Evidence-based Speaker Designs

[Ilkless]

Would you be able to change the level at exact certain places?
If so, it should be possible.

However, I'm uncertain what level and distortion it could achieve below the midrange. While one can use a separate subwoofer for the lowest frequencies, the mid - and upper bass is still crucial for a great design IMO.

I was actually imagining something like a RAAL ribbon stretched out and curved as one piece, with the amplitude shading built in at a driver level (eg. by varying the field strength as suggested by @12B4A) so that you'd only need one amplifier channel connected to one driver input. Doesn't have to be run that low, but as a mid-tweeter crossed to cone drivers as used currently.

Basically a one-piece transducer/radiating surface that follows the curvature of the CBT, rather than a HF array of multiple spaced drivers.

 

[Frank Dernie]

Is a one-piece curved ribbon with smoothly decreasing drive level (eg. through the physical design of the motor) feasible as an extreme implementation?

Probably not. A true ribbon isn’t tensioned so basically hangs vertically. The apogee had a foam wedge snubber preventing the ribbon getting out of the magnetic gap on mine but certainly the only way to support a curved ribbon would be a multiple of these, then it wouldn’t be a ribbon...

 

[Bjorn]

We have briefly looked at using a wide band ribbon driver in a CBT. Most likely we will need a custom ribbon driver.

Something to keep in mind though is that when you're using multiple drivers you can make a super driver out of mediocre drivers. The energy, the resolution, the clarity, etc. with even very low cost drivers in a CBT (as long as they're sufficiently small) is way above a traditional speaker. So I guess what I'm saying is that I don't know how much you would actually win by using high quality ribbon drivers in such a design. It would certainly be interesting trying it though!

 

[Ilkless]

We have briefly looked at using a wide band ribbon driver in a CBT. Most likely we will need a custom ribbon driver.

Something to keep in mind though is that when you're using multiple drivers you can make a super driver out of mediocre drivers. The energy, the resolution, the clarity, etc. with even very low cost drivers in a CBT (as long as they're sufficiently small) is way above a traditional speaker. So I guess what I'm saying is that I don't know how much you would actually win by using high quality ribbon drivers in such a design. It would certainly be interesting trying it though!

Was thinking that a single curved radiator could eliminate the need to consider vertical CtC spacing, at least in the midrange and treble, and it seemed like it could be a natural offshoot of large curved radiators that have already been brought to market such as the Muraudio curved electrostat and MBL Radialstrahler. Point taken though!

Also, are Horbach-Keele filters relevant to CBTs? (eg. could a CBT with horizontal pair-wise symmetric driver configuration be designed using Horbach-Keele filters while retaining the curved enclosure and the Legendre shading on the vertical axis?)

 

[Habu]

With the speakers that there are with measurements, buy one that does not have them is like playing Russian roulette. The same with amplifiers and others audio hard.

Hello from France,

I totally agree with "maty"
An evidence-based speaker or audio hardware need a lot measurements during it's design, so the maker has just to release the last measurements at the end since they exist and they can proove how good the speaker or audio hardware....
And independant reviews could verify that measurements are true.
But if the maker is not able to give access to measurements, maybe these measurements are not good, maybe it was designed by ear, or it is a big brand really proud of it's amazing products...

Sincerely yours

Habu

 

[Bjorn]

Was thinking that a single curved radiator could eliminate the need to consider vertical CtC spacing, at least in the midrange and treble, and it seemed like it could be a natural offshoot of large curved radiators that have already been brought to market such as the Muraudio curved electrostat and MBL Radialstrahler. Point taken though!

It's a neat idea. Thanks for mentioning it.
FIY: There's is no CtC spacing issue in the midrange to consider with a CBT. It's only a matter in the very highs and mostly only in near field, depended on the size and spacing of the high freq. driver.

Also, are Horbach-Keele filters relevant to CBTs? (eg. could a CBT with horizontal pair-wise symmetric driver configuration be designed using Horbach-Keele filters while retaining the curved enclosure and the Legendre shading on the vertical axis?)

Not that I know of.

 

[Bjorn]

One really cool way is illustrated below. We have called this "The Matrix".
View attachment 21597

Maybe we should call it "The Dalek"!

 

[maty]

This CBT speaker we're designing now works very well without any subwoofer by the way but does require an external crossover and 4 channel amplifers...

What do you think about tweeters with modern class D amps?

 

[Bjorn]

What do you think about tweeters with modern class D amps?

Considering I'm involved in a design of a power amplifier based on Ncore NC500 (with bridging, gain adjustments, trigger etc.) I think that answers your question
Both Ncore and Anaview AMS class D modules are terrific. Transparent with ability to drive difficult loads.

 

[Ilkless]

http://nuke.nonsoloaudiofili.com/HEDDType05SeriesONENearfieldStudioMonitor/tabid/504/Default.aspx

Excellent teardown with measurements of the HEDD Type 05. ICEPower biamplification, a beefy midwoofer, waveguide AMT, DSP crossover (with 8th-order low-pass and 6th-order high-pass). However, I don't know the measurement conditions, so take them skeptically as merely indications of design choices, such as crossover slope.

There is a curious 900Hz peak that can be trivially equalised out (doesn't vary off-axis). There is also some curious response shaping baked into the DSP. Nonetheless, the dispersion is matched at crossover and the response is still +/- 2dB within the band.

That this speaker, with all the technology and features, is made in Germany and sold for $1200 (under $1000 sold in Europe) is an indictment of the ubiquitous MDF passive boxes in a similar form factor. Remarkable that such passive boxes can cost that much even with offshore production.*

*Note that I'm not a skeptic of offshore production in China, Taiwan, Indonesia, Malaysia etc. With appropriate oversight they can make some amazing equipment like the Revel Performa3. However, there are so many hi-fi darlings like the Monitor Audios of the world providing far less for a similar price, despite offshoring. Which puts into question whether the purported cost savings are substantially translated into more competitive products.

 

[jhaider]

The woofer on that HEDD speaker looks like it might be Wavecor. That's a pretty expensive driver for a $600 speaker.

 

[maty]

HEDD Type 05, active

vs passive

New Studio Monitor, GR Research (aka Danny Richie). 6.5" + ribbon. For now passive but active in the future too.

[IMG, link] http://www.gr-research.com/pics/studio1.jpg

https://www.audiocircle.com/index.php?topic=160647.msg1729409#msg1729409

KEF Q100 5.25" coaxial speakers, passive.

 

[Ilkless]

HEDD Type 05, active

vs passive

New Studio Monitor, GR Research (aka Danny Richie). 6.5" + ribbon. For now passive but active in the future too.

[IMG, link] http://www.gr-research.com/pics/studio1.jpg

https://www.audiocircle.com/index.php?topic=160647.msg1729409#msg1729409

KEF Q100 5.25" coaxial speakers, passive.

If you want to be an apologist for backwards, antiquated loudspeaker design, I suggest you don't choose CSDs of different scales (ignoring that different measurement setups can't be compared) to push your agenda.

The HEDD CSD is on a 40dB. The lower limit is almost twice as low as the 24dB scales both others are using.

 

[maty]

You are right with the scale, but you can see the curves ant not only at 0.0 ms. I do not like the spectral decay of the HEDD ribbon.

 

[maty]

You can see what I say here too with the directivy. I know, scale again.

HEDD Type 5

KEF Q100

KEF LS50

 

[Ilkless]

You are right with the scale, but you can see the curves ant not only at 0.0 ms. I do not like the spectral decay of the HEDD ribbon.

The time resolution is also higher (number of "slices") on the HEDD CSDs. The curves are scarcely comparable when one is significantly. What is clear is there are several design choices realistically impossible with speaker-level passive crossovers, such as the 6th and 8th-order slopes.

You can see what I say here too:

HEDD Type 5

KEF Q100

KEF LS50

The colour divisions used on the KEF graphs are double (8dB) that of the HEDD ones. Again you use graphs that can't be compared to push for regressive, antiquated engineering methods, against all evidence - eg. given in the active v passive thread on this very forum.

 

[maty]

We all know about the problem with the directivity of ribbon tweeters. It is unfair to compare them with a good coaxial speaker.

But if we compare decay curves, the HEDD ribbon suffers from that problem, unlike, apparently, the other ribbon design. This second has a decay more like a good coaxial.

 

[Ilkless]

We all know about the problem with the directivity of ribbon tweeters. It is unfair to compare them with a good coaxial speaker.

But if we compare decay curves, the HEDD ribbon suffers from that problem, unlike, apparently, the other ribbon design. This second has a decay more like a good coaxial.

I'm nor quite sure what you intend to do comparing decay curves of vastly different vertical scale and time-resolution (look at the number of "slices" and gaps between them), all from different measurement setups. This is a bizarre interjection.

 

[maty]

More easy, do lou like HEDD spectral decay at HF? I do not.

 

[Ilkless]

More easy, do lou like HEDD spectral decay at HF? I do not.

The spectral decay is measured down to a lower floor. 40dB vs 24dB. You "like" certain decay characteristics of other speakers based on decay measured to a 24dB floor, and dislike that with a graph made to a 24dB floor. Are you even comparing them correctly? Are you looking at the peaks that pop up above the 40dB floor, that wouldn't even be shown on a CSD with a 24dB floor? This is ridiculous, and akin to saying a FR graph with a 2dB scale is rougher than that with a 5dB scale. Of course it'd be. Even the exact FR data would result in a "rougher" graph.

If you really, wanted to, go see the -24dB point on the HEDD and see the corresponding time. In the treble it ends well before 1.24ms, just like all the other speakers. I'm done commenting on wilful anti-intellectual behaviour (using incomparable data sets to make biased conclusions).