KEF LS50 Metas - dips around 3 & 6 kHz on both speakers - fault or am I measuring wrong?

[riiich88]

I bought KEF LS50 Metas a while back, but never quite fully engaged with them. I recently started experimenting with them (swapped amp, got a wiim ultra). Then in the process of adding a sub, I measured them this weekend and there's a big dip around 3 kHz on both speakers. Smaller ones again around 6 kHz and 9 kHz. My wife also pointed out the vocals sounded recessed, and I thought that listing to Netflix through them, the vocal sounded thin. I can't say though if this is something that's been there from new or has crept in.

Things I've tried, none of them changed it:

- moved the speakers forward, more toe in
- took the coffee table out
- blanket under the speaker, blanket around the mic
- speaker on the rug, off the stands
- measured at 20 cm, 1 m and 2.8 m
- mic pointing up and pointing at the speaker
- left and right on their own - same on both
- swapped the Nord amp for an old valve amp - same
- swapped the WiiM for a different DAC - same
- all EQ off, sub off

Then I put a Bose Bluetooth speaker in exactly the same spot with the same mic. Smooth through all three, no dips.

Blue is the KEF, pink is the Bose.

Mic is an EM-01 with the cal file loaded. SPL isn't calibrated so ignore the absolute numbers, I'm only looking at the shape.

So either I'm making a beginner mistake I can't see, or something is wrong with both speakers. Any ideas?

Best,
Rich

 

[DanielT]

Is there any warranty left on them?

Could your KEF LS50 Metas be broken?
On the other hand, it seems a bit odd if both to be broken—if separate measurements of each speaker show the same result, it would be strange for both to have failed or—for some reason—lost the exact same amount of performance.

Can you measure (ohms) each driver using a multimeter?

Have you ever played them really loud and fed them more power than they can handle?

My hypothesis (or rather, just a question) that the speakers might be broken could very well be wrong. I’m just throwing it out there.

 

[kemmler3D]

Looks like you are missing 10-20 dB in treble, to me that says damaged tweeter or something is wrong with the input, like inverted somewhere...

Agree with @DanielT that it would be unusual for both tweets to have the same problem like this, are you sure the amp is ok?

 

[b73i57dc3f]

Looks to me like the tweeters might be broken. 2.1 kHz is the crossover frequency according to KEF, and that's pretty much exactly where the response starts to drop off in your measurement.

t would be strange for both to have failed or—for some reason—lost the exact same amount of performance.

Wouldn't be all that strange for the tweeters to be broken in both speakers.

 

[DanielT]

Wouldn't be all that strange for the tweeters to be broken in both speakers.

riiich88 might have blown both tweeters during a high-volume listening session. It’s not exactly uncommon for speakers to get blown out when listening at high volumes.

@riiich88 Have they always sounded the way you described?
Sorry, I missed what you said:...but never quite fully engaged with them.
Did you buy them new or used?
(The previous owner might have blown the tweeters.)

 

[IngMishael]

that sounds like a killed tweeter.

 

[Duke]

Between 2 kHz and 10 kHz there are three dips at 3k, 6k, and 9k. Notice the pattern: Three dips, 3 kHz apart.

This could be a reflection with a path length of about 2.25 inches longer than the direct sound path, with the reflection energy arriving out-of-phase with the direct sound at those frequency intervals. And that reflection distance makes me think it might be the reflection around the perimeter of the coaxial driver.

You might try running your measurements again from about 20 degrees off-axis. If it's a reflection from the driver's perimeter, the off-axis microphone will smear out the reflection in time and those dips should largely fill in.

The reflection around the perimeter of a coaxial or round horn or waveguide is part of the game, and imo it should NOT be fixed with EQ because it's only happening at that microphone location. Change the geometry by moving back to a normal listening distance and the path length geometry changes enough to shift the dip pattern north. But even then if you correct for it via EQ, you simultaneously ruin all of the off-axis sound. Imo the better strategy is to sit off-axis by 15 degrees or more, and the dip pretty much goes away even in the direct sound.

Regarding the rest of the KEF's curve, I never trust "in-room" measurements. In this case, the cone's widening radiation pattern below the crossover frequency results in a lot more reflection energy arriving in the cone's passband, which misleadingly boosts the in-room curve in that region. (There could also be something else going on, but I still mistrust in-room measurements.)

 

[kemmler3D]

Between 2 kHz and 10 kHz there are three dips at 3k, 6k, and 9k. Notice the pattern: Three dips, 3 kHz apart.

This could be a reflection with a path length of about 2.25 inches longer than the direct sound path, with the reflection energy arriving out-of-phase with the direct sound at those frequency intervals. And that reflection distance makes me think it might be the reflection around the perimeter of the coaxial driver.

You might try running your measurements again from about 20 degrees off-axis. If it's a reflection from the driver's perimeter, the off-axis microphone will smear out the reflection in time and those dips should largely fill in.

The reflection around the perimeter of a coaxial or round horn or waveguide is part of the game, and imo it should NOT be fixed with EQ because it's only happening at that microphone location. Change the geometry by moving back to a normal listening distance and the path length geometry changes enough to shift the dip pattern north. But even then if you correct for it via EQ, you simultaneously ruin all of the off-axis sound. Imo the better strategy is to sit off-axis by 15 degrees or more, and the dip pretty much goes away even in the direct sound.

Regarding the rest of the KEF's curve, I never trust "in-room" measurements. In this case, the cone's widening radiation pattern below the crossover frequency results in a lot more reflection energy arriving in the cone's passband, which misleadingly boosts the in-room curve in that region.

This is a good point, but OP mentioned measuring at different distances, and I think the missing energy in the treble is too extensive to be caused by reflections off the cone. The whole tweeter range is already down by 5-10dB even before we start looking at the dips.

 

[kyuu]

It is odd that the subsequent dips are harmonics of the first dip at around 2.8kHz, and then the tweeter level recovers to just about where it should be at 10kHz before dropping off again. Have you tried factory resetting the speakers to ensure that nothing weird is unknowingly enabled in the speakers' settings?

 

[Cote Dazur]

You could play a frequency sweep and measure just the tweeter with the highest frequency.

 

[kemmler3D]

It is odd that the subsequent dips are harmonics of the first dip at around 2.8kHz, and then the tweeter level recovers to just about where it should be at 10kHz before dropping off again. Have you tried factory resetting the speakers to ensure that nothing weird is unknowingly enabled in the speakers' settings?

I looked again and OP mentioned an amp, these are the passive LS50 Metas, not the wireless. Rules out a factory reset, at least.

 

[DanielT]

Take a paper towel tube:

Hold it against the tweeter and listen to see if you hear anything/the tweeter.

The driver doesn't seem too difficult to unscrew. Pry off the trim ring covering the screws and unscrew them. The video shows the cables connecting to the tweeter; test with a multimeter.

 

[Curvature]

I bought KEF LS50 Metas a while back, but never quite fully engaged with them. I recently started experimenting with them (swapped amp, got a wiim ultra). Then in the process of adding a sub, I measured them this weekend and there's a big dip around 3 kHz on both speakers. Smaller ones again around 6 kHz and 9 kHz. My wife also pointed out the vocals sounded recessed, and I thought that listing to Netflix through them, the vocal sounded thin. I can't say though if this is something that's been there from new or has crept in.

Things I've tried, none of them changed it:

- moved the speakers forward, more toe in
- took the coffee table out
- blanket under the speaker, blanket around the mic
- speaker on the rug, off the stands
- measured at 20 cm, 1 m and 2.8 m
- mic pointing up and pointing at the speaker
- left and right on their own - same on both
- swapped the Nord amp for an old valve amp - same
- swapped the WiiM for a different DAC - same
- all EQ off, sub off

Then I put a Bose Bluetooth speaker in exactly the same spot with the same mic. Smooth through all three, no dips.

Blue is the KEF, pink is the Bose.

Mic is an EM-01 with the cal file loaded. SPL isn't calibrated so ignore the absolute numbers, I'm only looking at the shape.

So either I'm making a beginner mistake I can't see, or something is wrong with both speakers. Any ideas?

Best,
Rich


View attachment 558741

Measure one speaker at a time with a frequency dependent window set for 2 cycles. https://www.roomeqwizard.com/help/help_en-GB/html/analysis.html

This will give you the direct sound above 1kHz.

To be clear:

1. Put speaker A on a stand in the middle of the room.
2. Put the mic up to 1m away.
3. Sweep & measure.
4. Replace speaker A with speaker B.
5. Sweep & measure again.

Distance doesn't matter and neither do a lot of other niceties because we are looking for gross defects.

 

[Duke]

This is a good point, but OP mentioned measuring at different distances, and I think the missing energy in the treble is too extensive to be caused by reflections off the cone.

The apparently missing treble energy may be a separate issue from the dips.

If the dips move slightly up in frequency as the microphone is moved further away, that would be consistent with a reflection causing cancellation dips.

The whole tweeter range is already down by 5-10dB even before we start looking at the dips.

Yes it looks like a pretty big discrepancy.

I wonder whether that discrepancy is consistent regardless of microphone distance, or does it change as the direct-to-reverberant ratio changes with distance?

 

[kyuu]

I looked again and OP mentioned an amp, these are the passive LS50 Metas, not the wireless. Rules out a factory reset, at least.

D'oh. Anytime I see LS50 I immediately think of the active ones and completely forget that a passive version exists.

 

[Gruesome]

I hope the 116dB SPL measurements in your REW list (#10 and #11) were not taken with the speakers, because that probably would have broken them. 116 dB SPL corresponds to about 1000W for a speaker like the LS50 Meta with a sensitivity of 85 db SPL per Watt at 1 meter; the speaker specs list a maximum sound pressure level of 106 dB at 1m, corresponding to about 125W. I do not know whether this speaker can actually withstand that power level in a narrow frequency band, as would happen during a frequency sweep.

 

[riiich88]

Thanks so much for all the replies.

Have you ever played them really loud and fed them more power than they can handle?

I hope the 116dB SPL measurements in your REW list (#10 and #11) were not taken with the speakers, because that probably would have broken them. 116 dB SPL corresponds to about 1000W for a speaker like the LS50 Meta with a sensitivity of 85 db SPL per Watt at 1 meter; the speaker specs list a maximum sound pressure level of 106 dB at 1m, corresponding to about 125W. I do not know whether this speaker can actually withstand that power level in a narrow frequency band, as would happen during a frequency sweep.

Alas yes, it was with the speakers - I was using Claude to guide me through it and probably mis-read what it was trying to tell me (regarding signal to noise). However I went back through the earlier measurements and the pattern was still there.

Agree with @DanielT that it would be unusual for both tweets to have the same problem like this, are you sure the amp is ok?

Have tried 2 different amps with the same result, so pretty sure its not that.

You might try running your measurements again from about 20 degrees off-axis. If it's a reflection from the driver's perimeter, the off-axis microphone will smear out the reflection in time and those dips should largely fill in.

Tried taking a couple of off-axis measurements but got the same result.

Hold it against the tweeter and listen to see if you hear anything/the tweeter.

Measure one speaker at a time with a frequency dependent window set for 2 cycles. https://www.roomeqwizard.com/help/help_en-GB/html/analysis.html

Will try both of these later.

I wonder whether that discrepancy is consistent regardless of microphone distance, or does it change as the direct-to-reverberant ratio changes with distance?

I went back through the other measurements at different distances and it's pretty consistent. The gap is also consistent at different test volume levels.

I am going to try getting them looked at under warranty. I can't find the receipt, but I found the demo date in a WhatsApp message so hopefully the dealer can dig it out. It was 2022 and looks like the driver is covered by a 5 yr warranty. If not covered I'll try opening them up and trying a multi-meter.

 

[phoenixdogfan]

Thanks so much for all the replies.


Alas yes, it was with the speakers - I was using Claude to guide me through it and probably mis-read what it was trying to tell me (regarding signal to noise). However I went back through the earlier measurements and the pattern was still there.


Have tried 2 different amps with the same result, so pretty sure its not that.


Tried taking a couple of off-axis measurements but got the same result.



Will try both of these later.


I went back through the other measurements at different distances and it's pretty consistent. The gap is also consistent at different test volume levels.

I am going to try getting them looked at under warranty. I can't find the receipt, but I found the demo date in a WhatsApp message so hopefully the dealer can dig it out. It was 2022 and looks like the driver is covered by a 5 yr warranty. If not covered I'll try opening them up and trying a multi-meter.

I am experiencing the same thing with my LS 50 Metas. It happened, I think, when I got the Asci A6b's and I moved the Metas to surrounds. Anyway, I think I might have gotten some kind of power surge through the Behringer A500 that somehow took out both tweeters. I noticed they sounded subdued, and when I did the measurement for Dirac, I saw they took a dive around 2k. At first I thought it might have been amp related, so I tested with another amp, and had the same issue. I subbed in my OG LS 50's which I had mothballed, and the highs came right back. I guess I will get the Metas fixed sometime soon, but I'd kind of like to know what that's like, especially since they're off warranty. And I'd like to have a ballpark estimate of what it costs as well as whether the tweeter or the whole drive unit needs replacing when the tweeter blows. So maybe you could keep us posted on your effort?

 

[OldHvyMec]

They are crossed at 2.1KHz if the tweeters were blown, there would be a drastic drop from 2.1Khz to no sound at all by 3KHz or even less. Tweeters that are blown usually make no noise at all; you can often smell something with a sniff and look and see something blown out through the waveguide, or the VC housing or wires will look burnt/blue.

Bad tweeters seldom produce sound @ 15KHz

If it were a physical crossover problem again, the inside of the speaker could have a smell unless it was ages ago, and you would likely see a problem on the XO tweeter circuit board/section. I've never seen 3, 6, 9KHz dips and then peaks from anything blown.

An open space and testing will likely show a vast improvement, or you have a weird mic thing going on.

Lots of luck ay.

Regards

 

[riiich88]

And I'd like to have a ballpark estimate of what it costs as well as whether the tweeter or the whole drive unit needs replacing when the tweeter blows. So maybe you could keep us posted on your effort?

I contacted the service centre and will take them in next week. Warranty on the drivers is 5yrs, electronics 2yrs. I bought mine in 2022 so hoping it’s the drivers and they overlook a couple of toddler dings on them! Will post back on how I get on.

Tweeters that are blown usually make no noise at all; you can often smell something with a sniff and look and see something blown out through the waveguide, or the VC housing or wires will look burnt/blue.

An open space and testing will likely show a vast improvement, or you have a weird mic thing going on.

I took the driver out of one speaker and had a look - nothing smelly or burnt looking. It scratched the paint on the way out and was a pain to reseat so reluctant to do too much poking about inside whilst still a chance of warranty cover. Might give the outside test ago, although it’s not completely open space. The Bluetooth speaker tested fine so don’t think something off with the mic.