KEF LS50 Meta vs HEDD Type 07: An (un)fair comparison

After my DIY project based on Tang Band W6-2313 coaxial drivers disappointed me I wanted to compare it to two other speakers I've been wanting to hear for a while:
One, the HEDD Type 07. These are the Series One without the built-in DSP. I know frenchbat has these speakers and I know Voldemike uses Adam A77Xs. I also liked what I heard from the Adam A7Xs and the larger S3H when I compared the studio monitors available at the justmusic store near me. I thought the HEDD Type 07 would essentially be an improved Adam A7X, so I was very interested in trying them.
Two, the KEF LS50 Meta. It's been getting very good reviews ever since it got released and it seemed to fix the tonality I disliked about the original LS50. I always found the LS50 boomy, shouty and sizzly in the treble. The LS50 Wireless already fixed the tonality and in addition managed to achieve minimum phase response through its crossover region, which I think improves imaging.

I wanted to replace my 'DreiZwo Sat' speakers on my desk with either of those speakers, but I mainly compared them on speaker stands in front my my large OB speakers, since the desk isn't a very optimal place to put speakers.

TL;DR: I tried to rate the speakers from 0-15 in a couple areas that matter to me so you can get a glimpse of their performance quickly:



Without linearising the frequency response magnitude and phase:

HEDD Type 07:
First up, these are not studio monitor neutral. They have a pretty large upper midrange dip (2-3kHz) and while it's tastefully done I wouldn't rely on their stock performance to judge tonality. Sadly the Lineariser goes too far for my liking, with a measurable elevation around 3kHz at the listening position. More on that later.
The AMT here seems the most improved compared to the Adam A7X. Gone is the grit that plagued it at times. However I think it's also a bit too smooth sounding (which I think was also a problem with the Adam AMT). It reminds me a bit of those lightweight planars in its treble presentation. This is a minor complaint however and something that I only noticed after some time. It's also easy to adjust to.
What disappointed me the most was that it consistently omitted some low level information. Certain claps on Hotel California from HFO were missing compared to the other two speakers, for example. I still think it resolves at a high level and from memory I think it outresolves the Adam monitors I heard, but only by a small margin. This really surprised me, but I think most studio monitors just aren't that resolving. Maybe ATCs, which I haven't properly heard yet.
The imaging and coherency also suffer from the transition from a 7" woofer to the AMT. I don't think the woofer can quite keep up with the AMT. This is an area where I would expect the HEDD Type 20 and Type 30 to be noticeably better. I got to hear the HEDD tower mains for a short period of time at High End and from memory those were markedly better in that respect. The imaging just ends up lacking depth, something that is helped a bit by the Lineariser, but not as much as I'd hoped. The crossover also seems to be very steep between the woofer and tweeter (effectively 36dB/oct acoustically), which I think hurts the coherency.
Subjectively the HEDD still manages to sound like a very fast speaker, no doubt helped by its lightweight AMT. It's also a much more complete speaker than the LS50s, with very respectable bass. While I wished the bass tuning was a bit deeper at times (which could be achieved by plugging one of the ports to effectively lower the tuning frequency) to help with some boominess I got both on the stands and on my desk, the bass quality here is on another level compared to the other speakers. The bass is taut and authorative, impressively so for a speaker of its size. The 3rd order bass distortion was impressively low at higher levels in my measurements. Generally distortion was low at higher levels. These are speakers you can push hard and they won't give up on you.
Timbre is somewhat two-sided: The AMT seems light on its feet and the woofer seems to have a somewhat thicker timbre to it. It's still a good compromise. I'm glad they didn't put in a metal woofer here since the AMT is so smooth sounding. It actually makes for a pretty good match.

Hiss: It's audible from most normal listening distances, but not bothersome. In my room I could hear hiss at up to 4m distance. On the desk it was very much audible at all times, but I never found it annoying. I measured the hiss at roughly 10dB(A) at 1m distance. I don't think it's a problem, but those sensitive to hiss will probably be better off with passive speakers. I don't think many active speakers have less hiss than these. To put things into context the hiss I heard on my desk was subjectively similar to the hiss of the HD800 + Vali 1 combination.

KEF LS50 Meta:
It's been a while since I last heard the LS50, but immediately the Meta seems to fix my main complaint about the LS50: Its tonality. The LS50 Meta is IMO a very neutral sounding speaker with a very lifelike midrange reproduction. If I'm nitpicking I'd say the air region past 16kHz is a little elevated for my liking and obviously they lack bass extension. In my measurements I also saw a small 1-1.5dB dip around 3kHz.
What really stands out is how transparent and resolving these speakers sound. If I'm honest I don't remember any KEF speaker being this resolving and I have some experience with the KEF Blade and Reference 1, aswell. That may sound like an outlandish claim and it could be, but I expect the other ranges to be upgraded with the metamaterial tech in the future.
It makes listening to it very involving, much more so than I expected. It's also the smoothest coaxial I've heard from a frequency response perspective. There are no treble response issues even listening to it on-axis, something that is a big issue with many other coaxials (including my DIY one).
Where it disappoints the most is, of course, bass response. I find the bass quality disappointing even when considering its small size. The HEDDs have a much more substantial, more realistic bass response. They're truly on another level compared to the puny LS50s. This is where the coaxial suffers from its smaller surface area (compared to other 5" drivers) and its special surround optimized for treble response. OSMTs likely beat these in their bass performance.
However the midrange stayed free from congestion at all but the highest SPLs. This is an area where the Meta is supposed to be an improvement over the original LS50 aswell. My measurements show very low distortion in the midrange and treble from 200Hz on at all levels.
The LS50 is well known for its imaging and I'm not sure if the LS50 Meta is as good as the LS50 here. It may be due to the different crossover with what seem to be 24dB/octave slopes instead of 12dB/octave acoustically, but I found it somewhat lacking in depth, especially when compared to my OBs. Not really a fair comparison and it definitely beats the Type 07 both in image precision and depth placement, but images seemed a bit diffuse and spread out aswell.
Timbre is pretty life-like on these speakers to me, especially considering the metal drivers. There doesn't seem to be much metal sound in the midrange as is so often the case with metal midranges (the DreiZwo doesn't fare too well here, either) and the tweeter strikes a good balance between being too smooth and being too hard sounding. The cohesion between both drivers is also very, very good.


With FIR filters that linearize the magnitude and phase response:

HEDD Type 07:
The Lineariser plugin is one of the selling points of the Type 07. Unfortunately, while I liked what it did to the treble response (lowers the upper treble), I thought the correction for the upper midrange dip was a bit overdone. I still prefer the sound with the Lineariser plugin, but I also made my own file to be used with Equalizer APO. It still corrects the phase response, but with a slightly different target for the frequency response magnitude.
To my ears the effect of correcting the phase response was very pronounced with these speakers. Subjectively they sound even faster, much more cohesive and the imaging improves drastically. It's still no match for the LS50 in terms of imaging, but it's much more realistic this way. I also really liked being able to fill in the upper midrange and dial back the upper treble a bit. Otherwise the upper treble from the AMT can get somewhat overbearing at times.

KEF LS50 Meta:
The frequency response correction here was very minor: +1dB at 2800Hz, Q4 and -2dB at 16kHz, Q4. The effect of the frequency response changes alone is very subtle, but the change from fixing the phase response is not subtle. As with the Type 07 the imaging becomes more pin-point and with better depth-localization. Images that seemed stretched out in their depth are now concentrated to a point. The effect isn't as strong as with the HEDDs, but it's there nonetheless.

Quick notes on desk usage:
HEDD Type 07:
Bass got a bit too thick on my desk in the corner of my room. I didn't try, but I'm sure plugging one of the ports would've been a good idea. At these close distances the integration between the woofer and the AMT suffered a bit, but not as much as I'd expected. Also the AMT is almost high enough as is without extra speaker stands when leaning back in my chair.

KEF LS50 Meta:
I ended up using the outer foam to make the diameter of the port smaller, lowering the tuning frequency. It's a very nice touch that KEF offers multiple bass tunings this way. Speaker desk stands are absolutely necessary. As expected the coaxials work very well on the desk, still offering superb imaging even in cramped listening environments. Listening at closer distances solves some of the power handling issues these speakers have with deep bass at higher volume levels.

Conclusion:
As you might've guessed, I'm keeping the LS50 Metas. I'm still not entirely sure about it since the performance upgrade from my DreiZwo Sat is actually not night and day, while the price difference is very large. Especially considering the amp I'm driving them with on my desk and the acoustics there it's sort of a waste. But the LS50 Meta is the upgrade in microdynamics and overall transparency I was looking for. I think it's also a good reference to keep around and compare other speakers against. While its bass performance is nothing to write home about, the midrange and treble are very good considering its price.
Next post: measurements

 

Reserved for measurements. Probably tomorrow.

Frequency response:

Spoiler: Frequency response

KEF LS50:


HEDD Type 07:

Distortion:

Spoiler: Distortion

KEF LS50 Meta:


HEDD Type 07:


Note that the HEDD's distortion measurements are somewhat limited by the HEDD's noise, especially the 70 and 75dB plots. Also note that with each increase of 5dB SPL the scale also changes by the same 5dB. Where the HEDD has somewhat high 3rd order midrange distortion at lower levels, it has stellar distortion results at 90dB. Take that how you will. The KEF has very high distortion below 200Hz, but the midrange and treble distortion are nicely low with very low higher orders. Only my Voxativ OB has less higher order distortion, but it does have 2nd order distortion spikes. More distortion plots here: https://imgur.com/a/3PpN4qg

I think low distortion at 70dB is more important in the midrange than low distortion at 90dB. 90dB in the midrange is very loud. At 700Hz and 70dB SPL the HEDDs have 15dB more distortion than the LS50 Metas.
Also the LS50 Meta has comparable 2nd order distortion to my Voxativ OBs (apart from their distortion spikes), but has 10-15dB higher 3rd order distortion throughout the midrange.

Step response:

Spoiler: Step response

KEF LS 50 Meta


HEDD Type 07:


Note these are in-room plots with reflections visible. Really just to confirm the crossover slopes.

 

BTW, the KEF-Coax drivers can be bought in France if one wants to diy.

https://setelec-shop.fr/148-haut-parleur-kef?order=product.price.desc&page=2

 

Although I wasn't able to do a direct comparison, I've been living with the LS50 Wireless II (active version of LS50 Meta) and KC62 for a couple weeks and they are stunningly good for the price. I don't hear that typical Brit-fi sound characterized by shouty treble and metallic timbre.

 

If you want detail from studio monitors, you need:

1) soft domes, preferably passive with a good amp, or ATC/Quested actives. The typical plate amps are dogshit. No constant directivity, computer modeled waveguide bullshit. On axis. 6.5” woofer to 1” dome or the Questeds. Or three ways w/ the ATC or PMC dome. Most other dome mids suck balls and are overdamped crap. These are thousands of dollars.

Or

2) Old school horns. Not modern JBL bs, old school horns done right, eg the late 90s and 2000s Tannoys with the super tweeters, Radian drivers, Great Plains audio Altecs. These are a couple thousands dollars retail for the drivers alone.

and

3) no cheaping out on interface, converters, cables, volume control, and amps! No class D! No quad cables! No offbrands! No Behringer!

 

Of all the things you rant about, this one is new to me. You don't like star-quad cables?

 

Yes they kill detail and deaden the sound. They are for live use and rf rejection only!

 

Yup, my thoughts exactly. I'm toying with DIY cables, and found the same results. The very best I could do was an aluminum wire (yes, aluminum, for it's corrosion resistance), with custom extruded teflon and air dielectric. Geometry was 9 rectangular, thin solid core wires for XLR cable (5 of them just for ground). I'm trying to do the same with 70% air custom extruded eptfe and a silver/palladium single crystal alloy, but I've been quoted 6 figures for materials 0_0 + laser welding to connectors made of the same silver/palladium alloy. So yes, cables matter, a lot.

@Serious As always, a pleasure to read your thoughts about speakers. The comparisons are very clear, and I'm glad Kef improved the LS50 because it was, in my opinion, an excellent small speaker held back by it's metallic tonality.
The meta material is extremely interesting, as it can bring new possibilities to speaker designs that were just impossible before. For example, to absorb a bass woofer backwave requires a huge sealed t-line, like the B&W Nautilus...with this material, it may be possible to absorb the entire woofer backwave without having to spend a fortune in a custom mold or a huge wood translam cabinet. I'm curious about how they'll trickle up this tech to their more advanced models, such as Blade. That speaker, with some "metamaterial" care and some crossover mods should be end game, IMHO.

 

Will you be 'servicing' your Metas so we can get a closer look at the gadget?

 

Curious, what are your recommendations for quality cables, interconnects and speaker ? I’ve been reading/listening to options but always open to opinions.

 

Yup, I probably should've bought the old LS50 driver instead of the Tang Band W6-2313. The measurements at hifi-sound.de look quite a bit better. Just look at the distortion of the TB Coax. Unfortunately they don't seem to have the Meta drivers (yet).

I've tried prying off the ring hiding the driver screws, but didn't manage to do so. One thing I could see is that while the original LS50 had a coated surround (as can be seen on the site numbercube linked), the LS50 Meta doesn't. Even so the response in the midrange seems smoother, both in the NRC and stereophile measurements. Maybe they found another way to smoothen the midrange response without relying on coating the surround (which can lower the efficiency).

I'd expect a metamaterial absorber for bass frequencies to become very, very large aswell. Not sure it could help at bass frequencies.

I especially wonder about an updated R3 and Reference 1, aswell. I auditioned the Reference 1 and R5 with my gear and they just don't resolve very well. No idea if it's the metamateral, but something about the new LS50 is different from the other KEF speakers.

How do you like the KC62? It's the first sub I've seen which finally uses an inverted driver model in its DSP to lower distortion and according to the KEF specs it works really well. From 20% to 5% at 20Hz and roughly 90dB in-room.

 

You got the correlation right, but not the cause. The problem isn't the waveguides, the problem is that they're usually employed by companies who have a bad case of measurement myopia. They focus only on the simple measurements, often skimp on drivers/electronics and rely too much on DSP-esus to take the wheel. Having talked to some engineers, they straight up confess that they're afraid to venture into the immeasurable even if their gut tells them that it sounds more right that way. We can probably thank the line art adoration club for it.

TAD? Sure. But not much else. Horn design even now is super hard and more often than not their forte isn't qualities present in good hifi dome tweets. Usually impedance curves are a dead giveaway for nasties, if you can get/measure them.

Here's a DIY'er sweetheart, the B&C DE250:



Like wtf is that top octave shite? For a hifi dome that wouldn't fly in a 100 years. And good luck doctoring that out with any kind of filtering, your best bet is to just get old and lose HF hearing. Now for something tastier, the FaitalPro HF1440:



See the absence of any crud over 2kHz? Some guy at DIYA did comparisons with their prized Be TAD's and ended up preferring this. Oh, and did I mention that the HF1440 is a chonky 3.4" coil driver? It'll happily do 120dB continuous at 1m, if you'll let it. The problem is availability and the fact that it's around 400$. Drivers like these find themselves only in five figure systems.

 

1) that is true but the waveguides tend to color the higher treble themselves. They refuse to go by math even though going by the theoretical math developed to level cities in ww2 is what made digital telephones possible in the 60s and digital audio in the 80s. Accounting for theoretical possibilities is what made good drivers possible in the 80s and 90s and good electronics in the 60s-90s.

for dsp, they are trying to defeat the laws of physics and failing. Almost all good, aka actually working, dsp comes from people who know about and respect immeasurables. The people who don’t or claim it doesn’t matter, make mediocre gear and processes. Like the audio scientists on ASR and Reddit reject dither. Well if you reject dither, the math is wrong which means that you don’t level the city, the cruise missile misses the terrorists, and the stereo image is destroyed. Not that they hear it.

2) Horn and coax type design done right. There is very little a most of it is just drivers, and most of it is very expensive. Great Plains audio drivers, radians, the last tannoys pre behringer with the super tweeter, TAD. Most diy designs suck and they have no idea what they’re doing and sound worse than manufactured speakers with cheaper parts. Most horns are pretty much “detail free”

 

OUTDOOR KEF LS50 META MEASUREMENTS:

Port half blocked, which gave more energy at 40Hz. Wanted to get my own measurements to see how they'd compare to the others. FWIW I got less funky wiggles in the midrange than Amir with his "better than anechoic" NFS system.

Could be the peak at 2kHz was a problem and that's why there are dips at 1.5kHz and 2.7kHz.










Distortion (as per above):




I also attached an extended range CSD for fun.
Impedance: Look elsewhere, lol.

 

I realized they're now gone, so I uploaded them on my page. The Canton Reference 3K is similar to the HEDD with highish D3.

https://www.seriousaudiodesign.de/blog-etc/speaker-distortion

 

I just realized the Spectrogram visualisation has one major advantage over the CSDs: There is no hidden iniformation and as such you can see exactly where the peak energy is. This allows you to get an idea of the group delay caused by the filters of a multi-way speaker such as the KEF LS50 Meta.


Neat.

 

I took those measurements with the Ragnarok. IIRC. However the Ragnarok really isn't a measurement amplifier and will quickly distort driving low impedance speakers to high SPLs:
https://www.superbestaudiofriends.org/index.php?threads/schiit-ragnarok-distortion-analysis.14136/
As a result the distortion measurements here aren't accurate. I realized this after looking at Amri's measurements where he got much less D3 than I did.

I did a quick test at 85dB on my desk:


Very low D3 considering it's a metal driver. Distortion at 300Hz to 1kHz is slightly higher than Hathor all around, but past 1kHz it's much lower. Also bass distortion is quite high, as expected.