• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

KEF LS50 Bookshelf Speaker Review

trivium

Active Member
Joined
Jan 26, 2021
Messages
169
Likes
48
84dB/watt is not "hard to drive", speakers which are hard to drive are those with low impedance and/or big phase shifts which demand a lot from an amplifier and some modest amps can not deliver the required current.
A low efficiency speaker with a benign impedance curve is not hard to drive it just needs more power for a given loudness and the small low impedance speakers tend to have a lower maximum loudness simply because the amount of power you can put into a small driver before damage is usually less too.

My LS50s are not hard to drive but they are small, they will go louder than the LS3/5a they replaced but nothing like bigger speakers.

Gotcha well I should be fine with my little amp than, this is why I like this site because you guys use science and not magic, lol. Everyone I’ve talked to said they LS50's need 200+ watts with no reasoning behind their statement, unless im running bass boost (which i wont be, if anything ill run a crossover at 100hz) and listen at normal volumes of 75-85DB I dont see how that adds up. Since this is really a 4 ohm speaker would that not be better for me because my amp puts out 60 watts at 4 ohms instead of 45 at 8 ohms. Its harder on the amp but its designed for it so i dont see the problem here and the fact that its 4 ohm seems beneficial.

im really after a 3D/holographic soundstage and imaging, so im hoping the ls50s have that. Im also looking at the totem Skylight but i heard on here they sound bad, on the flip side the formal reviews i read said they were great and offer up 3 dimensional sound. In either case im less concerned about total accuracy and more interested in the former.
 
Last edited:

Angsty

Major Contributor
Forum Donor
Joined
Apr 11, 2020
Messages
1,862
Likes
2,215
Location
North Carolina, U.S.
im really after a 3D/holographic soundstage and imaging, so im hoping the ls50s have that.
The “holography” of this speaker is largely room and placement dependent. I use mine in a 10’x12’x9’ office with a decidedly non-optimal, asymmetric placement. In that configuration, they seem to be tonally accurate, but not holographic.
 

Skeeter

Member
Joined
Sep 10, 2020
Messages
45
Likes
63
Location
Norfolk, UK
I am struggling to understand how a review methodology that rates products primarily on measurements can measure the KEF LS50’s, get superb results and then give a ‘not recommended’ rating. Is this a case of ‘objectivist fantasy vs subjective reality’?

I‘ve had a pair of these since 2017. Firstly, they are small book shelf monitors. They won’t produce deep bass (anymore than any BBC monitor size speaker will). In a small room with some corner reinforcement however the bass response is pretty remarkable. If you use them in a bigger room, a sub bass is a must. You also need a decent 100 Watt or more amp to drive them because they are not efficient. One thing I learnt is that they sound and image better if the stands are about 500 mm off the floor and not higher.

The measurements speak of an exceptionally well engineered speaker. The response is flat, reflections, edge effects and resonances are all well controlled (John Atkinson’s Stereophile measurements found the same).

So why would you buy them? The imaging on these speakers is superb. I listen to a lot of classical music and older (Ella Fitzgerald, Duke Ellington etc) jazz recordings on CD and vinyl and the LS50’s generate a huge left<>right and front to back sound stage. To folks that listen to rock/pop they may sound a bit soft, but that’s simply because these speakers are very neutral. My other speakers are B&W 703’s. These have a great open, relaxed sound with lots of bass, but the top end is not as good as the KEF’s. If you've ever been to a classical music concert, and you listened to the two speakers side by side, you instantly recognize that one is accurate, the other not.

I have listened to a lot of gear at audio shows both as a visitor and participant showing gear. A lot of speakers nowadays are very forward (Revel, Dali, newer B&W for example). These are great for modern music but don’t handle what I call ’natural acoustic’ material realistically. There are others, like Kerr Acoustics and Leema Acoustics that sound ‘hyper real’, with an impressive etched sound, but no orchestra, brass band or big band sounds like these speakers make them. Great for modern music though.

I use a B&W ASW610 sub bass with KEF’s dialed in for frequencies about 80 Hz and below in my medium sized room. The KEF’s are used for classical and jazz and the 703’s for pop, rock and electronic music. I have a 240W per channel amp, 100W and a 25W class A amp that I use, regularly rotating them out.

if you are a rock/pop music person and like loud music, these would not be my recommendation. For other types of music, with a decent amplifier, they are hard to beat - but please note the comments about room size and use of a sub-bass.
 
Last edited:

Descartes

Major Contributor
Joined
Dec 27, 2020
Messages
2,104
Likes
1,077
This is a review and detailed measurements of the famous KEF LS50 speaker. It was kindly loaned to me by a local member. The LS50 costs US $1,500 but seems to be on sale for US $1,300 on Amazon and elsewhere.

The LS50 nails the look of the speaker with the bright color and shape of the concentric driver:



80% of the success of a speaker in this super crowded market is the industrial design and KEF nails that. The enclosure itself feels like it is solid all the way to its core!

Not much exciting on the back other than nice sized speaker terminals and a port that has benefited from good simulations:


My company became a KEF dealer for a short period of time when the LS50 had just come out. We set it up with a Revel sub optimized for the room with DSP and it sounded good in our little room. Since then I have heard it at shows but no opportunity for careful analysis of it.

Of course there are tons of reviews and measurements of the LS50. The idea of us testing it is to have a reference set of measurements using the same system (Klippel NFS) and graphics as I present for other speakers.

Measurements were performed at around 57 degrees at elevation of 14 feet above sea level. All scents were neutralized in the lab as to eliminate their effect on measurements (had to take a shower myself). To make sure the system was ready for precise results, a five axis laser convolvometer was used to determine the acoustic center of the speaker. Independent lab data is available for the calibration system on request.

Spinorama Speaker Measurements
Our "bible" for how a speaker may sound based on measurements it the spinorama as standardized in CEA/CTA-2034 standard:
View attachment 47771

Focusing on the on-axis graph (in black) and its close relative, the listening window (dashed green line), we see that there is a good sized drop in bass frequencies between 50 and 150 Hz. Post crossover there is a boost in mid-frequencies starting at 2.2 kHz. The ideal here is a flat line and unfortunately we don't have it.

Good news is that the reflections as represented by the dashed blue line has less variation in it so be sure to let those play out instead of covering your walls with absorbers.

Predicted in-room response assuming you let the reflections be therefore looks like this:

View attachment 47772

Doesn't look too bad other than the droop in low bass.

In summary, you are going to have somewhat anemic bass with some exaggerations of frequencies.

Basic Speaker Measurements
The impedance chart is presented to figure out if the amplifier is "difficult" to drive and hence requires beefy amplification:

View attachment 47775

Lowest impedance above 20 Hz is at 216 Hz. Phase angle is high at the same time so that is a difficult load although music energy is also rather low there.

At 40 Hz where a lot of music peaks are the impedance is higher to above 10 Ohm (good). Below it though we dip back down to 5 ohms or less.

Sensitivity is specified at 85 dB at 1 meter/2.83 volts. We get:

View attachment 47776

At mid-frequencies we get there. But in bass, you drop way down to some 75 dB. So don't go running standard calculators online on how loud a speaker gets based on the 85 dB one. Bass is where it is at and there, you need tons more power than 85 dB would indicate.

Advanced Speaker Measurements
For people who like to spend hours on measurements, here is all you need to pass that time.

Let's start with contour plots. In previous measurements, I showed them at 5 degree resolution. If we are going to measurebate, might as well go out with 1 degree granularity to put to shame anyone measuring speakers at 10 degree increments manually! :D

View attachment 47777

View attachment 47800


Here is the waterfall/"CSD" graph using the same scale as I presented in my last review.
View attachment 47781

Distortion Measurements
These are all in-room measurements at the same SPL as the spinorama measurements:

View attachment 47782

Here is the same data but presented in the more familiar THD ratio/percentage:

View attachment 47783

High frequency polar plot
I went up and down the frequency range to find anything interesting to find but there wasn't any. The coaxial driver nicely generates a balloon of sound even around crossover region. I thought we must have some eye candy though so I thought I present this plot at 20 kHz:

View attachment 47784

Science can be beautiful! Unlike the typical measurements performed, we have a full 3-D scan of the speaker radiation so we can present graphs like this with high resolution to show all their details.

A bit on the useful side, the same info plotted in 2-D (inset) shows that our acoustic center (tweeter) matches the measured acoustic center.

Really, really informal listening tests
I had just listened to the Harbeth M30 speakers so I thought I swap the two and listen to the same tracks. Well, this was a let down. There was no bass to speak of. Turning up the volume resulted in distortion as I could visibly see the little woofer trying to keep up. Definitely not a good choice for a large space. A subwoofer with proper EQ is well advised. Overall, I could not get excited over the sound I was getting.


------------
As always, questions, comments, recommendations, etc. are welcome.

Bought on rack to store speakers on. Felt good for a few hours until this morning when half a dozen new and huge boxes arrived! Have to buy another rack to store them! Needless to say, I am not happy spending money on anything I can't eat. So please help make me happy by donating food money using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Thank you Amir for your review!
I am curious how do you explain that these were rated Speakers of the year and have done so well outselling countless other brands?
 

BYRTT

Addicted to Fun and Learning
Forum Donor
Joined
Nov 2, 2018
Messages
956
Likes
2,452
Location
Denmark (Jutland)
...measure the KEF LS50’s, get superb results...

...The response is flat...

No problem LS50 can be a excelent acoustic part'e'ner :).. that said think you should relax a bit call it superb and flat performance on objective paper, it is so that there is a computed Preference Rating list based Amir's acoustic scans where "R3" is rated second "Q350" rated as seventeen "Q100" rated as thirtyeight and for LS50 it is a fiftyseven place, based that analyze one can argue LS50 is not superp or flat on objective paper, here is link for that Preference Rating list (LINK: https://docs.google.com/spreadsheet...i_eE1JS-JQYSZy7kCQZMKtRnjTOn578fYZPJ/pubhtml#).

009_KEF_LS50_2.png
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
2,656
Likes
5,819
Location
US East
I would caution not reading too much into the "Olive preference rating". The preference score should be viewed as a "maximum likelihood estimate" with a considerable amount of variance (or uncertainty). A correlation of 0.86 actually suggests it isn't that precise as a predictor.

Below is the comparison between the measured preference and the predicted preference of the data used in the original study. For example, a "predicted rating" of 5 can be anywhere from a little over 2 to a little under 7 in the actual measured rating. We also don't know whether the distribution will tighten or spread out when more different speaker samples and types are added.

Olive.PNG
 

Descartes

Major Contributor
Joined
Dec 27, 2020
Messages
2,104
Likes
1,077
It seems good measurements are not always a guarantee for commercial success!

Otherwise Revel M106 would win when compared with KEF LS50! I doubt Revel sold as many M106 as KEF did with the LS50.
 

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,403
Likes
5,296
Location
Somerville, MA
It seems good measurements are not always a guarantee for commercial success!

Otherwise Revel M106 would win when compared with KEF LS50! I doubt Revel sold as many M106 as KEF did with the LS50.

KEF is very strong in the brand appeal and industrial design department. Revel speakers are nicely made but do not look very special unfortunately. The JBL offerings are pretty tacky looking in my opinon, especially the HDI series (for the money).
 

Descartes

Major Contributor
Joined
Dec 27, 2020
Messages
2,104
Likes
1,077
KEF is very strong in the brand appeal and industrial design department. Revel speakers are nicely made but do not look very special unfortunately. The JBL offerings are pretty tacky looking in my opinon, especially the HDI series (for the money).

Agreed as I mentioned in an other post Harman needs to buy Sonus Faber and put their engineering inside then they will have winners!

The HDI are so ugly and look so cheap, and yet they are expensive!
 

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
1. The stats part... @NTK already pointed out the model's uncertainty. There are other issues as well, including ambiguous formulas which have been discussed elsewhere on this forum. Not criticizing the researchers here, they did a fine job in a non crowded field but the data set is limited by the practical constraints, its size, listener variation... And even if we had a higher confidence in some "80 percent prefer" result, don't forget you may not be in those 80 percent...

2. If you already own the speakers, why worry about what reviews say? Resale value? With the numbers sold and the speed at which KEF is pushing out improved models, it certainly isn't going to be great. I do own the LS50 btw, along with a bunch of others speakers, some costing 20 times more, and I really enjoy them. No need to worry about anything else imho, if they give you pleasure just relax and enjoy them. Don't let a guy on the Internet, even if it is Amir, or a bunch of numbers, spoil the fun.
 

richard12511

Major Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
4,335
Likes
6,700
I am struggling to understand how a review methodology that rates products primarily on measurements can measure the KEF LS50’s, get superb results and then give a ‘not recommended’ rating. Is this a case of ‘objectivist fantasy vs subjective reality’?
.

The final score is 90% based on the subjective listening portion. A speaker can measure quite poorly and still get a 5/5 (Revel M55XC) if it does well in the subjective listen. Likewise, a speaker can show excellent measurements and get a "not recommended" 3/5(SVS Ultra) or even 0/5(ELAC Unifi 2) if it does poorly in the subjective listen.

I actually prefer it this way, as by far the biggest criticism I hear on other forums is that Amir's recommendations are based entirely on measurements, which couldn't be further from the truth.
 

Descartes

Major Contributor
Joined
Dec 27, 2020
Messages
2,104
Likes
1,077
1. The stats part... @NTK already pointed out the model's uncertainty. There are other issues as well, including ambiguous formulas which have been discussed elsewhere on this forum. Not criticizing the researchers here, they did a fine job in a non crowded field but the data set is limited by the practical constraints, its size, listener variation... And even if we had a higher confidence in some "80 percent prefer" result, don't forget you may not be in those 80 percent...

2. If you already own the speakers, why worry about what reviews say? Resale value? With the numbers sold and the speed at which KEF is pushing out improved models, it certainly isn't going to be great. I do own the LS50 btw, along with a bunch of others speakers, some costing 20 times more, and I really enjoy them. No need to worry about anything else imho, if they give you pleasure just relax and enjoy them. Don't let a guy on the Internet, even if it is Amir, or a bunch of numbers, spoil the fun.

Very well said! We all have different taste and ears!
Now in near field the LS50 sound great but in a big room and 12 feet away from them they don’t sound as good. Yes at this distance one hears the room more than 5 feet away
 

Angsty

Major Contributor
Forum Donor
Joined
Apr 11, 2020
Messages
1,862
Likes
2,215
Location
North Carolina, U.S.
Very well said! We all have different taste and ears!
Now in near field the LS50 sound great but in a big room and 12 feet away from them they don’t sound as good. Yes at this distance one hears the room more than 5 feet away
My LS50’s sound, in near field, remarkably like my much larger Thiel CS6 speakers which also have a coaxial tweeter/midrange. However, in my larger living room, I much prefer the Thiels.
 

goldenears

Active Member
Joined
Oct 10, 2019
Messages
110
Likes
134
I haven't read all the pages of this thread, so apologies if this has already been discussed, but how does the LS50 end up with a preference score of 4.56 when the Q100 gets 5.04?
 

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,739
Likes
16,170
I haven't read all the pages of this thread, so apologies if this has already been discussed, but how does the LS50 end up with a preference score of 4.56 when the Q100 gets 5.04?
Because it is a rather rare case of while having smoothing directivity not having an optimal frequency response due to its crossover, with EQ it can be improved to 6.36, see here.
 

goldenears

Active Member
Joined
Oct 10, 2019
Messages
110
Likes
134
Because it is a rather rare case of while having smoothing directivity not having an optimal frequency response due to its crossover, with EQ it can be improved to 6.36, see here.

OK, interesting. Thanks for that.

Would Audyssey do anything similar to that?
 

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,739
Likes
16,170
OK, interesting. Thanks for that.

Would Audyssey do anything similar to that?
It could, especially if you can manipulate the target curve, basically you mainly need to reduce the 2dB bump between 2 and 5 kHz.
 

goldenears

Active Member
Joined
Oct 10, 2019
Messages
110
Likes
134
OK, great.

I know it would complicate things but maybe there should also be a "preference score with EQ", since that is what room correction that a lot of people will be using does.
 
Top Bottom