KEF Reference speakers - channel balance question

[AToMe]

Hello all, I purchased a pair of ex-demo KEF Reference 3 speakers a year back. I was too busy with my work the whole year around that I hardly had much time left to listen to them for more than 2-3 songs per week

Recently when I was bit free, I noticed that, both the speakers were not sounding the same in the midrange area. To me the tone of voice from left and right are different.(KEF has L/R markings on the speakers). It feels like one (L)sounds bit louder in upper midrange(or shouty) compared to the other. I kept both of them at the same spot and did bit of measurements using UMIK mic. When I measured one speaker I marked the position on the floor where it was kept. and also marked the position of the second speaker on the floor. Then I swapped them for the measurement. Measurements were done using the same channel of the amp. My pre amp has digital volume control, so the volume level was not altered even accidentally during the measurements.

I feel there is a problem with one of them(Possibly with L even though it look more flatter). The Left speaker sounds like its vague, bit uncontrolled, shouty and bit lacking in treble details. The R speaker sounds very agile, easy to listen and tonally balanced. I am really torn here as with imaging, some times the vocals appear bit off centre(towards left) than dead centre. It doesn't happen with every single vocal.

Is this measurement within KEF's standard reference tolerance? How much should be the tolerance between L/R? Or is this worth a repair?

What does those 2nd harmonic and 3rd harmonic distortion measurement from REW mean? I noticed the numbers are different between the channels here.

 

[AToMe]

Can some other KEF Reference user confirm if this measurements are within the tolerance?

 

[NTK]

Not a KEF Reference user and don't know about the previous Reference series. But for the current Reference Meta, pair matching is supposed to be within 0.5 dB (assuming of course the measurements are done properly).

https://assets.kef.com/documents/reference/kef-reference-meta-manual.pdf

 

[BrokenEnglishGuy]

Iooks Iike the measurement was taken wrong, dont worry about it.
But when you are doing rew at Ieast 85dB is necessay

 

[phoenixdogfan]

To me, it looks like the right channel is 2-3 db flatter above 1000 hz. I would not consider that acceptable in a $15k speaker system. Especially a flagship system from a premier brand. If anything, I think increasing the volume would only exacerbate the imbalance.

 

[goat76]

If the measurements are correctly done, there is definitely something wrong with one of the speakers. I think you should contact the dealer or KEF depending on if there is still a warranty on your speakers and figuring out how to solve the problem.

 

[BrokenEnglishGuy]

guys pease take a momment to read correctIy the graphs...........................

 

[goat76]

guys pease take a momment to read correctIy the graphs...........................
View attachment 383423

I’m not sure what you referring to but the SPL is sufficient at around 70-75 dB. Don’t look at those numbers as they just show the SPL where the mouse pointer happened to be placed on the graph, which is at 19.8Hz. Is that what you are looking at?

REW themselves recommends a level of around 75dB.

 

[Tom C]

Fundamental for L is 24dB, 495%(!) THD
Fundamental for R is 38dB, THD 70%
Something seams amiss, no?

 

[ban25]

Contact your dealer and then KEF about a warranty repair for the left speaker. They should take care of you.

 

[BrokenEnglishGuy]

Fundamental for L is 24dB, 495%(!) THD
Fundamental for R is 38dB, THD 70%
Something seams amiss, no?

the measurement was incorrecty taken... was human error for what i see.
No ONE here read the post? come on guys! this is a shame. PeopIe here insist about technnicaIIy support from KEF and the measurements from the OP is not even near to be correcty taked
Is kind of funny how everything is focusing about the Iitte thing in the mids but the whoIe bass is destroyed XD pus the odd distortion test

 

[NTK]

The distortion numbers shown were at 19.8 Hz (with the SPL at 23.5 dB). The signal was so low (and probably overwhelmed by ambient noise) at this low frequency that the distortion numbers from REW were meaningless. Using the SPL curves of the fundamental is perfectly fine for the purpose of comparing the FR's of the left and right speakers, especially as the harmonic distortion curves (the THD, H2, H3, etc.) are all more than 20 dB below fundamental and are nowhere to be seen for the frequency range of concern.

However, I would like to know why the LF roll off for both speakers started at ~200 Hz.

 

[AToMe]

the measurement was incorrecty taken... was human error for what i see.
No ONE here read the post? come on guys! this is a shame. PeopIe here insist about technnicaIIy support from KEF and the measurements from the OP is not even near to be correcty taked
Is kind of funny how everything is focusing about the Iitte thing in the mids but the whoIe bass is destroyed XD pus the odd distortion test


View attachment 383568

My binding posts were removed and the measurements were taken only for the UNIQs. I am fine to know if I had taken the measurements wrong. Can you let me know why it is necessary to have at least 85 db's for a measurement? Also it was a question here, why the distortion is so high in the measurement for one channel. I did the sweep multiple times for that channel and the results were "more or less" the same. Can someone post here the right method for UMIK measurements for identifying the frequency difference between left and right channel?

 

[AToMe]

guys pease take a momment to read correctIy the graphs...........................
View attachment 383423

Can you point out why this could be wrong? I am really sorry , I don't mean to offend you here, I only want to know what is wrong here and how can correct here. I don't want KEF's technical support to waste their time on this, if this is a problem by my measurements. I can hear a problem, and that's why I bought the mic to confirm it. Also the measurements show a deviation in frequency response at the same volume level. I understand home measurements can be tricky. I am getting those high percentages on the speaker which I think is the one which had the problem, although I never understood what they mean.

 

[goat76]

The distortion numbers shown were at 19.8 Hz (with the SPL at 23.5 dB). The signal was so low (and probably overwhelmed by ambient noise) at this low frequency that the distortion numbers from REW were meaningless. Using the SPL curves of the fundamental is perfectly fine for the purpose of comparing the FR's of the left and right speakers, especially as the harmonic distortion curves (the THD, H2, H3, etc.) are all more than 20 dB below fundamental and are nowhere to be seen for the frequency range of concern.

However, I would like to know why the LF roll off for both speakers started at ~200 Hz.

The main thing here is not the overall frequency response, the important thing is that the microphone wasn't moved between the measurements and the speakers were positioned at the same spot, which seems to be the case as he marked up the position on the floor. If he had made an error with the microphone position or the speaker's positioning, we would likely have seen larger changes in the lower frequency response and not just for the problematic area higher up.



Earlier this year, I made a similar measurement to see if the manufacturer's claimed pair matching of 0.5dB was true for my speakers. I didn't care much about the choice of the measuring spot for the speakers as I figured that the measurements would be affected the same by the room reflections as long as the microphone wasn't moved, and I carefully placed both the loudspeakers at the marked-out spot on the floor. So... don't pay too much attention to the frequency response in the below measurement as it's highly affected by the in-room response, just look at how the responses match each other.

Except for the possibility that the speaker's positions may have a deviation of a few millimeters, I think the manufacturer's claimed pair matching of 0.5 dB is fairly accurate.

ATC SCM40 (v2), left and right speaker frequency response. The same amp channel was used. 1/6 Smoothing.

 

[AToMe]

The main thing here is not the overall frequency response, the important thing is that the microphone wasn't moved between the measurements and the speakers were positioned at the same spot, which seems to be the case as he marked up the position on the floor. If he had made an error with the microphone position or the speaker's positioning, we would likely have seen larger changes in the lower frequency response and not just for the problematic area higher up.



Earlier this year, I made a similar measurement to see if the manufacturer's claimed pair matching of 0.5dB was true for my speakers. I didn't care much about the choice of the measuring spot for the speakers as I figured that the measurements would be affected the same by the room reflections as long as the microphone wasn't moved, and I carefully placed both the loudspeakers at the marked-out spot on the floor. So... don't pay too much attention to the frequency response in the below measurement as it's highly affected by the in-room response, just look at how the responses match each other.

Except for the possibility that the speaker's positions may have a deviation of a few millimeters, I think the manufacturer's claimed pair matching of 0.5 dB is fairly accurate.

ATC SCM40 (v2), left and right speaker frequency response. The same amp channel was used. 1/6 Smoothing.

View attachment 383583

Actually same was my goal. It wont matter to me if it measure flat at where I measured it. I did neither touch the mic, nor change position of anything in the room during the measurement session also I made sure, I stood at the exact same spot during the measurement, as I am also an "object" which can affect the measurement. I used masking tape on the floor to mark the positions of the speaker which was being measured and speaker which was not under measurement. Basically I swapped them, between measurements. But as @BrokenEnglishGuy pointed, if something is wrong in the way I used REW, I am willing to do it all over again. I directly went to disortion measurement option of the REW since it gives both frequency response and disortion in one shot. I did not change any settings between measurements.I was also wondering why the distortion % was so high on one speaker, but by the time I realised it from the screenshot, I had moved the speakers back to their original spots.

It's very difficult as the speakers weigh 51 Kgs each and I need to get help from my neighbour again to move them around. By the way the neighbour was absent in the room while the measurements were being taken.

 

[goat76]

Actually same was my goal. It wont matter to me if it measure flat at where I measured it. I did neither touch the mic, nor change position of anything in the room during the measurement session also I made sure, I stood at the exact same spot during the measurement, as I am also an "object" which can affect the measurement. I used masking tape on the floor to mark the positions of the speaker which was being measured and speaker which was not under measurement. Basically I swapped them, between measurements. But as @BrokenEnglishGuy pointed, if something is wrong in the way I used REW, I am willing to do it all over again. I directly went to disortion measurement option of the REW since it gives both frequency response and disortion in one shot. I did not change any settings between measurements.I was also wondering why the distortion % was so high on one speaker, but by the time I realised it from the screenshot, I had moved the speakers back to their original spots.

It's very difficult as the speakers weigh 51 Kgs each and I need to get help from my neighbour again to move them around. By the way the neighbour was absent in the room while the measurements were being taken.

I don't see that you have done anything wrong here.
An SPL of around 70-75 dB is enough to see if the two speakers have a matched response or not, and as you say, you didn't move the microphone or anything else and just swapped out the speakers to the same position.

As your loudspeakers are heavy as hell, I hope you will be able to just send in both the UNIQ drivers to KEF, so that they can figure out what is wrong with one or the other and make sure you get a matched pair. No matter if this will be on you or if it's going under warranty, I guess you want it fixed either way. But you should of course reach out to the dealer first if you bought it from a store, it's just a bit unfortunate that it took some time before you realized there was something wrong with the speakers.

Sorry @BrokenEnglishGuy, but you are focusing on irrelevant things that have very little to do with what AToMe wants to investigate with his measurements.

 

[AToMe]

I don't see that you have done anything wrong here.
An SPL of around 70-75 dB is enough to see if the two speakers have a matched response or not, and as you say, you didn't move the microphone or anything else and just swapped out the speakers to the same position.

As your loudspeakers are heavy as hell, I hope you will be able to just send in both the UNIQ drivers to KEF, so that they can figure out what is wrong with one or the other and make sure you get a matched pair. No matter if this will be on you or if it's going under warranty, I guess you want it fixed either way. But you should of course reach out to the dealer first if you bought it from a store, it's just a bit unfortunate that it took some time before you realized there was something wrong with the speakers.

Sorry @BrokenEnglishGuy, but you are focusing on irrelevant things that have very little to do with what AToMe wants to investigate with his measurements.

Actually I am happy to be wrong here. Since if something is wrong with my measurement settings, I want to know what is wrong that I can take reliable measurements to KEF for their consideration. I had sent the same measurements to KEF and they asked me to contact the dealer since I did not buy it with the original box and transporting them to the service centre has to be arranged by the showroom people. The person at the store, is extremely nice and he offered me to pick it up and they would send the speaker to the service centre. But the real problem is, I am not sure if this is within their tolerance or not. If the service centre says it's within the factory tolerance, then I would have to bear all the costs from home to Netherlands( I am typing this from Germany) and also the charges for the diagnosis. I saved up for some time to buy these speakers(after reading ASR for some time, I was 100% sure I wanted a KEF). Initially I aimed for R11 and then at the store these were ex-demo units and came with the whole KEF warranty. I paid little bit above the price of R11, and for that price still on theory it was a deal. I had only owned very old speakers prior to this, but fortunately none had the difference in sound between left and right until now. Also, I studied Mediatechnology for some years back and back then I had worked at labs with multiple Genelecs and ME Geithain monitors. I have a fair idea about how a neutral sounding speaker sounds. Since I did not work on the area of loudspeakers, but mostly on binaural audio reproduction, my hands on experience in loudspeaker area is very limited. Definitely less than many people here. And with speaker measurements, using REW - it was my first experience, so I am thinking there should be something wrong in what I am doing too.

 

[AToMe]

The main thing here is not the overall frequency response, the important thing is that the microphone wasn't moved between the measurements and the speakers were positioned at the same spot, which seems to be the case as he marked up the position on the floor. If he had made an error with the microphone position or the speaker's positioning, we would likely have seen larger changes in the lower frequency response and not just for the problematic area higher up.



Earlier this year, I made a similar measurement to see if the manufacturer's claimed pair matching of 0.5dB was true for my speakers. I didn't care much about the choice of the measuring spot for the speakers as I figured that the measurements would be affected the same by the room reflections as long as the microphone wasn't moved, and I carefully placed both the loudspeakers at the marked-out spot on the floor. So... don't pay too much attention to the frequency response in the below measurement as it's highly affected by the in-room response, just look at how the responses match each other.

Except for the possibility that the speaker's positions may have a deviation of a few millimeters, I think the manufacturer's claimed pair matching of 0.5 dB is fairly accurate.

ATC SCM40 (v2), left and right speaker frequency response. The same amp channel was used. 1/6 Smoothing.

View attachment 383583

Which software generates this graph?

 

[goat76]

Which software generates this graph?

It's the “All SPL” window in REW. If you have 2 or more measurements opened you will see them all overlapping each other in that window, and at the bottom you can uncheck the measurements you don't want to see.