he accomplished is amazing. A speaker with a response that flat over that wide of an angle is amazing. However, I just believed that I wouldn't personally like the speaker.

I can't help but conclude that there is more to good sound than these small design details. Is a smooth response important? Absolutely. Is an early reflection response that follows the shape of the direct response important? Absolutely. Does that early reflection response need to be attenuated significantly to sound good? Well, now I'm questioning that one.

And what's more interesting is that this speaker actually challenges some of the "rules" established in some of the research produced by Harman, likely because Harman has never tested a speaker quite like this (Olive and Toole have actually hinted that this is true). Because the power response of this speaker is so even and flat, it's in-room response is far flatter than that of a typical speaker. It sounds good, but that means applying room correction or adding subwoofers require that you no longer follow the typical rule of thumb. This speaker actually sounds good with a relatively flat in-room response.

http://www.philharmonicaudio.com/BMR Philharmonitor.html

What this speaker struggles with, which is really not bothersome under most circumstances, is dynamics. It's not nearly as efficient as my normal speakers, and the BMR midrange and ribbon tweeter both show signs of compression at modest levels.

And this isn't the only speaker that has challenged my ideals. It's probably the most extreme example, but others exist for sure.

 

[Kerry Armes]

Wayne you raise a great point. I think arguing the technical merit of a speaker is great fun, I do it all the time. But my dirty little secret is that I've heard speakers that I think sound absolutely amazing that do not, in any way, meet my criteria of an ideal speaker. That shouldn't be if one design is right and others are wrong. Yet that is my lived experience.

Since the embrago is now over I can finaly share that I am currently lucky enough to get to use the Philharmonic BMR Philharmonitor speakers. This amazing 3-way speaker is an ideal example of what I think is a "wrong" speaker. If my view that a controlled dispersion speaker with an elevated DI is the right speaker, then this speaker, one with an extremly wide dispersion, a totally flat DI that is very low, must be a "wong" speaker. Yet, I have to say, its a glorious sounding speaker. The soundstage isn't wrong, it isn't bad, its amazing, it's just different than what my Gedlee speakers produce.

It is a far more room dependent speaker, it's placement within the room isn't hard to get right, but it doesn't sound good against a wall. Moving acoustic panels in and out of the room's sidewalls make obvious differences. That isn't so with my Gedlee speakers. I really wanted to not like these speakers. I have huge respect for their designer, Dennis Murphy, and I think what he accomplished is amazing. A speaker with a response that flat over that wide of an angle is amazing. However, I just believed that I wouldn't personally like the speaker.

I can't help but conclude that there is more to good sound than these small design details. Is a smooth response important? Absolutely. Is an early reflection response that follows the shape of the direct response important? Absolutely. Does that early reflection response need to be attenuated significantly to sound good? Well, now I'm questioning that one.

And what's more interesting is that this speaker actually challenges some of the "rules" established in some of the research produced by Harman, likely because Harman has never tested a speaker quite like this (Olive and Toole have actually hinted that this is true). Because the power response of this speaker is so even and flat, it's in-room response is far flatter than that of a typical speaker. It sounds good, but that means applying room correction or adding subwoofers require that you no longer follow the typical rule of thumb. This speaker actually sounds good with a relatively flat in-room response.

http://www.philharmonicaudio.com/BMR Philharmonitor.html

What this speaker struggles with, which is really not bothersome under most circumstances, is dynamics. It's not nearly as efficient as my normal speakers, and the BMR midrange and ribbon tweeter both show signs of compression at modest levels.

And this isn't the only speaker that has challenged my ideals. It's probably the most extreme example, but others exist for sure.

Matt,

There is an interesting thread on the diyaudio.com forum discussing this topic if you haven't already seen it.

https://www.diyaudio.com/forums/multi-way/330741-preference-direct-radiators.html

There are a number of ways to skin a cat in this hobby. The biggest constants in good sound in my opinion are flat on-axis, smooth off-axis/power response, and low distortion in that order. I think you can sometimes even get subjectively good (or at least decent) sound breaking some of these rules, but if you follow those three things, the only thing that can ever make the speaker sound bad is the room.

The waveguide systems remove more of the room from the equation but produce a different sound to my ears. If you have a good room, I would argue that you don't necessarily need this. If your room is extremely lively, then it is going to be a big help.

I personally prefer wide and smooth over narrow and smooth directivity based on all the systems I've heard in almost all of my rooms. I did however, live in an apartment for a short period that had terrible accoustics and a large waveguide system definitely helped a lot.

 

[AudiocRaver]

Matt,

There is an interesting thread on the diyaudio.com forum discussing this topic if you haven't already seen it.

https://www.diyaudio.com/forums/multi-way/330741-preference-direct-radiators.html

There are a number of ways to skin a cat in this hobby. The biggest constants in good sound in my opinion are flat on-axis, smooth off-axis/power response, and low distortion in that order. I think you can sometimes even get subjectively good (or at least decent) sound breaking some of these rules, but if you follow those three things, the only thing that can ever make the speaker sound bad is the room.

The waveguide systems remove more of the room from the equation but produce a different sound to my ears. If you have a good room, I would argue that you don't necessarily need this. If your room is extremely lively, then it is going to be a big help.

I personally prefer wide and smooth over narrow and smooth directivity based on all the systems I've heard in almost all of my rooms. I did however, live in an apartment for a short period that had terrible accoustics and a large waveguide system definitely helped a lot.

Kerry, you really boiled that down nicely. Great summary.

My own preference is about like yours, wide (or dipole) and smooth, low distortion, and make the room work with and for the speakers.

Matt, no doubt implementation and integration are key to the sound of a finished design. Having reviewed several MartinLogan hybrid designs in recent years (and owning two models), I have felt like they excel at hybrid integration.

 

[ddude003]

A bit of history quoted from Martin Logan: "In 1955, Peter Walker published three articles regarding electrostatic loudspeaker design in Wireless World, a British magazine. In these articles, Walker demonstrated the benefits of the electrostatic loudspeaker. He explained that electrostatics permit the use of diaphragms that are low in mass, large in area and uniformly driven over their surfaces by electrostatic forces. Due to these characteristics, electrostats have the inherent ability to produce a wide bandwidth, flat frequency response with distortion products being no greater than the electronics driving them."...

Something to think about... Hundreds of square inches of low mass surface area... 292 square inches in the smallest ML electrostatic hybrid design... What other tweeter design can claim this? Except it is not just a tweeter... It seamlessly manages from 500 Hz on up to 22,000 Hz in this configuration without a crossover...

 

[Matthew J Poes]

Matt,

There is an interesting thread on the diyaudio.com forum discussing this topic if you haven't already seen it.

https://www.diyaudio.com/forums/multi-way/330741-preference-direct-radiators.html

There are a number of ways to skin a cat in this hobby. The biggest constants in good sound in my opinion are flat on-axis, smooth off-axis/power response, and low distortion in that order. I think you can sometimes even get subjectively good (or at least decent) sound breaking some of these rules, but if you follow those three things, the only thing that can ever make the speaker sound bad is the room.

The waveguide systems remove more of the room from the equation but produce a different sound to my ears. If you have a good room, I would argue that you don't necessarily need this. If your room is extremely lively, then it is going to be a big help.

I personally prefer wide and smooth over narrow and smooth directivity based on all the systems I've heard in almost all of my rooms. I did however, live in an apartment for a short period that had terrible accoustics and a large waveguide system definitely helped a lot.

Wow I read through that when it first started and stopped following, it has grown. I think it was less than 5 pages when I last looked.

I know Patrick and his thinking. We both bought and built Geddes speakers around the same time. He has to be the most obsessive tinker I’ve ever seen. Constantly trying interesting/goofy ideas.

I do think Patrick’s initial claim is flawed. What blind research has proven that people prefer “direct radiators”, what Falgott rightly pointed out is really a stand-in for wide dispersion. My opinion is that this preference is mostly just a sighted bias against horns and waveguides. I don’t believe that wider dispersion is widely preferred. If that would true it would poke holes in the research around the importance of room reflections and preferences for intact first reflections.

I think my views fall very in line with Tooles contribution on that thread. Yes I know that thread touches upon it but that isn’t saying anything new. How a speakers DI impacts its room interaction has been pretty well understood for close to 70 years.

My perfect speaker still would, conceptually, have a pretty high DI. But like I said, Ive heard plenty of speakers that look like the BMR’s and still sound amazing.

Vivid Audio is another one:
https://www.soundstage.com/index.ph...udio-giya-g2-loudspeakers&catid=77&Itemid=153

A great example of an ideal wide dispersion speaker. Very low and flat DI. Not what I think is best, but I’ve heard these and they sound pretty spectacular.

I think it would be interesting to have a chance to play with the B&O Beolab 90 with a MUSRA test. See not only how people detect the change in directivity under blinded conditions but how they rate their preference and why. It would be interesting to do the test in a room that could also blindly change the rooms reflections at early reflection points.