Kef R3 Bookshelf Speaker Review (Erin's Audio Corner)

[Frank Dernie]

Nice to see this quantified! And makes me wonder, when you have like LS50 cranked up and the woofer moving way back and forth, what is the effect on the tweeter response? That's the Achilles' Heel of two-way coax design and I've never seen it measured or even discussed much. This is especially so for models like Uni-Q or Uni-Fi, but we might still expect some effect from automotive-style coaxials where the tweeter is out on a bridge in front of a 6x9 or whatever cone.

I'm not sure how you'd move the woofer cone in/out to its extremes other than DC signal, or some kind of physical thing like jamming paper towels on the back side.

The automotive coaxial aren’t Hi-Fi anyway but people have been going on about problems like this in coaxial for decades.

I am sure if it was of any consequence the people who make them would have tested and if they found a problem would either never have gone into production or stopped making them decades ago, it isn’t as if it isn’t a blindingly obvious possibility, so we can be pretty confident it isn’t actually a problem.
IMHO

 

[thewas]

By the way Uni-Q has found its way already into OEM automotive and it seems soon more will come:

KEF delivers first car audio system with Uni-Q® driver array in partnership with Group Lotus

Lotus and KEF have announced a partnership to bring high-end audio into the next generation of Lotus sports cars and premium electric lifestyle vehicles, commencing with the new Lotus Emira.

uk.kef.com

564 Recording Engineer jobs in United States (28 new)

Today’s top 564 Recording Engineer jobs in United States. Leverage your professional network, and get hired. New Recording Engineer jobs added daily.

www.linkedin.com

 

[Sancus]

Nice to see this quantified! And makes me wonder, when you have like LS50 cranked up and the woofer moving way back and forth, what is the effect on the tweeter response? That's the Achilles' Heel of two-way coax design and I've never seen it measured or even discussed much.

I made a post about this here.

The TLDR; is that the manufacturers at least do believe it's significant, and consider it a substantial justification for building 3-way coaxials instead of 2-ways. What few measurements and tests have been done show it as enough to be audible.

That said, there isn't that much data or measurements in public sources, especially for newer 2-ways like the LS50 Meta. But I can't imagine the problem has been completely eliminated, or Kef would not have 3-ways for practically all of their high-end lineups.

 

[KMO]

I made a post about this here.

The TLDR; is that the manufacturers at least do believe it's significant, and consider it a substantial justification for building 3-way coaxials instead of 2-ways. What few measurements and tests have been done show it as enough to be audible.

That said, there isn't that much data or measurements in public sources, especially for newer 2-ways like the LS50 Meta. But I can't imagine the problem has been completely eliminated, or Kef would not have 3-ways for practically all of their high-end lineups.

But most of the reasons for being 3-way are the same as for a non-coaxial speaker - 30Hz to 2kHz is just too much for one driver. You can't get high sub bass levels from a single driver that will still work well at high midrange.

You can't really infer that the 3-way choice is because of the IMD effect. Do coaxial manufacturers even use 3-way designs noticeably more than other people?

 

[Sancus]

You can't really infer that the 3-way choice is because of the IMD effect. Do coaxial manufacturers even use 3-way designs noticeably more than other people?

I didn't infer or extrapolate anything, I literally quoted from both the Genelec white paper which states, and I quote: "Particularly for coaxial designs, a three-way construction can also significantly reduce the issue of Doppler distortion in the coaxial transducers." and from the Harman source who stated, and I quote: "...this effect will probably be mitigated in a 3-way system."

I should note that Genelec(IMO the best coaxial manufacturer in the business) does not even bother to make a 2-way coaxial AT ALL, despite the fact they could definitely make one in the price range of their (not cheap) 2-ways if they wanted to. And that it would certainly be more practical to make a 2-way the size of the 8331A than a 3-way.

 

[KMO]

I didn't infer or extrapolate anything,

You said "I can't imagine the problem has been completely eliminated, or Kef would not have 3-ways for practically all of their high-end lineups."

In other words, inferring that KEF is using 3-way specifically to avoid IMD, and if they'd eliminated it they wouldn't be using 3-way.

 

[Sancus]

In other words, inferring that KEF is using 3-way specifically to avoid IMD, and if they'd eliminated it they wouldn't be using 3-way.

No, not what I meant, they would obviously still have 3-ways for the larger speakers. Rather, if they thought it wasn't an issue I'd expect them to have a small 2-way bookshelf model in the R-series if not the Reference. Similar to how Revel has 2-ways even in their high-end and premium lineups like the M126Be. But Kef doesn't.

For standard speakers, manufacturers usually switch up to 3-way once you reach the ~8" woofer size(or for multiple woofer floorstanders), because that's when directivity matching with a normal tweeter becomes an issue. But 3-way coaxials are common even with small woofers. Kef, Genelec, and Kali all make 3-ways with woofers in the 5"-6.5" size.

Anyway, it is 100% clear to me based on the evidence quoted in my post that doppler distortion is a major *additional* reason to build a coaxial 3-way. On top of all the normal reasons. If you don't agree that's fine, I'm not trying to convince you in particular. I was merely providing an answer to the question of what happens when the midrange/woofer modulates the tweeter response. I've presented the evidence in my other post, and people are free to come to their own conclusions.

 

[KMO]

No, not what I meant, they would obviously still have 3-ways for the larger speakers. Rather, if they thought it wasn't an issue I'd expect them to have a small 2-way bookshelf model in the R-series if not the Reference. Similar to how Revel has 2-ways even in their high-end and premium lineups like the M126Be. But Kef doesn't.

Okay, that's clearer, but isn't the LS50 occupying that product space?

That's their flagship 2-way, and it's somewhere in between the R series and Reference series...

Indeed, its existence is a testament to their belief in 2-way designs - having this "boutique" 2-way which is almost Blade-esque in design concept.

 

[nc535]

Okay, that's clearer, but isn't the LS50 occupying that product space?

That's their flagship 2-way, and it's somewhere in between the R series and Reference series...

Indeed, its existence is a testament to their belief in 2-way designs - having this "boutique" 2-way which is almost Blade-esque in design concept.I

 

[nc535]

And what is their and any number of posters here offering of subs as companions to the LS50 and other 2-ways a testament to?

 

[KMO]

And what is their and any number of posters here offering of subs as companions to the LS50 and other 2-ways a testament to?

The fact that small 2-ways can't do deep bass? Again, nothing to do with the coaxial IMD effect.

I'm just trying to point out there's no evidence that KEF are noticeably avoiding 2-ways compared to others, so you can't infer from this non-fact that there is a significant flaw in 2-way coaxial designs compared to other 2-ways.

 

[nc535]

They aren't avoiding 2 ways; they let you choose/make your own tradeoff and they have price points to cover with a range of products. But the science is clear. If you ask for too much bass from a coax 2-way you will get IMD. So either sit close and/or listen at low levels or get a sub or a 3 way or like me, R3s with subs. I considered LS50s for my desk because I liked the looks but chose the R3 out in the office (not on the desk) because I understood the tradeoffs.

 

[Sancus]

I'm just trying to point out there's no evidence that KEF are noticeably avoiding 2-ways compared to others, so you can't infer from this non-fact that there is a significant flaw in 2-way coaxial designs compared to other 2-ways.

You're right that Kef seems less concerned about this problem than others. Which leads to the fact that Harman tested their 2-ways and found them to underperform relative to what was expected from their(good) spinoramas, and believes it's due to the coaxial IMD problem.

Perhaps their justification for the 2-ways is well, they can't play loud anyways(due to small woofers), so who cares if they get worse than other small 2-way speakers when playing loud. Go up to our 3-ways if you want to play loud. That's a totally reasonable decision. Or "market wants speakers in these price ranges and the most cost effective way to make them is 2-ways, so we'll just do it and if they're not the best that's OK."

I'm not really interested in any further back and forth about this, because it's tangential and doesn't significantly impact the main point that coaxial IMD is a real problem. The fact that Kef makes *some* 2-ways is not an indication they've somehow defeated physics and made it so that a moving waveguide for a tweeter is a good thing, contrary to all available evidence.

 

[nc535]

In hindsight, reviews of KEFs or other coax's should include IMD tests. I don't recall having seen any. Did I miss them? "standard" IMD is multi-tone cross products which can occur on any woofer or mid cone. With a coax we can dynamic frequency response distortion when the mid cone excursion causes treble reflections at its edge, not just when the trim ring isn't positioned correctly. I would love to see measurements of either type.

 

[thewas]

Indeed, its existence is a testament to their belief in 2-way designs - having this "boutique" 2-way which is almost Blade-esque in design concept.

The LS50 was also supposed to be an homage and modern successor of the legendary 2-way LS3/5a as it could be seen in its original white paper and marketing, another reason for its implementation choices. Many audiophiles, especially in some countries in Asia or Europe have a big desire for compact 2-way loudspeakers where they listen usually in not too high distances and not very loud or bass heavy music.

You're right that Kef seems less concerned about this problem than others. Which leads to the fact that Harman tested their 2-ways and found them to underperform relative to what was expected from their(good) spinoramas, and believes it's due to the coaxial IMD problem.

We don't know though which KEF models Harman tested back then, as for example the old LS50 had also some tonality problems. Also we shouldn't that Harman, same like Amir here, usually does mono listening which brings a loudspeaker quite earlier to its SPL limits. Also Harman research showed till now that there is not much correlation between transducer (loudspeaker & headphone) distortion and subjective rankings.

In hindsight, reviews of KEFs or other coax's should include IMD tests. I don't recall having seen any. Did I miss them? "standard" IMD is multi-tone cross products which can occur on any woofer or mid cone. With a coax we can dynamic frequency response distortion when the mid cone excursion causes treble reflections at its edge, not just when the trim ring isn't positioned correctly. I would love to see measurements of either type.

The German online magazine LowBeats does some and there for example the LS50 Meta doesn't really show significant disadvantages to other non-coaxial loudspeakers with similar small woofer size but often even the opposite, exemplary

Of course a 5" outer diameter woofer which has even less radiating surface due to the inner tweeter is not a loudspeaker for loud or far listening, but that imho has more to do with its size than having some extra IMD due to being a coaxial one.

 

[Sancus]

Of course a 5" outer diameter woofer which has even less radiating surface due to the inner tweeter is not a loudspeaker for loud or far listening, but that imho has more to do with its size than having some extra IMD due to being a coaxial one.

How do we know these IMD tests actually detect Doppler distortion which is different from IMD and really what we are discussing? IMD tests in general seem highly variable and nonstandardized.

We don't know though which KEF models Harman tested back then, as for example the old LS50 had also some tonality problems. Also we shouldn't that Harman, same like Amir here, usually does mono listening which brings a loudspeaker quite earlier to its SPL limits. Also Harman research showed till now that there is not much correlation between transducer (loudspeaker & headphone) distortion and subjective rankings.

Sure. I don't think that one quote is a smoking gun or anything. It was more the collection of all the different pieces of evidence that did it for me. I really think arguing individual bits is missing the point.

 

[thewas]

How do we know these IMD tests actually detect Doppler distortion which is different from IMD and really what we are discussing? IMD tests in general seem highly variable and nonstandardized.

According to Klippel who has probably done the most research on any kind of audio transducer distortions:

Multi-tone measurements gives a comprehensive fingerprint of all nonlinear distortion components

Source: https://www.klippel.de/fileadmin/klippel/Files/Know_How/Webinar/IEC*21/EN Handout/KLIPPEL LIVE Series1_Part9 - Intermodulation Distortion - Music is More than a Single Tone.pdf

If you see that paper also addresses Doppler distortion.

 

[Sancus]

According to Klippel who has probably done the most research on any kind of audio transducer distortions:

Multi-tone measurements gives a comprehensive fingerprint of all nonlinear distortion components

Source: https://www.klippel.de/fileadmin/klippel/Files/Know_How/Webinar/IEC*21/EN Handout/KLIPPEL LIVE Series1_Part9 - Intermodulation Distortion - Music is More than a Single Tone.pdf

If you see that paper also addresses Doppler distortion.

Unfortunately I don't have time to read all that now. Are you saying that German magazine uses the Klippel multi-tone test module though? That would add some credibility in my opinion.

That said, 86dbC is far too low level for any sort of distortion test in my opinion :/ If I'm listening to music with 20dB of dynamic range that means my average listening level has to be 66dB to stay within the bounds of this test...

 

[thewas]

Unfortunately I don't have time to read all that now. Are you saying that German magazine uses the Klippel multi-tone test module though? That would add some credibility in my opinion.

From what I see they just use REW which can do the same (I also use it for such), the Klippel module would do probably some more detailed analysis of the results.

That said, 86dbC is far too low level for any sort of distortion test in my opinion :/ If I'm listening to music with 20dB of dynamic range that means my listening level has to be 66dB to stay within the bounds of this test...

Those signal levels are close to the limit of the loudspeakers and quite loud for their size, for example most large floorstanders already distort more at approximately 10 more dB

As said it is obvious that almost all compact loudspeakers 5" are not for parties or very high listening distances, but for distances their directivity suits an immersive listening (usually till approximately 2 meters) you can often listen with quite some satisfying levels around 80-85 dB average, as in well implemented designs shot dynamic peaks are barely and audible problem.

 

[Sancus]

Those signal levels are close to the limit of the loudspeakers and quite loud for their size, for example most large floorstanders already distort more at approximately 10 more dB

Yeah, fair enough. I know the LS50 really don't do well when asked to play loud, I have heard them lol. To an extent that's why I think much of this is moot when regarding that particular speaker.

I do find it interesting that 1-2khz showed 40dB of IMD in this test. That is significantly more distortion than the THD in Amir's test. If that correlation holds at higher levels, it does imply that the IMD may become audible before the THD. Especially since the THD in that region is primarily 2nd harmonic at 96dB.