Evidence-based Speaker Designs

[direstraitsfan98]

Precisely - and some are so much more sophisticated than "hi-fi" speakers, but are crippled by their small size, price constraints and peculiar tuning choices. The natural question to ask is: what if such technology was harnessed in a less compromised form factor? And why hasn't anything like that emerged outside of DD or Kii?

Because that's not what the market is dictating. It's far easier to market a box with cheap drivers and a crossover as a passive design with lots of buzzwords and fancy patents and dress up said speaker to be much more then it is then to innovate. I love what Dutch and Dutch and Kii are doing but I don't think we'll ever see the industry you are wanting and expecting.

 

[Juhazi]

Would you care to comment on the following? Can you name the Tymphany (Peerless) part numbers? Provide their photos?

From https://en-de.neumann.com/product_files/1710/download:
"The 1” tweeter will become the new standard for all new monitors in the KH-series by Neumann. The selection was therefore done extremely carefully using countless prototypes and tests until the desired properties were reached.The final choice was a titanium sandwich cone. Developer Markus Wolff told us that the decision in favour of a metal cone instead of a pure fabric membrane was based on the significantly more pistonic motion of the metal cone.This usually comes with the disadvantage of a pronounced resonance, but this is above 30 kHz and in this case very well damped. Actually a resonance in this frequency range way above the audible spectrum would not be a problem anyhow. But if the resonance is started, it could lead to intermodulation distortion even within the audible frequency range, which makes suppression of the resonance worthwhile. A pure fabric tweeter cone does not have this problem, but would vibrate unevenly way below 20 kHz and therefore would lead to an uncontrollable dispersion behaviour."

Actually fabric tweeters have smoother dispersion without wg just because of this. The waveguide however takes advantage of the firm metallic membrane and those often are smoother and can go louder too. But every combination is unique, there are matches made in heaven and hell...
https://kimmosaunisto.net/WaveGuides/WaveGuides.html

 

[Sancus]

The natural question to ask is: what if such technology was harnessed in a less compromised form factor? And why hasn't anything like that emerged outside of DD or Kii?

My guess: Development cost and complexity.

Building something like the Apple Homepod is a complex and difficult challenge, far beyond the capabilities of most boutique, low volume speaker builders. They're rather specialized in putting drivers in a box. How many of them even have software engineers beyond the third party they contract to build their website?

An active speaker on the level of the Kii/DD involves custom PCB design and an entire team of software engineers. These development costs are prohibitively high unless you're selling enough units. This also directly conflicts with the 'lets make a zillion different variations' business model that so many boutique speaker companies seem to love. Mads from Buchardt said as much in his post here. They're buying a tunable OEM platform for the electronics of their active versions, which is what makes it possible. I suspect the availability of these kinds of solutions on the OEM side will make actives more common, but that will take time.

Honestly, I think a lot of the boutique hi-fi makers are really just not that interested in fundamental technology changes. They're selling a myth, not a functional product. It's a big risk to invest a lot of money in technology when you're not really selling technology to begin with.

 

[Ilkless]

They're selling a myth, not a functional product. It's a big risk to invest a lot of money in technology when you're not really selling technology to begin with.

Nail on head IMO. And the enthusiast market isn't critical or knowledgeable enough to be holding them to account for their regressiveness.

 

[Thomas_A]

Because that's not what the market is dictating. It's far easier to market a box with cheap drivers and a crossover as a passive design with lots of buzzwords and fancy patents and dress up said speaker to be much more then it is then to innovate. I love what Dutch and Dutch and Kii are doing but I don't think we'll ever see the industry you are wanting and expecting.

I think it has been the same story for many years. I recall the active Audio Pro A4-14 and B2-50 speakers in 1978 with a clever amp to manipulate the drivers (ACE-bass) and push-pull-to reduce distortion in the bass region; the measured very good and would be highly competitive even today. Those speakers sold good and made also quite some buzz in the AES at the time. And just to remark that active designs are not necessary to get good speakers, there are quite a few good speakers originating e.g. from Harman and others. In Sweden there is also speakers like the Ino Audio pi60 that has been around for 40 years, and the measure and sound as good as the best in the world. But sadly some other good speakers disappear, like models from the B/W matrix series, Snell and Dunlavys.

 

[digitalfrost]

I’d love to hear your thoughts on the two speakers if you get the chance. I’m about to build some er18dxts myself.

Ok so my setup is: I listen at the desktop. I've built DIY subwoofers that serve as stands for the LS50. I flipped the subwoofers on their side, this way the ER18DXT tweeter is only 3cm higher than the tweeter of the LS50 used to be. I compensate for that with my seat position so tweeters are at ear level.

My ER18DXT are closed box, not bass-reflex, always been a subwoofer guy.

Here's some old measurements of the LS50 and new measurements of ER18DXT, both after correction. ER18DXT is calibrated (upper lines), LS50 was not:



Not sure what's going on in the right channel there. I think I remember it had a notch filter around that frequency, maybe a cable has got loose. But I think we can conclude it looks pretty similar. I have calibrated both to 78dBC at the listening position.

Long story short: They're pretty close. Soundstage is a bit wider with the ER18, and I think they bring out details in the background more clearly, but that might be related to that frequency hump. You've given me a problem here, because I just don't know which I prefer.

Also I got a serious case of 'new speaker excitement', it's been a while since I listened to these and I'm having a lot of fun.

The LS50 do a better job of disappearing. With the ER18, I can more clearly make out the sound is coming from the speaker.

 

[FrantzM]

I think it has been the same story for many years. I recall the active Audio Pro A4-14 and B2-50 speakers in 1978 with a clever amp to manipulate the drivers (ACE-bass) and push-pull-to reduce distortion in the bass region; the measured very good and would be highly competitive even today. Those speakers sold good and made also quite some buzz in the AES at the time. And just to remark that active designs are not necessary to get good speakers, there are quite a few good speakers originating e.g. from Harman and others. In Sweden there is also speakers like the Ino Audio pi60 that has been around for 40 years, and the measure and sound as good as the best in the world. But sadly some other good speakers disappear, like models from the B/W matrix series, Snell and Dunlavys.

Speaking of which.. I remember fondly the rather impressive sounding (for me anyway at the times in my teens) Philips Motional Feedback speakers and I still remember how impressive most Dunleavy speakers remain. These remain competitive with the best... High End Audio didn't see well his views on cables and on amplifiers, he may not have been a good business manager or marketer either ... As for Snell its principal, Kevin Voecks moved to Harman to help produce the superlative Revel speakers line among others... One of my favorite speaker designer, (still trying to get my hand on one of his designs), Earl Geddes is rather indifferent about Active ...
My personal opinion is that active (line level) crossovers are the best solution in speakers. An approach taken by the late and immensely talented Siegfried Linkwitz... His speakers are passive, crossover-less with amplification driven by active crossovers. In these days of Hypex, IFI and Ice modules wiping the floor with most classic amplifers (A/B) designs , I see this as the better approach.. Interestingly the top of the line JBL M2 has taken it.. Lost in many discussions is the power afforded by some solutions on the DSP/Active crossover front.For exemple, miniDSP, offers some spectacularly powerful products. They have a variety of models to fill the DSP needs and wants of most enthusiasts. These may not have category leading measurements but seem for the most part below the hearing threshold of the most sensitive ears around in term of THD... miniDSP has competition in this space. I can't recall the brands and would appreciate the input of the collective. For those that are not faint of heart a combo of a good PC ( Intel NUC with an i5, for exemple for size, low-power consumption) or some other powerful Small Form Factor PC running windows or Linux and coupled with something like a Lynx 16-channel board or a FocusRite or RME or any reputable and solid multi-channel Pro Audio Interface will provide all the crossover , Room Correction and DSP the most sophisticated speaker designer can dream of... We are at an interesting threshold today.
Peace

 

[Thomas_A]

Speaking of which.. I remember fondly the rather impressive sounding (for me anyway at the times in my teens) Philips Motional Feedback speakers and I still remember how impressive most Dunleavy speakers remain. These remain competitive with the best... High End Audio didn't see well his views on cables and on amplifiers, he may not have been a good business manager or marketer either ... As for Snell its principal, Kevin Voecks moved to Harman to help produce the superlative Revel speakers line among others... One of my favorite speaker designer, (still trying to get my hand on one of his designs), Earl Geddes is rather indifferent about Active ...
My personal opinion is that active (line level) crossovers are the best solution in speakers. An approach taken by the late and immensely talented Siegfried Linkwitz... His speakers are passive, crossover-less with amplification driven by active crossovers. In these days of Hypex, IFI and Ice modules wiping the floor with most classic amplifers (A/B) designs , I see this as the better approach.. Interestingly the top of the line JBL M2 has taken it.. Lost in many discussions is the power afforded by some solutions on the DSP/Active crossover front. miniDSP offers some spectacularly powerful products. They have a variety of models to fill the SP needs and wants of most enthusiasts. These may not have category leading measurements but seem for the most part below the hearing threshold of the most sensitive ears around in term of THD... miniDSP has competition in this space. I can't recall the brands and would appreciate the input of the collective. For those that are not faint of heart a combo of a good PC ( Intel NUC with an i5, for exemple for size, low-power consumption) or some other powerful Small Form Factor PC running windows or Linux and coupled with something like a Lynx 16-channel board or a FocusRite or RME or any reputable and solid multi-channel Pro Audio Interface will provide all the crossover , Room Correction and DSP the most sophisticated speaker designer can dream of... We are at an interesting threshold today.
Peace

Active filter are good, but active speakers should also be done with amp-driver optimisation for lowest distortion. Just putting in an Hypex or Ice module is IMO wasting significant benefits that can be done, as was shown in the ACE-bass speakers and patents 40 year ago. I am not sure any of the active speakers here use similar solutions. Such benefits can otherwise partly be had with passive components, at least for mid-tweeter crossovers, or a combination of passive and active circuits.

 

[dc655321]

active speakers should also be done with amp-driver optimisation for lowest distortion.

What optimization(s) are you referring to?

 

[Thomas_A]

What optimization(s) are you referring to?

Impedance matching, current drive for example. With active amps you can do those solutions.

 

[KSTR]

Impedance matching, current drive for example. With active amps you can do those solutions.

Exactly. If one really digs into this it means finding the drive impedance profile (output impedance vs. frequency) that gives the best results overall with regard to harmonic and intermodulation distortion, large signal behaviour and overdrive recovery, power compression and robustness against dynamic and long-term driver parameter changes. A compomise, as usual.
Output impedance profile (part of which can be realized with passive components after the amp, eg. series caps and coils) is a powerful additional degree of freedom available in active designs that should be exploited, alas only few manufacturers currently do this (though some don't advertise these features because the actual circuit designers implement this silenty... like I did in the ADAM F-Series).

 

[jhaider]

Speaking of which.. I remember fondly the rather impressive sounding (for me anyway at the times in my teens) Philips Motional Feedback speakers and I still remember how impressive most Dunleavy speakers remain. These remain competitive with the best... High End Audio didn't see well his views on cables and on amplifiers, he may not have been a good business manager or marketer either ... As for Snell its principal, Kevin Voecks moved to Harman to help produce the superlative Revel speakers line among others... One of my favorite speaker designer, (still trying to get my hand on one of his designs), Earl Geddes is rather indifferent about Active ...

I may be in a minority, but the few times I heard Dunlavy speakers I thought they were better as a story than a product. One of my undergrad mentors had SC-IIIs and he expressed some regret at buying them unheard. I was able to hear the bigger multiway ones a few times, too. They never disappeared for me, though it is hard to forget you're listening to speakers when there are two huge boxes in front of you.

As for Snell, while I do not believe I have heard a Snell designed by Kevin, I have heard one designed by David Smith. His Snell speakers were built to a vertical array concept that started with a tight MTM using small midranges and a bowtie-shaped tweeter waveguide, with woofers spaced at a mathematically-derived interval and high-order crossovers all around. I remember them fondly and would like to hear them again, though they are also going on 20 years old now. Apropos of some earlier discussion, I think Smith works at Bose now.

 

[jhaider]

Would you care to comment on the following? Can you name the Tymphany (Peerless) part numbers? Provide their photos?

I cannot speak to manufacturer promotional material, either direct or conveyed through a reviewer. Such things have minimal worth.

I did not take photos of the drivers. The labels on them were from the OEM. I suspect the part numbers are unique to Neumann. That could mean functional changes, cosmetic changes, a different QC routine, or just the customer's desire for their own SKUs.

The main thing I remember about the tweeter is surprise it had a ferrite rather than neo magnet, given how small the cabinet is. It looked a lot like this one, except with bosses on the faceplate to align the dome in the waveguide:
https://www.parts-express.com/peerless-da25bg08-06-1-aluminum-dome-tweeter-6-ohm--264-1460
Tweeters all look very similar, so that may be wrong. Given the promotional text they may use Tymphany's standard anodized ("sandwich" in marketing terms) titanium dome that is also widely used by, inter alia, PSB.

The woofer I remember more clearly, because the stamped basket was distinctive. It looked like this one:
https://www.parts-express.com/peerl...9qbypcByWGsb8xpxRrOo7I_iiB1BBsKsaAtLhEALw_wcB

None of the above should be interpreted as reflecting badly on the KH120. Tymphany makes great drivers. The KH120 is a beautifully engineered minimonitor. I think it represents good value, too.

 

[Mashcky]

My ER18DXT are closed box, not bass-reflex, always been a subwoofer guy.
[...]
The LS50 do a better job of disappearing. With the ER18, I can more clearly make out the sound is coming from the speaker.

Thanks for the impressions. Could the difference in disappearing ability be due to the point source nature of the coaxial versus the traditional drivers? Otherwise great to hear the er18dxt perform so well beyond their parts cost, a win for diy!

I would go sealed-box also if I had the guts to put a couple of subwoofers in a small apartment. I’ll have to plug the ports later on when I scale up.

 

[pozz]

Thanks for the impressions. Could the difference in disappearing ability be due to the point source nature of the coaxial versus the traditional drivers? Otherwise great to hear the er18dxt perform so well beyond their parts cost, a win for diy!

I would go sealed-box also if I had the guts to put a couple of subwoofers in a small apartment. I’ll have to plug the ports later on when I scale up.

It's likely more to do with the difference in off-axis response, both vertical and horizontal, and the direct/reflected sound mix of @digitalfrost's room. The LS50's coincident drivers and box emphasize radial dispersion.

 

[KSTR]

From the first principles of acoustics, of course the ports will affect the radiated sound field. I haven't disassembled KH120, so I don't know precisely how long the ports are, and thus what their phase behavior is depending on frequency, but with sound wavelength at 1 KHz being an order of a third of a meter, the sound field in the 1 KHz to 2KHz range shall be affected asymmetrically by the KH120 asymmetrically placed ports.

I have. The port (and internal volume stuffing) design is aimed at lowest leakage, not highest efficiency. They did this by making the ports multi-path. The tubes have cutouts close to the baffle that are covered with some kind of felt. This reduces the port's pipe resonance Q (most important as the woofer is so close to the ports) and provides multiple paths for higher frequencies that allows for some cancelling to happen. The port lenght itself might have also been choosen for optimum leakage behaviour.
I would think no attempts have been made to tailor this further to get a controlled directivity. Rather they tried to minimize the leakage and make its character benign (non-resonant).

 

[Cosmik]

From the first principles of acoustics, of course the ports will affect the radiated sound field. I haven't disassembled KH120, so I don't know precisely how long the ports are, and thus what their phase behavior is depending on frequency, but with sound wavelength at 1 KHz being an order of a third of a meter, the sound field in the 1 KHz to 2KHz range shall be affected asymmetrically by the KH120 asymmetrically placed ports.

I'm interested in your views on ports, Sergei. You were very careful to discriminate between continuous sine waves and transients in the discussions on MQA. Do you extend this to your views on ports?

 

[jhaider]

Are you referring to the "line of sight" directivity principle?

No. I don't even know what that is.

False based on the line of sight directivity principle?

Demonstrably false based on actual measurements of real loudspeakers.Polar maps of real loudspeakers are too rare, but they do exist. Here are four published examples of 1" dome tweeters on flat waveguides:

1) Ascend Audio 6.5" 2-way, measured at Princeton Univ.

2) Outlaw Audio 5" 2-way, measured by James Larson (Audioholics), likely the best hifi loudspeaker reviewer working right now

3) RBH 5" 2-way, also measured by Larson

4) Monitor Audio 6.5" 2-way, my measurement. (Sorry, direct link to the polar map graph is blocked, search for "polar map.")
https://hometheaterhifi.com/reviews...-audio-silver-1-bookshelf-loudspeaker-review/

Note the general trend: nearly 180deg dispersion around the crossover, narrowing to about 140deg at 10kHz.

Now, KH120 (factory data):

Roughly 110deg around the crossover, progressively declining to about 90deg at 10kHz.

So yes, it is clear that a 1" dome tweeter on a flat waveguide exhibits wider dispersion than a 1" dome tweeter on a contoured waveguide. One need not delve into 4D esoterica to observe that fact.

AAgreed. A similarly-sized two-way monitor with a port opening under the woofer will exhibit qualitatively similar dispersion pattern. As much as I don't find this pattern attractive, such design has its uses: in my case three of the surround speakers had to be placed close to the walls, so the back firing ports or hot back plates weren't acceptable.

To be clear, this passage does not show agreement with what I wrote above. I have not seen evidence that port location - note that one of my examples in the previous post had a side mounted passive radiator - affects the vertical directivity. What clearly does is the center-to-center spacing. That is why coaxial speakers have substantially identical horizontal and vertical polars (affected by waveguide dimensions if not symmetrical). They have no center-to-center spacing. Steering the null by varying driver phase or delay can also affect the null resulting from non-coaxial drivers. It is common in both home and pro speakers to steer the crossover null downwards, to reduce floor reflections and decrease sit-stand frequency response variation.

 

[Sergei]

I have. The port (and internal volume stuffing) design is aimed at lowest leakage, not highest efficiency. They did this by making the ports multi-path. The tubes have cutouts close to the baffle that are covered with some kind of felt. This reduces the port's pipe resonance Q (most important as the woofer is so close to the ports) and provides multiple paths for higher frequencies that allows for some cancelling to happen. The port lenght itself might have also been choosen for optimum leakage behaviour.
I would think no attempts have been made to tailor this further to get a controlled directivity. Rather they tried to minimize the leakage and make its character benign (non-resonant).

Interesting information. Thanks!

I guess our earlier apparent disagreement with Jhaider can be settled then. I happen to prefer designs without ports, and consider the ports as generators of undesired - by me - irregularities in the sound field, which have to be justified by other advantages.

It appears Jhaider considers more gradations: perhaps there are "good" ports and "bad" ports, and the ones on KH120 are "good". I would still argue that they are "better than most" rather than unconditionally "good", yet at that point it would be hairsplitting.

I thought, based on my hands-on experimentation with KH120, available polar plots, and opinions of some reviewers, that the shape and placement of the ports had something to do with the directivity control, yet I won't insist on such interpretation.

How come the Neumann/Sennheiser designers haven't joined the ASR yet? We need some direct authoritative answers. Wait! This is a perfect excuse for Amir to contact them and invite them to the forum

 

[Sergei]

Now, KH120 (factory data):

I appreciate the detailed answer. However, it appears you are talking about the KH120 horizontal directivity plot. I was discussing the vertical:

and how the KH120 plot is "twirly" at negative angles, compared to smoother, more symmetrical plot of port-less KH310: