KEF LS50 Bookshelf Speaker Review

[BYRTT]

Hi @aarons915,

As preference rating plots by member MZMK into post #2 shows, normalize and you get 2kHz back and 4,5kHz disapear, looks clear in below overview that especiallly polars change alot when normalized where for example for Neumann KH 80 it makes nearly no difference. Graphs are based 10º steps as amirm shared so there is a big difference for the non normalized polar where he showed impressing 1º step resolution for LS50 .

 

[laurelkurt]

I have a pair of these. With a sub, they are awesome.

I have the LS50Ws with 2 subs in the TV room. I also think they sound great even at fairly loud levels. The imaging and especially vocals sound excellent. I also can now actually understand around 60% of what Mick Jagger is singing. Pretty impressive in that regard...I also went on to buy R900s. R600C, and R3s for my bigger basement system.

 

[amirm]

This Spin is fairly close to other measurements of the LS50 but one oddity I notice is the clear bump at 5k that I haven't seen in other measurements. Most seem to show an elevated region from 2-5k on axis but as you go off-axis, the 5K region smooths out while the 2k Resonance remains.

If I am following you correctly, that difference caused me to run the tests a second time with the same outcome. It could be some sample differences. Without having the identical speaker tested multiple ways, we will be guessing in this regard.

 

[laurelkurt]

Rythmiks are good, but at $399 on Amazon Prime, the SVS SB 12 NSD is hard to beat. You could get two for the price of one Rythmik or four for the price of two Rythmiks....

I own one SB12NSD. I'd definitely recommend them, especially 2 at the price. My only complaint is that the auto on isn't sensitive enough. I just leave it on always for that reason.

 

[aarons915]

Hi @aarons915,

As preference rating plots by member MZMK into post #2 shows, normalize and you get 2kHz back and 4,5kHz disapear, looks clear in below overview that especiallly polars change alot when normalized where for example for Neumann KH 80 it makes nearly no difference. Graphs are based 10º steps as amirm shared so there is a big difference for the non normalized polar where he showed impressing 1º step resolution for LS50 .

Yeah I saw that but I'm not talking about normalized plots, the Soundstage are probably the best measurements we had of the LS50 until this Klippel measurement and you can see a similar response except the 2-5k region is slightly different. It's not a huge deal and Amir is probably right about sample differences, oddly enough I bought a new pair of the Black edition about a year after my original pair and that region measured a bit different than my originals also, I figured it might have had something to do with the tweeter dome being painted black but that's just speculation.

 

[BYRTT]

@aarons915, over at C52 thread member SDX-LV hinted a link to university measurements for LS50 in relative high resolution of 5º steps and one can download them all in one zip folder, link is here https://www.princeton.edu/3D3A/Directivity.html . Have no experience or meaning of sample differences but you have a point for your Black edition being one year newer measured a bit difference . Think it could look like we get some better resolution from that wonder scanner than we used to and we need to adapt like it or not, below is adjusting same ratios as that university measurements covering 0,5-20kHz-, horizontals then verticals, NFS at top with 10º steps resolution shared at post #1, university measurements at bottom with 5º steps resolution.

 

[Kaiju]

You should really consider measuring Triangle Titus EZ costs cost similar to LS50.

 

[maty]

[Polish] https://audio.com.pl/testy/stereo/zespoly-glosnikowe/2517-triangle-titus-ez

to English:

* https://translate.google.es/transla...reo/zespoly-glosnikowe/2517-triangle-titus-ez

* https://www.translatetheweb.com/?re...reo/zespoly-glosnikowe/2517-triangle-titus-ez

 

[hardisj]

@aarons915, over at C52 thread member SDX-LV hinted a link to university measurements for LS50 in relative high resolution of 5º steps and one can download them all in one zip folder, link is here https://www.princeton.edu/3D3A/Directivity.html . Have no experience or meaning of sample differences but you have a point for your Black edition being one year newer measured a bit difference . Think it could look like we get some better resolution from that wonder scanner than we used to and we need to adapt like it or not, below is adjusting same ratios as that university measurements covering 0,5-20kHz-, horizontals then verticals, NFS at top with 10º steps resolution shared at post #1, university measurements at bottom with 5º steps resolution.

View attachment 48032

Extra data is nice. But I don’t necessarily see a big benefit in extra steps. The Klippel NFS page states it can measure as low as 1 degree increments so 5 degrees is possible but I don't know if the extra time to do so is worth it.

Also, keep in mind the one thing you’re seeing in the 3D3A Labs’ spectrograms that makes it look cleaner is they are using only a color steps of 7 (each step is 3dB). Which covers up some problem trends that a larger set of colors/higher resolution might help identify. Looking at the below examples where I plotted some data in matlab in 10 vs 20 color steps you can see how this makes a difference.

10 color gradients:

20 color gradients

That said, a finer range of values would be more important, meaning my legend really doesn't need to extend 80dB and that a tighter range of maybe 30dB would be more appropriate, but you get the idea here. I like their range... but I'm of the mindset that maybe they should have had each color step be representative of 1dB increments rather than 3dB increments. My $0.02, though.

 

[jaykay77]

To me these are the most overhyped speakers I’ve ever seen...Listening to them was the biggest let down!
I wanted to love them and almost got caught up in the hype.
Thanks for telling it like it is

 

[aarons915]

To me these are the most overhyped speakers I’ve ever seen...Listening to them was the biggest let down!
I wanted to love them and almost got caught up in the hype.
Thanks for telling it like it is

I felt the same way originally, I thought they sounded like crap in the showroom, the only reason I gave them a try at home was because of the hype. They were a completely different speaker in my smallish room and most importantly crossed over to subs. Since then I've EQ'd the 750Hz peak and the 2-5k region and they sound even better. Some would say why not just get a better measuring speaker to begin with and normally I would agree but there just isn't a speaker out there with even dispersion in all directions and with such a solid, diffraction-free cabinet, so I try to do what I can to make it work.

 

[BYRTT]

Hi @thewas_ or good morning above EQ settings x16 is alternative to using convolution engine, it should make below prediction before your room and furniture set in, on axis improve but listening 20º off axis help EQed or not, polars are not normalized so they represent on axis and EQ filters look help link contour lines inside +/-45º sector together, have fun .

 

[BogdanR]

@amirm
In the review you said:
“The impedance chart is presented to figure out if the amplifier is "difficult" to drive”
I believe you wanted to say speaker is difficult to drive.
Cheers!

 

[bigjacko]

View attachment 48189

Hi @thewas_ or good morning above EQ settings x16 is alternative to using convolution engine, it should make below prediction before your room and furniture set in, on axis improve but listening 20º off axis help EQed or not, polars are not normalized so they represent on axis and EQ filters look help link contour lines inside +/-45º sector together, have fun .

View attachment 48190

Is it good to eq speaker at high frequency?

 

[BYRTT]

Measurement plus listening will tell at one spot in space any EQ can work perfect relative to what a microphone hear high frequency or not, EQ exercise here is based on 71 measurements covering horizon and vertical axis for ASR's sample of LS50 in 10º steps, so as graphs show theoretical it should work, well at least for that sample.

 

[thewas]

Is it good to eq speaker at high frequency?

Yes, but only if its not linear before but has smooth directivity and measurements can be trusted.

 

[thewas]

Hi @thewas_ or good morning above EQ settings x16 is alternative to using convolution engine, it should make below prediction before your room and furniture set in, on axis improve but listening 20º off axis help EQed or not, polars are not normalized so they represent on axis and EQ filters look help link contour lines inside +/-45º sector together, have fun .

Thank you again very much , I applied them to my EQ Apo and compared them to no correction, a correction based on just the listening window of the same ASR measurements (to flat) and to my current correction based on moving microphone measurements to a Harman curve.
The good news is that they all don't sound very different and in my first tries my preference sequence is MMM > BYRT > listening window > no EQ.
The comparison with MMM isn't fair of course as the others don't consider the influence of close reflection surfaces below 1Khz, but even when I tried ignore the lower region I like the MMM correction a bit more as it gives a bit more sparkle above which may be also something related to my room acoustics. When I find time I will also try to combine your correction with room correction below the transition region.

 

[Fone]

Beautiful design. I love the coherent, time-aligned speaker arrangement. Great "reviews" but something is wrong with the ls50 IMHO. I have listened to these at least a dozen times.

I heard KEFs 3 driver bookshelf speaker last weekend and quite liked it (the R3)

 

[napilopez]

Beautiful design. I love the coherent, time-aligned speaker arrangement. Great "reviews" but something is wrong with the ls50 IMHO. I have listened to these at least a dozen times.

I heard KEFs 3 driver bookshelf speaker last weekend and quite liked it (the R3)

I think the Uni-Q arrangement really shines in three way designs, where the midwoofer doesn't have to handle so much so much of the range. I do like the R3 a lot.

 

[laurelkurt]

I think the Uni-Q arrangement really shines in three way designs, where the midwoofer doesn't have to handle so much so much of the range. I do like the R3 a lot.

The midwoofer is just the mid, and a good one at that, in the larger Kefs.