KEF LS50 thoughts

[PENG]

I'm not sure its a problem with directivity or resonance based on the whole set of NRC measurements.

Consider a cleaner view of the on axis FR from the distortion graph:
View attachment 23502
It seems to me the tweeter is clearly tipping things up between 2kHz and 5kHz relative to the mid/woofer level. It also seems to me that off axis, those relative proportions are maintained pretty well.
View attachment 23503
Drop the tweeter level 4dB or so, and all you'd really see is that off axis dip around 4kHz, and the rise afterwards as a result of the tangerine phase plug.

If that was due to resonance, what kind? Cabinet related or cone related or something else? The rise to 2K didn't really fall off (just very slightly,yes) after it peaked, and the impedance was rather low for it to be cone resonance. I really think it is more likely to be like you said, some sort of voicing, i.e. by design.

 

[TheWarrior]

No, I wanted to ignore the part of the measurement just above the part that you claim is a resonance, the part where it dips, not peaks. Doing so makes the peak look more like a consistent rise in treble/tweeter response, which is consistent with Steve's observations.




A coax in a symmetric cabinet radiates the same vertically as horizontally. In this case I think the cabinet isn't quite symmetric, but it's close enough to say that the vertical responses will be very similar to the horizontal responses. So we have a very good idea what the vertical response is for this speaker. If there's any doubt, look at JA's vertical plots for frequency response differential. It's nearly identical to the equivalent angles for the horizontal response. There are no huge dropouts in the crossover region.

You can't say that at all about dome-over-cone speakers.




Huh? This was the plots for the LS50. It's flat within +/-3dB on axis, and at 15 and 30 degrees off axis. Its listening window is as well. Same goes for the R100.

Surely you're not saying speakers need to be within +/-3dB at 45, 60, and 75 degrees off axis?

Counting from below 80Hz, well, show me any speaker - especially a small monitor - that's flat within +/-3dB below 80Hz. You've moved the goalposts substantially.

In the present text, any cone or dome based transducer will not radiate equally in all directions. JA's measurements are in-room at lower resolutions, according to Fig.4 and 5 on Stereophile. Despite that, there is still a clear difference between vertical and horizontal - although confusing as to why he didn't go 90deg vertical.

Sorry for the confusion on models. But I would not use the word flat to describe the LS50.

 

[Beave]

In the present text, any cone or dome based transducer will not radiate equally in all directions.

Why not? What do you mean by "in the present text?" Are you saying a woofer radiates differently into the vertical plane than in the horizontal plane? Same with a midrange? Sure, there are some drivers for which that is true - AMTs, ribbon tweeters, planar midranges, etc - but most cone or domes (without waveguides) radiate consistently in all directions.

When a speaker is put into a cabinet, things change. They change even more when there are stacked drivers overlapping in frequency ranges they cover.

But that doesn't apply to the coax in the KEF. The driver, in an infinite baffle, radiates symmetrically, ie, equally in all directions. In the LS50 cabinet, some small changes to that happen. But they are small.

JA's measurements are in-room at lower resolutions, according to Fig.4 and 5 on Stereophile. Despite that, there is still a clear difference between vertical and horizontal - although confusing as to why he didn't go 90deg vertical.

You have eagle eyes - or an active imagination - if you can compare his +/-30degree plots with his +/-90 degree horizontal plots and see a "clear difference." The speaker's output barely changes over the +/- 30 degree window, whether horizontally (as seen by the NRC plots) or vertically (as seen by JA's plots).

Sorry for the confusion on models. But I would not use the word flat to describe the LS50.

So your +/-3dB is good enough, except when it's the LS50 being discussed?

I agree it isn't the flattest I've seen. But I've seen far worse, too. But coming from the world of DACs, I wouldn't use flat to describe any speakers frequency response!

 

[PENG]

So your +/-3dB is good enough, except when it's the LS50 being discussed?

I agree it isn't the flattest I've seen. But I've seen far worse, too. But coming from the world of DACs, I wouldn't use flat to describe any speakers frequency response!

That's the point I wanted to make too, and I would agree the LS50's FR fits the +/- 3 dB approximately (may be 4, but then may argue 2.5) criteria, if there is such criteria. Regardless, +/- 3 dB in the listening window is actually quite flat relatively speaking if one reference to all the speakers measured by SoundStage Network, Stereophile and S&V.

Here's another review with lab measurements:
http://i.nextmedia.com.au/avhub/australian-hifi_reviews_2014_2014-01_kef_ls50_anniversary_speakers_review_and_test_lores.pdf

The reviewer claimed +/- 2.5 dB.

So far we have 3 sets of graphs from 3 different labs, Brent Butterworth of the S&V did one too but I couldn't locate the graphs, he also had high praise of the FR, especially the off-axis ones. None of the reviews highlighted the 2 KHz rise as a resonance issue. After hearing so much about that so called 2 Khz resonance, I hope owners won't be affected by expectation bias going forward.

 

[TheWarrior]

Why not? What do you mean by "in the present text?" Are you saying a woofer radiates differently into the vertical plane than in the horizontal plane? Same with a midrange? Sure, there are some drivers for which that is true - AMTs, ribbon tweeters, planar midranges, etc - but most cone or domes (without waveguides) radiate consistently in all directions.

When a speaker is put into a cabinet, things change. They change even more when there are stacked drivers overlapping in frequency ranges they cover.

But that doesn't apply to the coax in the KEF. The driver, in an infinite baffle, radiates symmetrically, ie, equally in all directions. In the LS50 cabinet, some small changes to that happen. But they are small.



You have eagle eyes - or an active imagination - if you can compare his +/-30degree plots with his +/-90 degree horizontal plots and see a "clear difference." The speaker's output barely changes over the +/- 30 degree window, whether horizontally (as seen by the NRC plots) or vertically (as seen by JA's plots).



So your +/-3dB is good enough, except when it's the LS50 being discussed?

I agree it isn't the flattest I've seen. But I've seen far worse, too. But coming from the world of DACs, I wouldn't use flat to describe any speakers frequency response!

If you take a raw driver, not loaded in a baffle, it will measure more consistently in either plane. Give it a baffle and all bets are off. Then the crossover design can cause inconsistent radiation as well. The other factor is that rooms are reflective spaces - the off axis sound is the dominant factor. So fully understanding what the speaker sent out requires a full orbit of measurements around the speaker - no speaker radiates sound equally in all directions (except designs that were purpose built to do so).


The LS50 costs $1500, it better be able to perform below 80 hz. Thats why I say it does not maintain that 6dB range.

That's the point I wanted to make too, and I would agree the LS50's FR fits the +/- 3 dB approximately (may be 4, but then may argue 2.5) criteria, if there is such criteria. Regardless, +/- 3 dB in the listening window is actually quite flat relatively speaking if one reference to all the speakers measured by SoundStage Network, Stereophile and S&V.

Here's another review with lab measurements:
http://i.nextmedia.com.au/avhub/australian-hifi_reviews_2014_2014-01_kef_ls50_anniversary_speakers_review_and_test_lores.pdf

The reviewer claimed +/- 2.5 dB.

So far we have 3 sets of graphs from 3 different labs, Brent Butterworth of the S&V did one too but I couldn't locate the graphs, he also had high praise of the FR, especially the off-axis ones. None of the reviews highlighted the 2 KHz rise as a resonance issue. After hearing so much about that so called 2 Khz resonance, I hope owners won't be affected by expectation bias going forward.

One set of measurements is low res and in room. The other is anechoic - comparing those graphs is really only useful to JA to help position his speakers (assuming he uses a higher resolution that doesn't mask resonances). But even in room, that resonance exists (what else would cause such a rise in amplitude, anechoic?) - Sorry I didn't record my phone call with Floyd, but Beave verified his comments from another forum. To be fair, many loudspeakers have resonances built in to them, this one happens to be the first that I called out immediately upon hearing it.

But in a thread asking for thoughts on a product, there's been a pretty extraordinary amount effort applied to convincing the masses that this resonance A, does not exist, and B, is inaudible or shouldn't be considered a factor. The anechoic measurements cannot be disputed, nor can mine and others listening experiences.

That is why I said previously, when I thought this was concluded, the take away should be that people go out and LISTEN and decide for themselves!

 

[PENG]

The LS50 costs $1500, it better be able to perform below 80 hz. Thats why I say it does not maintain that 6dB range.

That's your opinion. My in room measurements show the LS50 does maintain that range to below 80 Hz and I use it with a subwoofer most of the time.

Sorry I didn't record my phone call with Floyd, but Beave verified his comments from another forum. To be fair, many loudspeakers have resonances built in to them, this one happens to be the first that I called out immediately upon hearing it.

Not really, I thought his comments involved the typical caveat of "not enough info..." or something like that as I don't remember the exact quoted wording.

The anechoic measurements cannot be disputed, nor can mine and others listening experiences.

I DO NOT recall anyone disputing the anechoic measurements so far. You referred to the freq rise to 2 kHz as a "resonance" and you are very sure about it, while others are not so sure if it is due to other reasons. Not that it matters, but facts are facts, the measurements/graphs are not disputed on this thread. The shape of that little peak doesn't look like a resonance to me at all, but unlike you, I am not entirely sure about it. That's not a dispute on the graph itself, just different interpretation. As Dennis said, "the jury is still out on that one."

That is why I said previously, when I thought this was concluded, the take away should be that people go out and LISTEN and decide for themselves!

You probably could pick that "resonance" out just by listening to it because of expectation bias effects. In real world applications (not in an anechoic chamber) the LS50 is quite flat in FR. My own in room measurements don't even show that much of a peak and they don't sound bright to me at all, though obviously that's in the subjective territory.

 

[ematthews]

Well, I can definitely vouch for the bass of the Phil 3's. The highs of the ribbon are superb, and offer the most realistic reproduction of stringed instruments I have ever heard, including on beryllium, which is just barely ousted.

My only complaint of the ribbons is that I feel they audibly compress the sound of complex movie soundtracks, where beryllium really shines. My perception was that there was just too much being asked of the ribbon to reproduce, particularly for center channel duties.

You KNOW. Sometimes I feel that way with poorly recorded music and ribbons.

 

[yepimonfire]

I DO NOT recall anyone disputing the anechoic measurements so far. You referred to the freq rise to 2 kHz as a "resonance" and you are very sure about it, while others are not so sure if it is due to other reasons. Not that it matters, but facts are facts, the measurements/graphs are not disputed on this thread. The shape of that little peak doesn't look like a resonance to me at all, but unlike you, I am not entirely sure about it. That's not a dispute on the graph itself, just different interpretation. As Dennis said, "the jury is still out on that one."



You probably could pick that "resonance" out just by listening to it because of expectation bias effects. In real world applications (not in an anechoic chamber) the LS50 is quite flat in FR. My own in room measurements don't even show that much of a peak and they don't sound bright to me at all, though obviously that's in the subjective territory.

I heard it and I knew very little about the ls50 when I demoed it. Our ears are very good at picking out the direct sound of the speaker from the rooms contribution.

Sent from my LM-X210(G) using Tapatalk

 

[TheWarrior]

That's your opinion. My in room measurements show the LS50 does maintain that range to below 80 Hz and I use it with a subwoofer most of the time.



Not really, I thought his comments involved the typical caveat of "not enough info..." or something like that as I don't remember the exact quoted wording.



I DO NOT recall anyone disputing the anechoic measurements so far. You referred to the freq rise to 2 kHz as a "resonance" and you are very sure about it, while others are not so sure if it is due to other reasons. Not that it matters, but facts are facts, the measurements/graphs are not disputed on this thread. The shape of that little peak doesn't look like a resonance to me at all, but unlike you, I am not entirely sure about it. That's not a dispute on the graph itself, just different interpretation. As Dennis said, "the jury is still out on that one."



You probably could pick that "resonance" out just by listening to it because of expectation bias effects. In real world applications (not in an anechoic chamber) the LS50 is quite flat in FR. My own in room measurements don't even show that much of a peak and they don't sound bright to me at all, though obviously that's in the subjective territory.

Diffraction does not cause a 5db gain in amplitude, across a whole octave. It is a resonance. The only thing in question is exactly what part(s) of the speaker are causing it. And that requires more data.

Again, I looked at the measurements AFTER hearing the speakers. As did others. If you don't hear any issue, then enjoy your speakers!

 

[TheWarrior]

You KNOW. Sometimes I feel that way with poorly recorded music and ribbons.

I'd say thats true with any transducer! But I will add that my complaints were never with music, only action movie type things.

Again, I was being very, very nitpicky when I compared it to a beryllium tweeter, but the difference was audible. And as I am designing a custom center channel, a RAAL and a Beryllium Illuminator are about the same cost - the challenge to me is to get it to match the timbre of the Phil's.

Lots of outdoor testing will be happening over the next few months!

 

[PENG]

Diffraction does not cause a 5db gain in amplitude, across a whole octave. It is a resonance. The only thing in question is exactly what part(s) of the speaker are causing it. And that requires more data.

Again, I looked at the measurements AFTER hearing the speakers. As did others. If you don't hear any issue, then enjoy your speakers!

I don't know about you but I have a pretty good understanding of the definition and nature of resonance, not just in speakers but in all sorts of things. If you insist that is a resonance, though it does not look like one, you are entitled to your opinion but it is not necessarily fact. The best I can agree to is Dennis saying "the jury is still out...."

You won't hear from me again, not on this topic.

 

[Steve81]

@PENG @TheWarrior

FWIW, I dropped a line to Erin from Medleys Musings to see if he has any thoughts on the subject, given that he ran the LS50 drive unit through his battery of tests. Figure his input may be worth a little more than trying to speculate further one way or the other. He's a member here, so hopefully he's watching

 

[PENG]

I heard it and I knew very little about the ls50 when I demoed it. Our ears are very good at picking out the direct sound of the speaker from the rooms contribution.

Sent from my LM-X210(G) using Tapatalk

You heard what? That there was a "resonance" at 2 kHz? Wow, take a look of the many more speakers measured at the Stereophile, S&V and NRC facilities that are uglier and you'll see that +/- 3 dB is in fact very good. I am picky with speakers too otherwise I wouldn't end up with some many of them in my rooms and basement. Unlike amps, preamps, dacs, even the good ones are bad relatively speaking.

 

[PENG]

@PENG @TheWarrior

FWIW, I dropped a line to Erin from Medleys Musings to see if he has any thoughts on the subject, given that he ran the LS50 drive unit through his battery of tests. Figure his input may be worth a little more than trying to speculate further one way or the other. He's a member here, so hopefully he's watching

The NRC graph shows an approximately -2 dB trough between around 870 hz to 1500 hz, then it ramps up from 1500 hz to approximately 2000 hz by about 4 dB, it basically becomes flat to about 5,000 hz. That's not being disputed or speculated as it is there for all to see. We just are not sure whether that 4 dB rise are due to some sort of "resonance". It will be great if your expert can clear that up with explanation and supporting information.

I was looking for something that I would consider more obvious indication of a peak due to resonance, and found the following CSD plot. To me that one is clear cut, whereas the LS50's is not clear enough for us call it with certainty, though as you put it, we can speculate and only speculate.

http://novo.press/understanding-speaker-specifications-and-frequency-response/

 

[Beave]

If you take a raw driver, not loaded in a baffle, it will measure more consistently in either plane. Give it a baffle and all bets are off. Then the crossover design can cause inconsistent radiation as well.

What you said about drivers and baffles is just what I said. Before it seemed you were saying that applied to drivers, but you now seem to have backtracked to include drivers in baffles. Well, that can be true, but depends on the driver and the baffle. A concentric driver(s) with a symmetric baffle radiates symmetrically in the hor and vert planes. A dome over cone does not.

But how does a crossover cause inconsistent radiation? It doesn't by itself.

The other factor is that rooms are reflective spaces - the off axis sound is the dominant factor. So fully understanding what the speaker sent out requires a full orbit of measurements around the speaker - no speaker radiates sound equally in all directions (except designs that were purpose built to do so).

Kinda like the LS50 being discussed (assuming by 'all directions' you mean in the forward firing sense).

The LS50 costs $1500, it better be able to perform below 80 hz. Thats why I say it does not maintain that 6dB range.

There are plenty of small bookshelf speakers that don't meet your newly-moved-goalposts of being flat below 80Hz. Better tell Revel that their M105 (also $1500/pair and not as often discounted) isn't up to your standard. Better tell Dynaudio that their Excite X14 doesn't meet it either. Nor does Revel's way-more-expensvie Ultima Gem2. There are countless others. Expecting great performance below 80Hz in a small bookshelf speaker is simply unrealistic.

 

[Beave]

@PENG @TheWarrior

FWIW, I dropped a line to Erin from Medleys Musings to see if he has any thoughts on the subject, given that he ran the LS50 drive unit through his battery of tests. Figure his input may be worth a little more than trying to speculate further one way or the other. He's a member here, so hopefully he's watching

I looked at his measurements to see if there is a driver resonance, but his plots aren't really high enough resolution to see for certain. But what I did notice - and mentioned earlier - is that the tweeter response peaks right around where the LS50's NRC measurements show the peaking - at 2kHz, and from there to around 4-5kHz.

It sure would be nice to see plots of the midwoofer and its crossover, and the tweeter and its crossover. JA does that on speakers with buy-wiring connectors, but he doesn't do it for speakers that only have single-wiring connectors.

 

[yepimonfire]

You heard what? That there was a "resonance" at 2 kHz? Wow, take a look of the many more speakers measured at the Stereophile, S&V and NRC facilities that are uglier and you'll see that +/- 3 dB is in fact very good. I am picky with speakers too otherwise I wouldn't end up with some many of them in my rooms and basement. Unlike amps, preamps, dacs, even the good ones are bad relatively speaking.

I heard coloration in the midrange. Unfortunately, we can take lots of speakers measuring +-3dB and test them side by side and they will sound totally different. Even side by side speakers measuring +-2dB will sound different. We can theorize why this is exactly, but it's not always apparent why.

If you like the ls50s, great, I don't, and they don't sound neutral and transparent to me, and they have a narrow, boxed in sound. They could sound different near field as I previously stated, but I haven't had a chance to try them that way. I also have no need for a near field speaker.

I'm not a fan of klipsch sound or any other sound signature, I'd rather a speaker add as little of its own sound to the audio as possible. Assuming Toole is correct, most people will pick speakers that are neutral over ones that aren't, without preference to any sonic signature.

Sent from my LM-X210(G) using Tapatalk

 

[Beave]

... Unfortunately, we can take lots of speakers measuring +-3dB and test them side by side and they will sound totally different. Even side by side speakers measuring +-2dB will sound different. We can theorize why this is exactly, but it's not always apparent why.

Absolutely. Think of it this way: Even with two hypothetical speakers being +/-2dB, one might be +2dB from 80Hz to 1kHz, then -2dB from 1kHz to 20kHz. The other might be the opposite, -2dB from 80Hz to 1kHz, then +2dB from 1kHz to 20kHz. They will differ by 4dB in each range! They will definitely sound different, with the former being 'warmer' and the latter being 'brighter.'

 

[PENG]

I heard coloration in the midrange. Unfortunately, we can take lots of speakers measuring +-3dB and test them side by side and they will sound totally different. Even side by side speakers measuring +-2dB will sound different. We can theorize why this is exactly, but it's not always apparent why.

If you like the ls50s, great, I don't, and they don't sound neutral and transparent to me, and they have a narrow, boxed in sound. They could sound different near field as I previously stated, but I haven't had a chance to try them that way. I also have no need for a near field speaker.

I'm not a fan of klipsch sound or any other sound signature, I'd rather a speaker add as little of its own sound to the audio as possible. Assuming Toole is correct, most people will pick speakers that are neutral over ones that aren't, without preference to any sonic signature.

Sent from my LM-X210(G) using Tapatalk

You are stating the obvious that I think everyone can agree to. I have not said much about the LS50's sound at all so far. I do not listen to them from far away, normally about 8 ft. I like them for my application, but I don't like them better than my other speakers. As an example, I find my 1028 Be more transparent, not because they cost 5X. I also find my R900 more open, funny thing is they also cost more.

My comments are about TheWarrior's typical (sort of a pattern) quick conclusion on something he has some knowledge in, in this case "resonance", and I am not sure he fully understands the nature and cause of the claimed resonance either.. Regarding the +/- 2 to 3 dB being relatively flat is just factually speaking, I did not mean to say the speaker will sound good just because they meet the 3 dB criteria. I thought Beave brought that up first.. If you listen to them in an anechoic chamber, they may, and should, sound more similar if they are within +/-3 dB than +/-5 dB (still not a sure bet). I don't suppose you use Harman's turntable set up for you AB sessions, right?

Now, if we are talking about +/- one dB then it is a different story.

Your claim that the LS50 is colored is a matter of opinion (now maybe I am stating the obvious), ask Dr. Toole what he meant by "neutral"? Remember the Infinity Primus 360/363, that by Harman standard were "neutral"? I bought a pair after reading the reviews, not the subjective parts but the analytical parts, only to give them away a few months later. I seem to remember I wasn't the only one, ask ADTG if he did the same.

For argument sake, all speakers could be said to be colored, or neutral, in relative sense but of course I what you are saying and I read many of those articles at the Harman website too so I am within you 100% on that one. I am just saying that it is also all relative, and a highly subjective matter to say which one is more colored than the other.

 

[PENG]

Absolutely. Think of it this way: Even with two hypothetical speakers being +/-2dB, one might be +2dB from 80Hz to 1kHz, then -2dB from 1kHz to 20kHz. The other might be the opposite, -2dB from 80Hz to 1kHz, then +2dB from 1kHz to 20kHz. They will differ by 4dB in each range! They will definitely sound different, with the former being 'warmer' and the latter being 'brighter.'

He stated the very obvious, but still, good to demo it with examples for the few who may have doubts!