Magnepan LRS Speaker Review

[MZKM]

And therein lies the problem. All Magnepan speakers (including the LRS) were never intended to be measured (or listened to) "as if there was no room." The entire design is intended to utilize the room ambience to act in conjunction with the speaker output to provide the intended listening experience. This isn't a point source radiator that was designed in an anechoic chamber. Instead, the front and rear radiation are harmonious parts of the intended listening experience just as much as the room echo contribution is also. You've measured in a way totally opposed to the designers' intent and then made (grossly erroneous) conclusions based on your measurements.

Neither the cost of your test equipment nor the fact of robotic analysis can compensate for your deliberate disregard of the design intent.

Every room is different, and placement also effects it. So, unless you get 100 people who own these speakers to do the exact measurement and average them, getting an anechoic measurement is the next best thing.

 

[Vuki]

The listening test confirmed what the measurements showed, though.

Somehow I find it more relevant when measurements confirm listening impressions than when listening test confirms measurements.

 

[richard12511]

The difference between religion and a scientific theory is that any solid evidence against the theory produces doubt in it while in a religion, the doubt always falls on the evidence.

Perfect!

 

[Vuki]

What about your own listening tests? Do you think those might also be biased?

Everyone's listening test is biased but not everyone understand this.

 

[Blumlein 88]

Quite. I'd note, though, that part of the design of Maggies and other planar dipoles is that they use *acoustical* dipole equalization. That can't be accomplished with a Linkwitz design, since he has only one driver to work with in each frequency band. But a planar panel can be partitioned into underdamped resonant segments, and these can be tuned to compensate for the 6 dB/octave dipole rolloff. Thus Linkwitz uses active equalization (some of it programmed by Dave Reite in this thread), while Maggies don't require it, as the nearfield measurements show.

You must be looking at different measurements than I am in this review.

 

[Vasr]

(Wow, an even bigger vertical scale (dB)!) Nearly 160dB covered by Y-axis! (should be more like 50dB covered on the vertical scale)

The point does not change. It only accentuates a single position room mode.

This is what an early roll-off in the lows would look like as people interpreted the measurements here

 

[josh358]

I can imagine this is hard to deal with, and no doubt is responsible for much of this reviews criticism. I really do sympathize with owners.

That said, it is also possible that the measurements simply don't capture the magic of these speakers. Everyone who's heard them praises their imaging and soundstage.

But that's precisely it -- they don't capture the magic --- and as I said to Amir when the possibility of a review was first mentioned, that's what I was curious about, why they sound better than they measure. Maggies sound more realistic than speakers that would seem to measure better, and this implies that the measurements are incomplete, erroneous, or misinterpreted. And I'm personally interested in that.

One obvious way in which the measurements fail is the one you mentioned -- it's hard to get from the measurements to an understanding of the speakers' spatial qualities. I have a pretty good idea, but 99 percent of the people who see these measurements will not.

But the main thing that's ticking people off is that people are judging speakers that they've never heard. Whoa, Nellie.

I have some issues with Amir's listening methodology -- one speaker? You're throwing away one of the speaker's main advantages if you do that, its imaging. And no wonder he heard image wandering -- the speakers are mirror imaged for a reason. The image won't wander if the other pair is in place, it will grow a bit diffuse as it moves to the extremes. You're screwing up the bass as well, because these speakers were designed to be used as a pair -- the acoustical equalization is shared between the speakers, with two resonant sections on one and three on the other, and requires that both be used for correct midbass response.

But at least Amir listened, so has a useful opinion, e.g., re the rise in the lower midrange (not everything requires two speakers to hear).

 

[Blumlein 88]

But that's precisely it -- they don't capture the magic --- and as I said to Amir when the possibility of a review was first mentioned, that's what I was curious about, why they sound better than they measure. Maggies sound more realistic than speakers that would seem to measure better, and this implies that the measurements are incomplete, erroneous, or misinterpreted. And I'm personally interested in that.

One obvious way in which the measurements fail is the one you mentioned -- it's hard to get from the measurements to an understanding of the speakers' spatial qualities. I have a pretty good idea, but 99 percent of the people who see these measurements will not.

But the main thing that's ticking people off is that people are judging speakers that they've never heard. Whoa, Nellie.

I have some issues with Amir's listening methodology -- one speaker? You're throwing away one of the speaker's main advantages if you do that, its imaging. And no wonder he heard image wandering -- the speakers are mirror imaged for a reason. The image won't wander if the other pair is in place, it will grow a bit diffuse as it moves to the extremes. You're screwing up the bass as well, because these speakers were designed to be used as a pair -- the acoustical equalization is shared between the speakers, with two resonant sections on one and three on the other, and requires that both be used for correct midbass response.

But at least Amir listened, so has a useful opinion, e.g., re the rise in the lower midrange (not everything requires two speakers to hear).

What if your recording had mono bass? How does non matching panel help that? And are you sure that's how the LRS is functioning?

 

[Vasr]

The lower midrange/upper bass peaking has been a part of Magnepan's sound off and on over the years, along with some other "interesting" frequency response choices, but not really much worse or better than a lot of other speakers, particularly smaller ones.

I think this is the part that is missing in the interpretations of the measurements (not a problem with the measurements). The REW measurement I posted above of the SMGa also shows a low mid bump but not as exaggerated as the LRS.

Maggies (smaller ones) were never known to have a deep bass extension (the point at which the output falls off rapidly by a standard amount). That is not controversial. Unfortunately, the interpretation here appears to imply and has been stretched to the point that people believe that the LRS is bass-deficient starting at about 300hz or so. This is a really bad interpretation encouraged by the review. The MMG-W curve I posted above is closer to what a bass-deficient speaker with early roll-off would look like.

If you sat on a pimple, the face would look cratered. If you sat on the face surface, the pimple would look big. So, is the face cratered or the pimple too big? The review and ensuing discussion leads one to believe the former. This is a bad take on the LRS in my opinion.

One might think the above is a distinction without a difference but that is incorrect.

From a practical perspective, if there is an early roll-off, it is difficult if not impossible to correct it via some EQ processing. But you could just add a sub. If there are peaks or exaggerated areas and they are heard as problematic, you can hold it against the speaker or consider it the "signature" that you like/dislike, but they can also be reduced with EQ. Not very different from taming some brightness in treble. The LRS even comes with resistors to affect treble balance depending on the room and tastes.

I think it is perfectly fine to disagree on trade-offs speakers make and hold them to whatever standards one likes but at least let us get the interpretation of measurements right because these interpretations are all subjective.

 

[josh358]

You must be looking at different measurements than I am in this review.

The difference is that I know how to interpret them.

Look at Stereophile's measurement of the LRS:

See how the bass rises as frequency decreases? That's the acoustical dipole equalization of which I spoke, and it compensates for the 6 dB/octave dipole cancellation. The reason it shows up in John Atkinson's measurements is that the speakers were measured in the near field. In the far field, the low frequency waves diffract around the baffle and you get flat response, which seems to be the case with the LRS (the measurements I'm seeing here suggest that there's a peak in the midrange rather than a deficiency in the midbass -- so a different flaw).

 

[josh358]

What if your recording had mono bass? How does non matching panel help that? And are you sure that's how the LRS is functioning?

I discussed it once with Mark Winey, the president of Magnepan, and he confirmed that that's why the MMG's (the predecessor of the LRS) are asymmetrical.

The problem is that the LRS is a very small speaker and there isn't much room for the resonant segments used for the acoustic equalization, so they had to make a compromise and split the equalization between the two panels. IIRC, it goes from lowest on one to second lowest on the other, then third lowest on the first, then fourth and fifth lowest on the second. You can actually play a musical scale if you tap on the sections!

I don't think mono bass is much of an issue, because bass isn't very directional in a typical listening room. Still, it could cause some image instability.

Another possible issue is that as the image moves to the *side*, you're only going to hearing one speaker, so the bass will be less accurate. I suspect this wasn't a problem in LP days when bass was mixed to mono to avoid excessive vertical modulation, but now that that's no longer necessary, it could be an issue if a bass instrument was panned to the left or the right.

 

[TheGhostOfEugeneDebs]

Figure I may as well show my own measurement in-room of the 1.7i. Measurements are from Dirac and are the aggregate of 6 different locations around the seating area. The measurement only differs from my perceived experience in the sub 100hz range, where I don't perceive that dip. It actually sounds more like the way Atkinson's chart looks, with a peak and plateau below 100 to the drop off point at 45hz. It is more of a room problem than a speaker one though, they are very EQ-able in those frequencies and certainly not bass-light. I was able to position them a few different ways to get more bass (closer to the wall - though it got muddy) eliminate the midrange dip above 1k (closer to me, but then I was getting a weird echoic effect that I didn't like and couldn't quite find the source of in the room to dampen), boost the treble a LOT (switching the speakers so the tweeters were on the inside) and so on, but I lost those charts when Dirac updated and won't let me load them anymore lol. This represents the most compromise with regards to room placement with in room sound experience and the position I am ultimately EQing from currently.

Again, I'm more than happy with these speakers. Especially after I EQed them to my liking, they absolutely FILL the room in a way I never experienced with other speakers. These speakers are more than twice the size of the LRS though, so that may be helping that effect. I've never heard the LRS.

 

[bobbooo]

But that's precisely it -- they don't capture the magic --- and as I said to Amir when the possibility of a review was first mentioned, that's what I was curious about, why they sound better than they measure. Maggies sound more realistic than speakers that would seem to measure better, and this implies that the measurements are incomplete, erroneous, or misinterpreted. And I'm personally interested in that.

One obvious way in which the measurements fail is the one you mentioned -- it's hard to get from the measurements to an understanding of the speakers' spatial qualities. I have a pretty good idea, but 99 percent of the people who see these measurements will not.

But the main thing that's ticking people off is that people are judging speakers that they've never heard. Whoa, Nellie.

I have some issues with Amir's listening methodology -- one speaker? You're throwing away one of the speaker's main advantages if you do that, its imaging. And no wonder he heard image wandering -- the speakers are mirror imaged for a reason. The image won't wander if the other pair is in place, it will grow a bit diffuse as it moves to the extremes. You're screwing up the bass as well, because these speakers were designed to be used as a pair -- the acoustical equalization is shared between the speakers, with two resonant sections on one and three on the other, and requires that both be used for correct midbass response.

But at least Amir listened, so has a useful opinion, e.g., re the rise in the lower midrange (not everything requires two speakers to hear).

Is there any evidence of this supposed subjective 'magic' in the form of double-blind listening tests, showing higher listener preference than speakers that measure objectively better?

 

[josh358]

Bass response
A few simple considerations about the low frequency reproduction of the LRS. The low-frequency reproduction depends only on the volume of displacement (Vd).

So it's all about estimating the Vd for the LRS speaker. Anyone who has the speaker at home can provide the exact data, I can only give you a rough estimate.

The size of the"woofer panel" corresponds approximately to the marked area in the photo:
View attachment 83899
This corresponds to about 95cm x 19cm, so about 1800cm².
Of course, not all of the panel surface is membrane surface. My estimate is that about 40% of it is membrane area.
So 720cm² membrane area is available.

Next we have to estimate the "Maximum Linear Excursion (Xmax)" of the membrane. Someone who is better acquainted with the functioning of the Magnepan-Ribbon may correct me here, my estimate would be 0.25 - 1mm excursion.

With assumed 0.5mm Xmax the Vd corresponds to about 720cm² x 0.05cm = 36cm³.
This corresponds somewhat to the Vd of a typical 6'' chassis (10cm³ => 4'', 20cm³ => 5'', 30cm³ => 6'', 80cm³ => 7'', 150cm³ => 8'',...).
Then there are the "losses" due to the open baffle dipol design. This means less bass performance than a "closed box" 6'' chassis.

As already mentioned, owners of the LRS can certainly determine this value more precisely, but it should be clear that for physical reasons alone, not much is possible with the LRS in the low bass.



Multiton Distortion (MD) & decay behavior (CSD)
To be honest, I'm a bit surprised that nobody has determined the multitone distortion of the LRS.
Especially because of the low Xmax of the "Ribbon-Woofer" it would be very interesting to make measurements. Are the intermodulation distortions (IMD) particularly low (minor excursion of the membrane) or high (low Xmax of the membrane)?

The same applies to the decay behaviour of the loudspeaker. The manufacturer even explicitly points out resonance problems, which should be suppressed by setting "buttons" on the panel.
Therefore CSD measurements of the "woofer-panel" and the "tweeter-panel" would be very interesting.

You can easily measure close to the panel with the microphone (preferably at different heights of the panel), which should give useful results in a simple way.

The small baffle, the excursion limits of the driver, and the finite energy for acoustical equalization all set a limit on bass extension.

The behavior of the bass panels and the role of the hold downs ("buttons") is a bit complex, meaning that calculating the volume of displacement isn't as straightforward as it would be for, say, a dynamic driver.

The panels are divided into resonant sections and the motors underdamped to provide 6 dB/octave acoustical equalization to compensate for the 6 dB/octave dipole cancellation. (It is one of several techniques that are/can be used for this -- altering the tension of the film across the speaker is another, as were the trapezoidal drivers in the old Apogees.) This means that the motion of the diaphragms isn't simple. In a given frequency range, most of the contribution will come from a segment, rather than the whole diaphragm. And much of that motion will then be "wasted" by the 6 dB/octave dipole cancellation to which you refer.

There's another issue as well, which is that the diaphragm will tend to "dish," meaning that it doesn't make full use of the volume. The drivers are designed to accommodate that, in that the pole piece/magnet assembly is curved and Xmax is higher towards the center. This maximizes Bl product at maximum excursion. Some of the drivers also have double traces at the edges to reduce dishing (AFAIK) by increasing the Bl product there (AFAIK, the true ribbon ones don't). And acoustical damping can be added towards the center of the driver -- not sure if that was done in the LRS.

Anyway, the buttons aren't just there to spread the resonance (though they contribute to that), but to shape it for acoustical EQ.

Re IM, the tweeters on these little models share the diaphragm with the woofer, to increase the effective size of the woofer. This makes them more susceptible to IM. In practice, I heard intermodulation from the bass on the old MMG when the SPL's got into maybe the mid 90's.

In general, planars have high THD in the bass, and very low THD higher up. The low THD and IM may have something to do with the vaunted "transparency" of planar magnetic loudspeakers and ESL's -- I'm not sure, as there are other possible explanations (waterfall, directivity, etc.).

 

[Ilkless]

The difference is that I know how to interpret them.

Look at Stereophile's measurement of the LRS:

See how the bass rises as frequency decreases? That's the acoustical dipole equalization of which I spoke, and it compensates for the 6 dB/octave dipole cancellation. The reason it shows up in John Atkinson's measurements is that the speakers were measured in the near field. In the far field, the low frequency waves diffract around the baffle and you get flat response, which seems to be the case with the LRS (the measurements I'm seeing here suggest that there's a peak in the midrange rather than a deficiency in the midbass -- so a different flaw).

So... somehow the merits of the quintessential farfield speaker only emerge on nearfield measurements? A dipole-equalised speaker equalises for the response in farfield after the front and back waves have cancelled. Why would anyone equalise based on the front radiation before it is cancelled when it is the net radiation including the null that is interacting with the listener?

This is pseudoscientific apologism akin to the ATC types that refuse to see how their approach has been exposed by evidence-based methods.

 

[ctrl]

(I've noticed by the way that some people have been making the opposite assumption -- that the performance limitations of the LRS apply to the larger models. But of course they don't, which is why there are larger models! Bass extension for example ranges from 60 Hz in the LRS to 20 Hz in the high end 20.7 and 30.7. They're no different than a dynamic speaker line in this regard.)

With all bias, one should not drift into a fantasy world where everything is possible
The laws of physics still apply. It is simply about volume of displacement (Vd).

For loudspeakers in this price range ($30,000), two 10'' subwoofers or an 18'' PA woofer for horn systems are not unusual.
Which corresponds to about 800cm³ Vd. If my consideration in post#305 is not completely wrong, this would correspond to an area of 20 LRS panels.
So it should be clear that even a Magnepan 30.7 loudspeaker is far away from that in terms of low frequency dynamics and maximum sound pressure.

Look at Stereophile's measurement of the LRS:

The low frequency measurements of stereophiles should be considered with caution, as they do not contain baffle compensation.

 

[josh358]

Is there any evidence of this supposed subjective 'magic' in the form of double-blind listening tests, showing higher listener preference than speakers that measure objectively better?

Yes, but closely held by Magnepan. They subject speakers to blind A/B testing with two listening panels, one expert and the other not. For obvious competitive reasons, I doubt that they'll make the results of those tests public, and I've never seen them myself. But they won't release a new model unless it is judged improved in those A/B tests, and I think that has a lot to do with why they get such consistently good reviews.

I don't know of blind A/B comparisons that have been done publicly, but that's true of most speakers.

Part of the problem here is that all but the most naive listener would be able to identify which were the panels in the A/B test, so it would no longer be objective in that sense -- we aren't talking subtle differences of the "can you hear this" kind, line source dipoles and point source omni/cardioids have readily identifiable sonic signatures. So you'd still get expectation bias unless you used a naive listening panel -- the reason, I think, that Magnepan uses two panels in their in-house testing.

 

[ctrl]

The panels are divided into resonant sections and the motors underdamped to provide 6 dB/octave acoustical equalization to compensate for the 6 dB/octave dipole cancellation. (It is one of several techniques that are/can be used for this -- altering the tension of the film across the speaker is another, as were the trapezoidal drivers in the old Apogees.) This means that the motion of the diaphragms isn't simple. In a given frequency range, most of the contribution will come from a segment, rather than the whole diaphragm. And much of that motion will then be "wasted" by the 6 dB/octave dipole cancellation to which you refer.

But this would mean that the volume of displacement of the LRS would be much lower than I thought.

Since you know a lot about Magnepan, is there any information from the manufacturer about the Vd of the speakers or panels?

 

[bobbooo]

Yes, but closely held by Magnepan. They subject speakers to blind A/B testing with two listening panels, one expert and the other not. For obvious competitive reasons, I doubt that they'll make the results of those tests public, and I've never seen them myself. But they won't release a new model unless it is judged improved in those A/B tests, and I think that has a lot to do with why they get such consistently good reviews.

I don't know of blind A/B comparisons that have been done publicly, but that's true of most speakers.

Part of the problem here is that all but the most naive listener would be able to identify which were the panels in the A/B test, so it would no longer be objective in that sense -- we aren't talking subtle differences of the "can you hear this" kind, line source dipoles and point source omni/cardioids have readily identifiable sonic signatures. So you'd still get expectation bias unless you used a naive listening panel -- the reason, I think, that Magnepan uses two panels in their in-house testing.

So that's a 'no' then, to any scientific evidence of subjective preference that belies their objective deficiencies.

 

[josh358]

So... somehow the merits of the quintessential farfield speaker only emerge on nearfield measurements? A dipole-equalised speaker equalises for the response in farfield after the front and back waves have cancelled. Why would anyone equalise based on the front radiation before it is cancelled when it is the net radiation including the null that is interacting with the listener?

This is pseudoscientific apologism akin to the ATC types that refuse to see how their approach has been exposed by evidence-based methods.

With all due respect, you don't have the foggiest understanding of what I said, or why I said it; you are arguing with yourself. You are welcome to your interpretive fantasies, and to your insults; I don't propose to waste time on them.