loudspeakers L/R matching are sometimes terrible !

[Blumlein 88]

ELAC Adante AF-61 loudspeaker Lab Report

Modern floorstanders as large as the AF-61 typically record sensitivities of 89dB or 90dB so it’s a surprise to find that ELAC specifies 87dB, and that our measured pink noise figure is just under 86dB. Classically, you’d expect this to be accounted for by the AF-61 having greater bass extension...

www.hifinews.com

”Pair matching is disappointing at ±2.0dB, the largest disparities again occurring in the treble.”

True, but this and your previous post the differences are not even a db until past 8 khz or 10 khz. I'm telling you it is at those frequencies a fraction of an inch misalignment will cause these kinds of results. I don't know their methodology. I would like for them to measure one. Take it out of place, put it back and measure it again to show me they can get the same result twice.

And when your mismatch is all above 8 khz, it might make some tiny difference, but your basic stereo image is not going to be effected.

 

[Tangband]

True, but this and your previous post the differences are not even a db until past 8 khz or 10 khz. I'm telling you it is at those frequencies a fraction of an inch misalignment will cause these kinds of results. I don't know their methodology. I would like for them to measure one. Take it out of place, put it back and measure it again to show me they can get the same result twice.

And when your mismatch is all above 8 khz, it might make some tiny difference, but your basic stereo image is not going to be effected.

It can be a measurement fault for sure , but the fact is - nobody but Hifinews has measured this potential problem with matching the L/R to close tolerances , so I wanted to discuss this and also doing a vote poll about how serious this might be for stereo performance.

It seems like the opinions are split in this matter . As a former DIYer, I know that matching drivers with a dsp for correcting each driver is mandatory to get really, really good stereo response results . And believe me - different stock drivers sometimes differ more than 1 dB almost always at the crossover frequency . You need tools to correct this and its best done with a dsp crossover - ie an active crossover .
One such driver with some deviance that needed 1 dB different dsp correction L/R is the Seas er18rnx, that I used in my active projekt the HYBRID. And yes - the difference in sound was clearly audible when listening to good stereo recordings.

 

[Tangband]

I really wonder why theres is split opinions of the need to match the righ and left loudspeaker … ?

In the proffesional world in recording with only two microphones you really need the microphones to be matched the same.

In dac or amplifier testing, if one channel is 2 dB off , its gonna be considered as a headless panther. A broken unit.

I guess some people havent heard about the possible differences between L and R loudspeaker , because almost noone has been testing this except Hifinews.

Sometimes there are very small differences, as they have shown, and thats fine.

But…. If you have bought a pair of loudspeakers recently that you love the sound of - or you have built a good DIY loudspeaker ,- how do you know If the L&R are equal or not ? There might be sound quality gains with individual eq for each speaker with a good dsp.

I suggest that everyone who is unsure about this important issue, do their own measurements.

Maybe there is a sound quality problem, maybe there isnt.

 

[Blumlein 88]

I really wonder why theres is split opinions of the need to match the righ and left loudspeaker … ?

In the proffesional world in recording with only two microphones you really need the microphones to be matched the same.

In dac or amplifier testing, if one channel is 2 dB off , its gonna be considered as a headless panther. A broken unit.

I guess some people havent heard about the possible differences between L and R loudspeaker , because almost noone has been testing this except Hifinews.

Sometimes there are very small differences, as they have shown, and thats fine.

But…. If you have bought a pair of loudspeakers recently that you love the sound of - or you have built a good DIY loudspeaker ,- how do you know If the L&R are equal or not ? There might be sound quality gains with individual eq for each speaker with a good dsp.

I suggest that everyone who is unsure about this important issue, do their own measurements.

Maybe there is a sound quality problem, maybe there isnt.

I don't think there is a split opinion on the need to match them. Maybe some range of opinions about how close they need to be given that your position in the room will make more difference even for first arrival sounds.

I measured my F12 Revels at one time. Up to about 11 khz they overlaid on each other almost perfectly. Above that there were wiggles of a db or so between channels, but I also got that if I measured them a second time. I've measured some 305s I have, and they were roughly the same. I did my measurements on the tweeter axis from 1.5 meters. Maybe I just got lucky. I don't recall measuring this aspect of other speakers.

 

[thewas]

Yamaha NS-1000M loudspeaker Lab Report

Lab Report With speakers of this vintage – even a pair in such immaculate condition as these – there is always a question mark regarding ageing effects. So when I plotted the forward frequency responses of this pair of NS-1000Ms [Graph 1, below] measured at 1m on the tweeter axis and saw the...

www.hifinews.com

And another ….

”Pair matching over the same frequency range is poor at ±2.3dB”

Do you know the age of those loudspeakers?

ELAC Adante AF-61 loudspeaker Lab Report

Modern floorstanders as large as the AF-61 typically record sensitivities of 89dB or 90dB so it’s a surprise to find that ELAC specifies 87dB, and that our measured pink noise figure is just under 86dB. Classically, you’d expect this to be accounted for by the AF-61 having greater bass extension...

www.hifinews.com

”Pair matching is disappointing at ±2.0dB, the largest disparities again occurring in the treble.”

Yes, in the midrange where its important the difference looks to be less than a dB:

 

[Tangband]

I don't think there is a split opinion on the need to match them. Maybe some range of opinions about how close they need to be given that your position in the room will make more difference even for first arrival sounds.

I measured my F12 Revels at one time. Up to about 11 khz they overlaid on each other almost perfectly. Above that there were wiggles of a db or so between channels, but I also got that if I measured them a second time. I've measured some 305s I have, and they were roughly the same. I did my measurements on the tweeter axis from 1.5 meters. Maybe I just got lucky. I don't recall measuring this aspect of other speakers.

I dont say that Revel has a common problem - none of those tested seems to have big problems in hifinews reviews.
At least - you did measure yours, and I guess that you wanted to be sure both speakers were ok .

 

[Tangband]

Do you know the age of those loudspeakers?


Yes, in the midrange where its important the difference looks to be less than a dB:

This is not acceptable differences for me , but maybe for you ? In my opinion, if the measurements are correct, - you gonna hear those differences. It gonna give the listener a less perfect stereo image. A dsp crossover with individual corrections can ( anyway ) easily correct those faults.

One thought - I assume that all real hifi enthusiast always sits in the sweet spot in front of the two speakers, listening in stereo. I sure do. If not , I guess anything goes

 

[BenB]

I have typically always been impressed that a DIYer like me can mount some drivers in a box, measure those particular drivers in that particular box, design a crossover based on measurements, and then build the second speaker and get a near identical result. Then we can share our designs online, and others can build them and again get very, very similar results and great left-to-right matching. In my experience, there's very good consistency in passive crossover parts, and also very good consistency in drivers that are mass produced... until the manufacturer changes something in their process. From run-to-run, there's less consistency in drivers, but within a run of drivers manufactured with the same process consistency is usually very good. Admittedly, I don't use buy-out no-name speakers, and my experience is somewhat limited, so take that for what it's worth.

I'm not sure how helpful it is for testers to measure 2 speakers and for consumers to then assume anything about consistency within the various speaker lines. Two is still a woefully insufficient number for a reliable estimate. We can very easily have consumers avoiding a line with generally very good consistency that happened to include an out-of-spec speaker in the testing. We would also often end up with the appearance of good consistency in a line with poor consistency, just by chance.

 

[Tangband]

I have typically always been impressed that a DIYer like me can mount some drivers in a box, measure those particular drivers in that particular box, design a crossover based on measurements, and then build the second speaker and get a near identical result. Then we can share our designs online, and others can build them and again get very, very similar results and great left-to-right matching. In my experience, there's very good consistency in passive crossover parts, and also very good consistency in drivers that are mass produced... until the manufacturer changes something in their process. From run-to-run, there's less consistency in drivers, but within a run of drivers manufactured with the same process consistency is usually very good. Admittedly, I don't use buy-out no-name speakers, and my experience is somewhat limited, so take that for what it's worth.

I'm not sure how helpful it is for testers to measure 2 speakers and for consumers to then assume anything about consistency within the various speaker lines. Two is still a woefully insufficient number for a reliable estimate. We can very easily have consumers avoiding a line with generally very good consistency that happened to include an out-of-spec speaker in the testing. We would also often end up with the appearance of good consistency in a line with poor consistency, just by chance.

You got a point . I think the only thing one can do is :

1. Realise there might be a problem with the matching of L/R speakers. What we can see in hifinews test might also only be true for that specific pair of speakers - meaning that the speakers that are ” good ” , in reality maybe just got lucky with the pair tested.

So what can be done ?

2. The customer can buy measurement gear and do his own measurements om both L and R speaker . Then buy a good dsp and make the necessary corrections, or rebuild the passive crossover so L and R speaker are within 0,5 dB deviation . Maybe this is the only good scientific solution .

 

[sergeauckland]

You got a point . I think the only thing one can do is :

1. Realise there might be a problem with the matching of L/R speakers. What we can see in hifinews test might also only be true for that specific pair of speakers - meaning that the speakers that are ” good ” , in reality maybe just got lucky with the pair tested.

So what can be done ?

2. The customer can buy measurement gear and do his own measurements om both L and R speaker . Then buy a good dsp and make the nessesary corrections. Maybe this is the only good scientific solution .

Whilst doing one's own measurements is the best way, manufacturers who take pair matching seriously will allocate matched drivers and crossovers to the kits in the factory, and keep records of what went into what. This means that whilst any two pairs may be slightly different, within any one pair they should be closely matched.

It's somewhat easier with active 'speakers, especially DSP based actives as these can be matched after assembly and don't require carefully matched components. With DSP crossovers, this can be easily automated and correction factors loaded into memory.

There's really no excuse for poorly matched loudspeakers except for cost and being bothered.

S

 

[thewas]

This is not acceptable differences for me , but maybe for you ? In my opinion, if the measurements are correct, - you gonna hear those differences. It gonna give the listener a less perfect stereo image. A dsp crossover with individual corrections can ( anyway ) easily correct those faults.

Although I doubt that those deviations above 6 kHz would be significantly audible I also wouldn't tolerate such a loudspeaker because such usually represent some problems or large tolerances on the drivers. In this case more consistent quality drivers are the way to go, just correcting by a DSP crossover or DRC filter is unfortunately not the same.

One thought - I assume that all real hifi enthusiast always sits in the sweet spot in front of the two speakers, listening in stereo. I sure do. If not , I guess anything goes

Sweet spot is usually managed easily, but perfect room, placement and furniture symmetry unfortunately not as almost LP measurements show.

 

[DanielT]

Good thread Tangband. What you are addressing, matching speakers, that is. I have nothing to contribute regarding the subject but I read the comments with interest.

 

[Tangband]

According to the thead poll, there is only a slight advantage with 11 votes for audible differences between missmatch of L/R loudspeakers . 10 votes for not a big problem and not audible .

Iˋm slightly shocked by this result..
10 people obviously dont care for good stereo image, or lack the knowledge of good installation and maybe havent heard about the precedence effect, making the direct sound from the speaker very crucial even in listening position.

In my opinion and in my experience, because I have tested it for real , missmatches between loudspeakers bigger than 1 dB can be heard as a slightly diffused stereoimage.

For some of us this might be ok and perfectly fine, but not paying attention for a possible problem in loudspeaker matching is not gonna bring Hifi-science forward.

 

[Blumlein 88]

I had not noticed it was a poll. I just voted for #1. I think there is some ambiguity in how people are voting. One view is whether matching is good for stereo imaging. The other is whether or not anyone thinks matching between speakers is commonly a problem. Not the same thing, and I think most of those #3 votes are saying it usually isn't a problem, not that mismatching if it occurs is to be ignored.

 

[sergeauckland]

I had not noticed it was a poll. I just voted for #1. I think there is some ambiguity in how people are voting. One view is whether matching is good for stereo imaging. The other is whether or not anyone things matching between speakers is commonly a problem. Not the same thing, and I think most of those #3 votes are saying it usually isn't a problem, not that mismatching if it occurs is to be ignored.

Of course there are quite a few people who don't 'get' stereo, and who are happy to walk around a room whilst listening, listen off-centre, and/or position their 'speakers where most convenient, even if not optimal for stereo. There are also those for whom timbre and coloration is more important than stereo accuracy, so pair-matching doesn't matter a lot. Similarly, those who use loudspeakers in an AV environment, with pictures, won't be bothered much by accuracy of stereo, as the pictures create the image more than the sound.

I'm one of those who positions their 'speakers first, then the rest of the furniture, who sits in the optimal position and doesn't move a lot whilst listening. I also don't have a screen in the listening room, so for me, accuracy of stereo image is what sets the stage. Those like me, (poor souls) will be most likely affected by close matching.

S.

 

[MAB]

According to the thead poll, there is only a slight advantage with 11 votes for audible differences between missmatch of L/R loudspeakers . 10 votes for not a big problem and not audible .

Iˋm slightly shocked by this result..
10 people obviously dont care for good stereo image, or lack the knowledge of good installation and maybe havent heard about the precedence effect, making the direct sound from the speaker very crucial even in listening position.

The poll is:

"loudspeakers L/R matching are sometimes terrible !​

Is this an audible problem when listening to two loudspeakers ?"​

If loudspeakers are really terribly mismatched, then yes there is an audible problem.
But, I would have to say that loudspeaker L/R matching is actually not that bad based on the data in this thread. I think @thewas showed this comprehensively. So if I voted, it would be "No, L/R matching is not terrible, so everything else is irrelevant!".

For instance, Hifinews' data on the Elac AF-61 shows really good L/R matching until the trough of the dip at 6kHz. After that there are L/R mismatches with response artifacts at even multiples of 6kHz. I marked up the Hifinews graph to show those multiples.

I actually hesitate to speculate on what is causing this. I could imagine a mechanical resonance in one of the many components and assemblies making up the concentric pair, I would likely guess wrong... But for sure this measurement shows a matched L/R pair until 6kHz. I gotta think the 8 dB dip at 6kHz is the real problem here...
And, if it is a resonance, likely the Hifinews measurement is only capturing one aspect of it.
Stereophile tested the same speaker:

Stereophile data shows just a 3dB dip at 5.5kHz. They also have the cum spectral-decay plots and additional data that makes me wonder; what is going in the 6kHz region? They may sound great or not, I just can't tell. L/R mismatch would likely be far down the list of things that impact the sound here given the smallness of the mismatch. I bet with the concentric drivers combined with the really smooth overall response with good L/R matching to 6kHz, they actually image great. The suck-out at 6kHz and additional oddities at higher octaves may or may not be audible. I am too confused by the different measurements to make any of those conclusions.

So, the results are not exactly anti-hifi enthusiast. Perhaps the are is just a bit of misunderstanding on what is being asked in your two-part question! And the fact is that most of the speakers here are not terrible for L/R. Doesn't mean L/R is irrelevant, and it doesn't mean there are a bunch of anti-hi-fi enthusiasts here. To me your question is too open ended, and this conclusion is a bit of a leap!