This is a review and detailed measurements of the Magnepan LRS (Little Ribbon Speaker). It was kindly sent to me by a member and costs US $650.
NOTE: as you will see later, this is a special review with far more detail than I usually show in a speaker test. I thought it would be "fun" to see much more extensive treatment of these speakers. Result was three full days of measuring, processing, processing again and again, generating slides, generating those again, and again. I hope you appreciate the extra work that has gone into this review and no, I won't be doing this with future reviews.
Given the detail level, this is not for a casual reader of these reviews. As such, feel free to skip to subjective listening results and conclusions.
I was thankful for the thin and relative light weight of the LRS when I went to measure it and carry it to my listening room:
As you can sort of tell from above picture, there are two panels side-by-side. One to the left with wider size and traces and the other handling higher frequencies with much narrower width. Per feedback from membership, I selected the measurement axis/point as being more or less where the "X" mark is in red (center of the panel). Alas, I had to empirically move this [position (in software) as you will see later. I wanted the center of that axis to be on the tweeter but the auxiliary stand that I built was symmetrical and wouldn't allow me to go close to the right.
You see an indentation toward top. This is not normally visible but the reason for it is a large button screw that holds the panel down (I think).
A set of metal stands forces a some amount of lean back which I find surprising. By dropping a couple of rings, you can tilt the speaker less and is something that is recommended by the company if listening past 10 feet or so. I was worried about measuring the speaker so leaning back in its default position so I tiled it up mostly to the angle that would be if you used the secondary, more upright position.
I was fortunate enough to get a preview measurement that was performed by Klippel distributor in US, Warkwyn labs which was published in AudioExpress magazine. My measurements match theirs although the results as presented are different due to some improvements I made as you see later.
Warkwyn measurements showed that the Klippel system was struggling to characterize the sound field despite using over 2000 measurement points. I was going to use more measurement points only to realize it would take 5 hours just to do the 2000 point measurement! It was really strange to watch the system make a vertical set of measurements and move a millimeter and do that all over again! The tall speaker meant there was a lot of time lost moving up and down, lengthening measurements.
Not only was the measurement time long, but so was processing the 1.5 Gigabyte file to compute the sound field and various measurements. Computational time was in the order of half hour.
Klippel NFS Measurement Accuracy
As most of you hopefully know, the Klippel NFS makes a series of near-field measurements (in order to benefit from better signal to noise ratio) and then using those points, solves the partial differential equations that describe the wave propagation. Once there, we are able to predict the sound field in any point in space in the far field (where we more or less listen). The technique while wonderful, has limitations in that if the sound field becomes too complex, it requires many more measurement points and "orders of expansion." Despite using high measurement points used per above (more than 2X of any speaker I have measured), the sound field was too complex in high frequencies to get accurate readings. Klippel is able to determine the error by making additional "real" measurements that it then compares to what it has computed. The difference is expressed in dB of error. Usually the results are below -20 dB indicating 1% error in most or all of the audible band. That was not the case here:
In theory, this should NOT have happened. Lay intuition about this speaker is that they are two vertical drivers each sending out a perfect plane wave of their own. Reality is different in that there are clearly other sources of sound interfering with each other, creating a highly complex wave front. Fortunately, the error is not high enough to distort what we are interested in. Here is a Klippel graph showing the actual versus computed response at a certain point:
Focusing on the right, the actual measured response was in red but the blue is what is computed. As you see, as the frequencies go higher, error increases but fortunately it still more or less follows the response of the speaker which is dropping like a rock above 10 kHz or so.
Speaker Radiation Pattern
Normally I lead with our spinorama measurements but here, I thought we work backward and first look at the directivity of the speaker. Here it is in the horizontal axis. That is, you are facing the speaker and the graph shows what happens 360 degrees around the speaker:
Being a dipole speaker, the LRS radiates what you see in the front, in the back. The back response however is split into two halves (top and bottom of the graph) so not as intuitive. But hopefully you see it now with the annotations in place. We have a nice constant beam width up to low treble and then the pretty picture corrupts:
The notch around 1 kHz in directivity is due to offset nature of the tweeter panel. I have the "right" speaker so the treble has baffle amplification on the left but not right. At least this is what I assumed when I created the slide but I am not sure the tweeter is active there. Open to ideas from readers.
The most interesting part and what cost me a lot of time and effort was the vertical dispersion:
Yes, as you go above 1 kHz or though, the speaker starts to beam hugely, creating a super narrow angle of +- 10 degrees where you get the full response. Go up or down below this and you have massive loss of high frequencies. Indeed that is what I had when I first measured the speaker as did the Warkwyn plots. So I started to adjust the reference axis higher and higher until I got it in that sweet spot.
Spinorama Speaker Measurements
Here is our standard graph now:
To give you an idea of the response was before I changed it, here it is:
You basically get no highs or low!
Both graphs show odd modulations of on-axis response above 3 kHz as indicated by the jagged peaks. And the fact that this speaker doesn't generate much sound until you get to 300 Hz or so.
One of the "benefits" of a line source of this type is that you get little radiation above and below the speaker. We can see some of that effect in the early window and the levels of the floor and ceiling reflections:
And here is our predicted in-room response:
There was no good way to draw the trend line given the large drop in bass response. But I tried anyway.
Sound Field Visualizations
Let's get fancy and look at how the sound propagates from the LRS first in horizontal plane. I can only show this at one frequency at a time so let's start with 500 Hz or so that is squarely in the domain of the main low frequency panel:
The color shows sound pressure level (red the highest, blue the lowest). We see a wave front radiating from the driver.
Now let's step up to 2.2 kHz:
If my interpretation is correct, we now have both drivers radiating and creating interference patterns.
Going to 10 kHz isolates just the tweeter:
The tweeter panel is the right but we also get images on the left side of the speaker.
I also analyzed a vertical slice but this time, I will just show a static picture (above animations were a lot of work to create):
Imagine you are standing on the right looking into the speaker. We see that the radiation pattern on the back and front follow the tilt axis of the speaker as it was measured. Also indicated clearly is that we have a narrow range of radiation before we hit pockets that are nulls or have less power (no red color in them). These interference patterns must be due to multiple sources playing at once and creating the complex sound field I talked about at the start of the review.
Speaker Distortion Measurements
At the original point I was measuring the speaker, the highs were very low and as such, I could not get the LRS up to 86 dB let alone 96 dB. Here are what I got anyway:
Distortion seems to be very much in control at higher frequencies.
Notice how rough the in-room response is where I measured the speaker.
Impedance Measurements
Electrically panel speakers like LRS are simple resistive loads:
Impedance dips down to 3 ohm so you better have an amplifier with good current capability. The phase though is nearly zero. The vertical scale is only 20 degrees and hence the visually large variation.
Transition to Far Field
Klippel computers the response of the speaker and can show at what point we are approaching far field of the speaker. For LRS and at frequencies of 400 Hz and higher, that is 3 meters or 10 feet. So better not sit too close to them:
Subjective Speaker Listening Tests
I first positioned the panel right at me and started to play. What I heard sounded like it was coming from a deep well! I then dropped the little rings on the stand and repositioned the speaker as you see in the picture (less toed in). That made a big difference and for a few clips I enjoyed decent sound. Then I played something with bass and it was as if the speaker was drowned under water again. It wasn't just absence of deep bass but rather, quietness on top of that.
Even when the speaker sounded "good" you would hear these spatial and level shifts that was really strange. As the singers voice changed tonality, it would sometimes shift left and right. And change in level no doubt due to uneven frequency response. There was also some strange extended tail to some high frequency notes that would seem to go on forever.
Just when I thought I had the speaker dialed in, I leaned back some and the tonality got destroyed. You had to sit in the proverbial vice around your head to get the "right" sound out of LRS.
I applied a quick and dirty inverse fix to the response to get some semblance of neutrality:
The one PEQ shown, combined with an overall lift of the entire response made a huge difference. Speaker was no longer dull, lacking both bass and treble. Alas, after listening some, the highs got to me so I put in the right filter to fix that. And while the LRS could handle the boost in low frequencies well, bringing for the first time some tactile feedback, it did start to bottom out so I had to put that sharp filter for extreme lows.
Once there, I was kind of happy until I played the soundtrack you see at the bottom. Man did it sound horrid. Bland and some of the worse bass I have heard.
When it did sound good -- which was on typical show audiophile tracks -- the experience was good. Alas, every track would sound similar with the same height and spatial effects.
Conclusions
The Magnepan LRS is a hugely flawed speaker with moments of delight. If I could control what you listen to, e.g. in an audio show or dealer room, I could convince you it is much better speaker than it is. The best way I can explain this is that the designers solved 30% of the physics of building a speaker, and threw you in there to solve the rest! You take on the job of spending what must be a lifetime messing with location, tilt, EQ, etc. to get sound that is good for more than a few select tracks.
I am confident a better job can be done than what we see in LRS. Maybe making the panels smaller causes the beaming and interference patterns worse. I don't know. What I do know that this is not a product finished and fit for use by a consumer.
I wonder how much simulation and in field analysis was performed as I have shown here. Doesn't seem like much was done to find and remove issues with this speaker.
Needless to say, I can't recommend the Magnepan LRS.
------------
As always, questions, comments, recommendations, etc. are welcome.
Have to go and see if I can fix our dishwasher now. Too cheap to pay someone $500 or more to fix this German invention. If you want me to consider hiring someone to fix it, please donate generously using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
NOTE: as you will see later, this is a special review with far more detail than I usually show in a speaker test. I thought it would be "fun" to see much more extensive treatment of these speakers. Result was three full days of measuring, processing, processing again and again, generating slides, generating those again, and again. I hope you appreciate the extra work that has gone into this review and no, I won't be doing this with future reviews.
Given the detail level, this is not for a casual reader of these reviews. As such, feel free to skip to subjective listening results and conclusions.
I was thankful for the thin and relative light weight of the LRS when I went to measure it and carry it to my listening room:
As you can sort of tell from above picture, there are two panels side-by-side. One to the left with wider size and traces and the other handling higher frequencies with much narrower width. Per feedback from membership, I selected the measurement axis/point as being more or less where the "X" mark is in red (center of the panel). Alas, I had to empirically move this [position (in software) as you will see later. I wanted the center of that axis to be on the tweeter but the auxiliary stand that I built was symmetrical and wouldn't allow me to go close to the right.
You see an indentation toward top. This is not normally visible but the reason for it is a large button screw that holds the panel down (I think).
A set of metal stands forces a some amount of lean back which I find surprising. By dropping a couple of rings, you can tilt the speaker less and is something that is recommended by the company if listening past 10 feet or so. I was worried about measuring the speaker so leaning back in its default position so I tiled it up mostly to the angle that would be if you used the secondary, more upright position.
I was fortunate enough to get a preview measurement that was performed by Klippel distributor in US, Warkwyn labs which was published in AudioExpress magazine. My measurements match theirs although the results as presented are different due to some improvements I made as you see later.
Warkwyn measurements showed that the Klippel system was struggling to characterize the sound field despite using over 2000 measurement points. I was going to use more measurement points only to realize it would take 5 hours just to do the 2000 point measurement! It was really strange to watch the system make a vertical set of measurements and move a millimeter and do that all over again! The tall speaker meant there was a lot of time lost moving up and down, lengthening measurements.
Not only was the measurement time long, but so was processing the 1.5 Gigabyte file to compute the sound field and various measurements. Computational time was in the order of half hour.
Klippel NFS Measurement Accuracy
As most of you hopefully know, the Klippel NFS makes a series of near-field measurements (in order to benefit from better signal to noise ratio) and then using those points, solves the partial differential equations that describe the wave propagation. Once there, we are able to predict the sound field in any point in space in the far field (where we more or less listen). The technique while wonderful, has limitations in that if the sound field becomes too complex, it requires many more measurement points and "orders of expansion." Despite using high measurement points used per above (more than 2X of any speaker I have measured), the sound field was too complex in high frequencies to get accurate readings. Klippel is able to determine the error by making additional "real" measurements that it then compares to what it has computed. The difference is expressed in dB of error. Usually the results are below -20 dB indicating 1% error in most or all of the audible band. That was not the case here:
In theory, this should NOT have happened. Lay intuition about this speaker is that they are two vertical drivers each sending out a perfect plane wave of their own. Reality is different in that there are clearly other sources of sound interfering with each other, creating a highly complex wave front. Fortunately, the error is not high enough to distort what we are interested in. Here is a Klippel graph showing the actual versus computed response at a certain point:
Focusing on the right, the actual measured response was in red but the blue is what is computed. As you see, as the frequencies go higher, error increases but fortunately it still more or less follows the response of the speaker which is dropping like a rock above 10 kHz or so.
Speaker Radiation Pattern
Normally I lead with our spinorama measurements but here, I thought we work backward and first look at the directivity of the speaker. Here it is in the horizontal axis. That is, you are facing the speaker and the graph shows what happens 360 degrees around the speaker:
Being a dipole speaker, the LRS radiates what you see in the front, in the back. The back response however is split into two halves (top and bottom of the graph) so not as intuitive. But hopefully you see it now with the annotations in place. We have a nice constant beam width up to low treble and then the pretty picture corrupts:
The notch around 1 kHz in directivity is due to offset nature of the tweeter panel. I have the "right" speaker so the treble has baffle amplification on the left but not right. At least this is what I assumed when I created the slide but I am not sure the tweeter is active there. Open to ideas from readers.
The most interesting part and what cost me a lot of time and effort was the vertical dispersion:
Yes, as you go above 1 kHz or though, the speaker starts to beam hugely, creating a super narrow angle of +- 10 degrees where you get the full response. Go up or down below this and you have massive loss of high frequencies. Indeed that is what I had when I first measured the speaker as did the Warkwyn plots. So I started to adjust the reference axis higher and higher until I got it in that sweet spot.
Spinorama Speaker Measurements
Here is our standard graph now:
To give you an idea of the response was before I changed it, here it is:
You basically get no highs or low!
Both graphs show odd modulations of on-axis response above 3 kHz as indicated by the jagged peaks. And the fact that this speaker doesn't generate much sound until you get to 300 Hz or so.
One of the "benefits" of a line source of this type is that you get little radiation above and below the speaker. We can see some of that effect in the early window and the levels of the floor and ceiling reflections:
And here is our predicted in-room response:
There was no good way to draw the trend line given the large drop in bass response. But I tried anyway.
Sound Field Visualizations
Let's get fancy and look at how the sound propagates from the LRS first in horizontal plane. I can only show this at one frequency at a time so let's start with 500 Hz or so that is squarely in the domain of the main low frequency panel:
The color shows sound pressure level (red the highest, blue the lowest). We see a wave front radiating from the driver.
Now let's step up to 2.2 kHz:
If my interpretation is correct, we now have both drivers radiating and creating interference patterns.
Going to 10 kHz isolates just the tweeter:
The tweeter panel is the right but we also get images on the left side of the speaker.
I also analyzed a vertical slice but this time, I will just show a static picture (above animations were a lot of work to create):
Imagine you are standing on the right looking into the speaker. We see that the radiation pattern on the back and front follow the tilt axis of the speaker as it was measured. Also indicated clearly is that we have a narrow range of radiation before we hit pockets that are nulls or have less power (no red color in them). These interference patterns must be due to multiple sources playing at once and creating the complex sound field I talked about at the start of the review.
Speaker Distortion Measurements
At the original point I was measuring the speaker, the highs were very low and as such, I could not get the LRS up to 86 dB let alone 96 dB. Here are what I got anyway:
Distortion seems to be very much in control at higher frequencies.
Notice how rough the in-room response is where I measured the speaker.
Impedance Measurements
Electrically panel speakers like LRS are simple resistive loads:
Impedance dips down to 3 ohm so you better have an amplifier with good current capability. The phase though is nearly zero. The vertical scale is only 20 degrees and hence the visually large variation.
Transition to Far Field
Klippel computers the response of the speaker and can show at what point we are approaching far field of the speaker. For LRS and at frequencies of 400 Hz and higher, that is 3 meters or 10 feet. So better not sit too close to them:
Subjective Speaker Listening Tests
I first positioned the panel right at me and started to play. What I heard sounded like it was coming from a deep well! I then dropped the little rings on the stand and repositioned the speaker as you see in the picture (less toed in). That made a big difference and for a few clips I enjoyed decent sound. Then I played something with bass and it was as if the speaker was drowned under water again. It wasn't just absence of deep bass but rather, quietness on top of that.
Even when the speaker sounded "good" you would hear these spatial and level shifts that was really strange. As the singers voice changed tonality, it would sometimes shift left and right. And change in level no doubt due to uneven frequency response. There was also some strange extended tail to some high frequency notes that would seem to go on forever.
Just when I thought I had the speaker dialed in, I leaned back some and the tonality got destroyed. You had to sit in the proverbial vice around your head to get the "right" sound out of LRS.
I applied a quick and dirty inverse fix to the response to get some semblance of neutrality:
The one PEQ shown, combined with an overall lift of the entire response made a huge difference. Speaker was no longer dull, lacking both bass and treble. Alas, after listening some, the highs got to me so I put in the right filter to fix that. And while the LRS could handle the boost in low frequencies well, bringing for the first time some tactile feedback, it did start to bottom out so I had to put that sharp filter for extreme lows.
Once there, I was kind of happy until I played the soundtrack you see at the bottom. Man did it sound horrid. Bland and some of the worse bass I have heard.
When it did sound good -- which was on typical show audiophile tracks -- the experience was good. Alas, every track would sound similar with the same height and spatial effects.
Conclusions
The Magnepan LRS is a hugely flawed speaker with moments of delight. If I could control what you listen to, e.g. in an audio show or dealer room, I could convince you it is much better speaker than it is. The best way I can explain this is that the designers solved 30% of the physics of building a speaker, and threw you in there to solve the rest! You take on the job of spending what must be a lifetime messing with location, tilt, EQ, etc. to get sound that is good for more than a few select tracks.
I am confident a better job can be done than what we see in LRS. Maybe making the panels smaller causes the beaming and interference patterns worse. I don't know. What I do know that this is not a product finished and fit for use by a consumer.
I wonder how much simulation and in field analysis was performed as I have shown here. Doesn't seem like much was done to find and remove issues with this speaker.
Needless to say, I can't recommend the Magnepan LRS.
------------
As always, questions, comments, recommendations, etc. are welcome.
Have to go and see if I can fix our dishwasher now.
Too cheap to pay someone $500 or more to fix this German invention. If you want me to consider hiring someone to fix it, please donate generously using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
