MEH / Synergy Horns vs. direct radiators - deep dive w/ measurements

[stoneeh]

I recently formalized my personal objective data based findings on MEHs vs direct radiators, as well as anything substantial I could source from third parties. Here it is - enjoy:

 

[Chris A]

I think you forgot about IMD of the direct radiating mode of the drivers vs. the same drivers being used in the MEH aperture (usually you will see a 20-25 dB difference). This is not only measurable, but highly audible. THD may be higher in the SH-50, but is essentially inaudible at home hi-fi levels. PWK pointed this out many times in his career.

At the time that Erin did the SH-50 measurements, he didn't have all the Klippel S/W modules he wanted (cost). After multiple exhortations to provide phase and group delay data, he produced a "hand hack" of the data I requested, which was not at the resolution that the Klippel NFS provides. Additionally, Erin had trouble positioning the SH-50 for NFS measurements due to his inexperience (at that time) using an NFS for measuring anything but nearfield direct-radiating drivers (which his SH-50 measurements were not, since he measured at the mouth of the horn aperture.).

Another thing to keep in mind--the SH-50 was designed for using passive crossovers in a PA application--not hi-fi, and not designed to use hi-fi drivers. As such, its three-way design was a compromise for PA duties. You can see this in its amplitude response of each of its three ways, with significant compromises in its crossover design (really a balancing network with many notch filters but no crossover filters, per se) to support 127 dB continuous output level. Other hi-fi versions of the MEH having two-way designs and DSP crossing of ways (woofer to compression driver) are much cleaner and smoother sounding: https://community.klipsch.com/topic/161404-a-k-402-based-full-range-multiple-entry-horn/

Bottom line: if you are going to make pronouncements of MEH designs in general, based on the limited experience of the PA-duty SH-50 and the other third party MEH here (for home hi-fi duty), I think you need a lot more data, and you need to look at the higher hi-fi Danley models--like the HRE-1 and ILE3 designs (among others, like his Studio Monitor 2, etc.).

YMMV.

Chris

 

[Waxx]

There are indeed much better MEH's around now, mainly build by diy people on math and measurements, and tuned for hifi. That would be a more honest comparation than this (in the meantime old) design in the early stages of this technology. On diyaudio.com the builders come together and share measurements, and ways to calculate and sim (with BEM software like ATH and ABEC.). They work further on the tech by Tom Danley. The reason you don't see much commercial builds yet is that the basic patent only recently expired. And the technology behind it is also not fully mature yet, it's still getting finetuned. Modern MEH's are also done with DSP, as you can finetune it a lot more (mainly on time and phase issues) than passive like in the SH-50.

 

[Chris A]

The reason you don't see much commercial builds yet is that the basic patent only recently expired.

There has been a lot of FUD propagated over the US8284976 ("Synergy") patent owned by DSL...which essentially replaced the US6411718 ("Unity") patent owned by Sound Physics Labs (Tom Danley worked at each company). The Unity patent actually expired in 2014 due to non-payment of fees (see second link). Essentially all the newer patent does is preclude crossing the higher frequency drivers below the so-called first notch frequency on the MEH aperture. This is something that I find is not essential in MEH design, although crossing at a frequency below the first-notch frequency of the lower frequency drivers can be used to help with consistency of polar coverage at the first notch frequency. (YMMV.) I think a lot of DIYers have inadvertently infringed on this, but it's difficult to spot and enforce legally, since most DIY MEH designs are enabled by DSP crossovers, whose settings can be switched in the blink of an eye.

https://www.diyaudio.com/community/threads/why-are-their-so-few-synergy-unity-builds.287990/

And the technology behind it is also not fully mature yet, it's still getting fine tuned.

I think this statement is really aimed at those smaller boutique companies doing home hi-fi designs--whose engineers and designers don't really understand modern horn design. Among the larger JBLs and Klipsch companies in this country, I think they are fully aware of the technology, but choose not to go down the "Unity Horn" path due to the not-invented-here (NIH) organizational syndrome. JBL has designed their trough-type line array modules making full use of the MEH concept, e.g.,

Chris

 

[Chris A]

I recently formalized my personal objective data based findings on MEHs vs direct radiators, as well as anything substantial I could source from third parties. Here it is - enjoy:

By the way, Danley publishes virtually all its loudspeaker design EASE and CLF measurements, so there is very little here that is truly useful in terms of "new insights".

In fact, I might go as far as to say that this video doesn't look anything like a fair assessment of the MEH capabilities...which is precisely why I commented above.

Chris

 

[Moonlightshadow]

Thank you @stoneeh for your work!

I understand the SH-50 was not volunteered by Danley to Erin. Now I understand why.

By the way, Danley publishes virtually all its loudspeaker design EASE and CLF measurements, so there is very little here that is truly useful in terms of "new insights".

Practically none of the types of measurements, especially decay, distortion, and compression, referenced in this review, are included in a GLL or CLF.

Your "there are better MEH out there, I have no evidence" objection above is equally ridiculous.

This type of quality of response will not encourage anyone to share their valuable data here.

 

[gnarly]

Interesting, thx.
I like how you give an intro to what a synergy/unity/meh is physically. And I like how you start with their acoustic goal, smooth constant directivity.
That said, I don't think you can extrapolate the measurements from the two samples, the SH50 & Solana, into being a fundamental part of the MEH concept.

If I saw a narrow distortion peak like on the SH-50 has at around 1kHz at such a low SPL level, I'd be looking for an issue. On anything I've ever built, which includes 11 versions of MEHs, and about the same number of straight direct radiator designs, a small narrow distortion peak like that means an issue....somethings loose, trash in voice coil, something.
So I tend to think It's hard to have such a narrow peak without something being wrong...like you yourself said, you've never seen it before either.

I don't think you can extrapolate the SH-50's 0.5dB compression at 102dB into higher SPLs. It will need to be measured. The apparent compression at 102 dB may not be compression in the usual sense of thermal compression....if fact, I'd think likely not. Synrgy drivers "see" each other and have acoustic interactions. My guess from my measurements of my syns at different SPLs, is that the driver interactions have a little 'frequency response altering SPL dependency'....not a desirable trait for sure, imo, but not one that leads to compression. I haven't experienced significant compression on any of mine, as far as I've been able to measure/hear.

The comparison of the mid drivers response on the Solana vs your direct radiators, I think shows valid info which will apply to all mehs.
It's impossible to not create some diffraction or issues with any kind of structure in front of a radiator. The only question is are the issues correctable over the constant directivity pattern. This ends up being individual speaker model dependent. So I give a 50% vote on this part of your presentation presenting a fundament property of mehs. Maybe yes, maybe no, but with increased likelihood of yes.

I can't see IMD being an inherent property of mehs, because i think reducing IMD with conventional drivers is primarily a function of reducing driver sections' bandwidths, so it comes down to how many driver sections does either a meh, or a direct radiator build have. The big meh advantage here is how to physically do that, minimizing center-to-center driver spacings necessary for smooth constant directivity in both axis.

Ok, now for a purely personal perspective on what is a synergy/unity/meh....I think they need to be relatively large to take full advantage of their concept.
It's a matter of how low in frequency can pattern control be held. The difference in the SH-50 and the Solana is striking to me....unfortunately size matters.
I think when a meh gets small, it's in essence competing with similarly sized traditional coaxials, and needs to show real improvements over then to make sense.
We both know the guys making this effort and they are doing some great stuff. Watching the progress, so far leads me to think it's trickier to get things right on the smaller mehs, than on larger. I'm eagerly watching their research.
The point of this personal perspective is to say, I think it's a mistake to be extrapolating too much from the Solana example yet, ......in terms of stating universal meh properties.

 

[stoneeh]

"there are better MEH out there, I have no evidence" objection above

I agree that claims without data don't warrant a serious response, but I will still say that I had considered that point going in myself - could another MEH, implemented differently, possibly not suffer from the issues I discovered? The answer quickly became clear: normal (frontloaded) horn speakers do not show these symptoms, and the only significant difference between them and MEH is that the drivers don't excite the horn directly, but via small openings. Since this is the only meaningful physical difference, it is the only possible reason for the increased compression & distortion. The openings, and especially their small size, are an inherent feature of an MEH, and can't be changed by a meaningful degree, because they're required for pattern control, impedance matching, etc. - and for the reason of them being an inherent design feature, that has to be present in a similar manner for all MEHs, a general conclusion can very much be derived.

 

[Chris A]

Here's data on the SH-50--the amplitude response--in each of the three ways:

As you can see from this plot, the cone midrange drivers are the ones that cover the area where the rise in second harmonic distortion (largely inaudible due to masking) is found.

Its passive crossover is probably the major player in any apparent compression distortion.

I would not expect that the amount of compression distortion seen in Erin's NFS measurements would be any worse at higher on-axis SPL.

Chris

 

[gnarly]

I agree that claims without data don't warrant a serious response, but I will still say that I had considered that point going in myself - could another MEH, implemented differently, possibly not suffer from the issues I discovered? The answer quickly became clear: normal (frontloaded) horn speakers do not show these symptoms, and the only significant difference between them and MEH is that the drivers don't excite the horn directly, but via small openings. Since this is the only meaningful physical difference, it is the only possible reason for the increased compression & distortion. The openings, and especially their small size, are an inherent feature of an MEH, and can't be changed by a meaningful degree, because they're required for pattern control, impedance matching, etc. - and for the reason of them being an inherent design feature, that has to be present in a similar manner for all MEHs, a general conclusion can very much be derived

Like said previously, hopefully politely and with reason, I don't think you have sufficient data for your conclusions.
Mores samples needed, more measurements needed, even on the only two examples provided.
Bluntly, too much thinking, not enough measuring..

We need to see real measurements at higher SPL on any MEHs that you want to extrapolate MEH properties from..... both for compression and distortion.
Pls get that data, and I'll take it seriously.

 

[Chris A]

So no comments on SH-50's real measurement data? No one else here actually owns one to measure its performance?

Pity.

I suppose that the revealed purpose of the thread isn't actually "neutral science" after all...but something else.

Comparing direct radiator loudspeaker performance to well-designed MEH performance isn't a winning formula for direct radiator designs, unless suppression of key performance capabilities that correlate to the psychoacoustic preferences of humans listening to MEHs is the means of the "investigation" to arrive at the argument that is attempting to be made.

Chris

 

[Moonlightshadow]

I agree that claims without data don't warrant a serious response, but I will still say that I had considered that point going in myself - could another MEH, implemented differently, possibly not suffer from the issues I discovered? The answer quickly became clear: normal (frontloaded) horn speakers do not show these symptoms, and the only significant difference between them and MEH is that the drivers don't excite the horn directly, but via small openings. Since this is the only meaningful physical difference, it is the only possible reason for the increased compression & distortion. The openings, and especially their small size, are an inherent feature of an MEH, and can't be changed by a meaningful degree, because they're required for pattern control, impedance matching, etc. - and for the reason of them being an inherent design feature, that has to be present in a similar manner for all MEHs, a general conclusion can very much be derived.

Sensible self-check. The logic I can also only agree with.

I've seen that comment on the video referring to the Hinson MEH and your response, in addition to some of the absurdity here that I have already responded to. The lengths some people will go to try to make unpleasant facts about their favorite design go away ..
Anyway, I'm sure you realize that your research is appreciated by the serious audience.

 

[stoneeh]

It's a recognizeable pattern. I've been publishing research for about a decade now - always starting with personal curiosity about a topic, then feeling the responsibility to share the results when/if I discover something important - always data-driven, impartial, and very self critical for my own sake (because I really want to know the answer to the question, not something tainted or subjective). The public response widely varies. The pattern has shown itself to be purely the emotional response of the viewer - if the data concludes in something he likes / prefers / enjoys, I, the conveyor, get praise, and if the opposite, the response is equally negative.


More data is also the reason for this reply. For the 2nd DUT, the Solana, I did perform multitone (IMD) measurements, but didn't realize I had also done the same measurement for the midwoofers as direct radiators (simply forgot, it's been a while and I'm always busy). So for completeness' sake, here's that comparison as well.

Basic documentation on my multitone measurements can be found here. The measurements were performed for a full multi-way system each, so to isolate the midwoofers, I cut the graphs to the frequency spectrum they were filtered / XOd to (200-1000 Hz). SPL was 110 dB RMS for the complete system each.

Left direct radiators, right MEH:

As in the harmonic distortion comparison, the MEH's distortion is vastly increased vs. the direct radiator's, by as much as 20 dB. However, very interestingly, the harmonic distortion increase (shown in the video) progressed towards lower frequencies, while the IMD distortion increase concentrates itself at the upper end of the MEH's midwoofer frequency range, specifically right around the bandpass peak of the openings. As per Klippel, IMD and HD have partly similar, partly differing mechanical and electrical causes - in this case, the latter seem to dominate.


Finally, here, on the left, is the complete measurement of the Solana, and on the right I threw in a similar total membrane area & market class (vs. the 6ND430 midwoofers, which are still top of the line) conventional PA top, a TW Audio M12, as comparison. Crossover frequency of the Solana is 1 kHz, which is very visible in the graph.




My only remaining question would be why it has taken so long, and a private, non-commercial effort, to attain this kind of data & comparison. Manufacturers, like obviously most prominently Danley, could have always thrown a large part of their portfolio towards Erin, the Production Partner, or similar independent, data-focused reviewers. Most of the serious PA companies like JBL, d&b, L'Acoustics, etc etc., have. Hell, even Funktion One let the Production Partner test some of their speakers, they measured horribly, and I'm pretty sure it hasn't made a dent in their sales. And the MEH DIY community could have supplied these types of measurements as well - many of them have suitable measurement equipment, and performing these types of measurements reasonably accurately isn't that much of an obstacle. Someone must have been as curious about the actual (not assumed, or perceived) performance of their speakers long ago, I'd have thought.

 

[Audio Science Pal]

I recently formalized my personal objective data based findings on MEHs

Your video showed an image of what looked similar to an SH-96. https://www.danleysoundlabs.com/products/sh96/

I hope to have a pair of SH-96 one of these days. Has your research exposed you to any data on that model, which I imagine should measure better than the SH-50.

If there are reasons to cross those off my wish list, it might be good to know before I hear and like them.

 

[Chris A]

The reason why I commented above was due to this kind of statement that inevitably ensues (either implicitly...or explicitly like it is here):

...If there are reasons to cross those off my wish list, it might be good to know before I hear and like them...

I don't really mind much if someone wants to believe that up is down, and down is up--but only if that person doesn't go to the web and start trying to convince other people that they have scientific data saying their belief is "the truth". I do have an issue with that.

And cherry picking certain measurement types while completely disregarding a raft of other measurements. implying "look, look...a defect!...right HERE!" in an attempt to disqualify certain loudspeaker designs--isn't too cool, either, especially when they didn't take those measurements.

(Does that mean that those that do that have their own axe to grind?)

The Klippel NFS is not really what I'd pick for acoustically measuring MEHs since the measurements made on MEHs are not really the same "near field" as they are with direct radiating loudspeakers. I instead recommend ground plane or elevated (upwards directed) measurements at 90-100 dB at the measurement microphone, or better yet-- measurements in an anechoic chamber. Get back away from the horn mouth by one or more metres when taking measurements, i.e., those that correspond to what the human hearing system actually perceives.

There are issues that can arise with certain MEH design approaches. For instance, in my experience, using separate cone-type midrange drivers instead of dual diaphragm drivers that are able to physically get their two internal diaphragms within 1/4 wavelength of each other at the crossover frequency.

Another one is in using compression drivers with diaphragms that exhibit significant non-pistonic motion above 10 kHz (e.g., 1.4"-->2" titanium dome drivers), i.e., that audibly "chatter".

Another one is using an MEH driver crossover schema that inserts all-pass phase growth, i.e., using typical non-MEH IIR crossover filters.

Another one is building MEHs that do not control their polars below 400-500 Hz--and calling them "small MEHs".

These are design decisions that I personally do not recommend.

But I don't try to convince someone that using these design approaches that I don't particularly recommend are "tagged out" just because they use them in their designs. I do say that I wouldn't personally use those design approaches. And I don't generally try to cherry pick esoteric, edge of the SPL performance envelope data comparisons to try to imply that one design approach is valid and the other approach isn't.

Chris

 

[Waxx]

I agree with Chris, i'm quiet sure MEH have disadvantages also. But your way of using data, from a few devices and in a way we can't controll is not the way to show those. You need a lot more samples and types of data before you can make such a general conclusions. Your review is at most an indication that there is an issue at that point, but it needs a lot more data to prove that.

 

[Tell]

Another one is building MEHs that do not control their polars below 400-500 Hz--and calling them "small MEHs".

These are design decisions that I personally do not recommend.

But I don't try to convince someone that using these design approaches that I don't particularly recommend are "tagged out" just because they use them in their designs. I do say that I wouldn't personally use those design approaches. And I don't generally try to cherry pick esoteric, edge of the SPL performance envelope data comparisons to try to imply that one design approach is valid and the other approach isn't.

I've been temped for quite a while building one of those "small MEHs" and will hopefully so that relatively soon, but I'm curious why you don't recommend them?

 

[Chris A]

but I'm curious why you don't recommend them?

I wouldn't use a smaller horn aperture than ~32" horizontally and ~19" vertically...because you're giving up the real advantages of doing MEHs by using a smaller horn mouth.

And building or buying a larger MEH mouth is bird feed in terms of added costs. Why not just find a way to integrate into your room? If your room REALLY IS too small for larger horn mouths--first consider using headphones instead. Your listening room is too small.

(I don't like to use any listening room less than 13'-14' in depth/horizontal dimensions because of problems of first bounce reflections from the back of the room that come too soon to the listeners.)

Right now, I can go out and buy a Celestion Axi2050 and put it on a K-402 horn, and not do any crossing to lower frequency drivers--and I get down to 200-250 Hz at the lower breakpoint frequency using only one amplifier channel and amplitude (SPL) EQ only--and it will perform just as well or better in-room for home hi-fi audio as an MEH built to go down to that frequency (or higher).

You don't need to have all the "MEH rigamarole" to get down to 200 Hz. The real reason why I think MEHs are used is because their horn apertures support at least horizontal polar control down to ~200 Hz or lower.

Before doing your "small MEH", first go listen to a full-range MEH that have been dialed-in--like the ones that Danley makes in its Synergy series...in a room that's treated for early reflections (less than 4 ms). Listen to the neutrality of sound quality that comes with flattened amplitude AND phase response...AND full-range directivity down to the room's calculated Schroeder frequency.

You'll never want to go back after hearing it.

Chris

 

[mwmkravchenko]

I watched the video when you first released it. I thought that the information was well presented. The information was useful as well.

I do have a few questions. Measuring a horn at the apex of the mouth is not going to allow the best results of the horn to be measured. Generally you need to be a few wavelengths of the lowest frequencies that you are expecting the horn to reproduce. Roughly 2.5 metres if you expecting around 100 hertz as the low end of the horn. The horn itself messes with the nearfield response many times.

Mark

 

[Waxx]

like said, you need a horn that is big enough to support the whole passband to have all the advantages of a MEH, which is next to a single point source effect and controlled directivity. If the horn is not big enough the directivity will not be controlled low enough to make the effort.

And today you got compression drivers like the Axi2050 or the B&C DCX464 that go low on their own (with the right horn), and they often do it better in their passband. If i would make a MEH, i would like to use such a driver, and woofers to get the horn low enough to couple to subwoofers. That means the horn must be very big (like the K-402 or even bigger) to make that work. Small MEH's that only do treble and mid are often worse sounding than that type of compression drivers (also made by some others like BMS) and not necesairly cheaper than those (quiet expensive) drivers to build.

If you want to test another one, test the design that Scott Hinson made, that is wel documented, tested and proven by many diy builders. That is considered a good hifi MEH, not those Synergy horns of Danley (those are ment for big pa systems). And like said, you need to measure the horn from a distance, not in the hornmouth.