Magnepan LRS Speaker Review

[KaiserSoze]

I'm going to partially disagree with you on this, speaking as someone who owns LX521's and has used mainly dipoles and planar speakers for the past 20 years. I think they'd measure much better than this low-end Magnepan did but I wouldn't be surprised if a Revel or cheap JBL measured as well or better. OTOH, I think that if someone were to listen to a pair of properly engineered dipoles properly setup in a room they'd hear things not conveyed by a positive Klippel NFS analysis. No science to back up my hunch as I don't have a Klippel NFS system setup in my garage.

Some reading to backup my hunch.

I am puzzled by the fact that you started by saying "I'm going to partially disagree with you on this...", but then did not say anything that, so far as I could tell, was in disagreement with what I had said. For you to have done that, you would had to have said that the LX521s would most likely measure bad on the Klippel instrument. If that's what you intended to say, it did not come across to me, which leaves me puzzled as to why you would start off this short post by writing, "I'm going to partially disagree with you.." Hell, if you're say that I'm wrong about something, at least show me the courtesy of saying what it is that you think I was wrong about.

 

[KaiserSoze]

in short the thing you describe can still be a point source no matter how many woofers and or bass ports it depends on what frequency they produce. everything low can be still be further away from each other , but i must say i read some more in the paper and i understand only 1/4th of it (or less). so for all i know they found a solution to do it correct. and assumed according to methods normally used... you cant do it like this. so i might be wrong. still everything i said i think is true. be it i might utterly incorrect if the klippel works the way people say it work. and i need to do some home work on the measurement system used in this case.

It isn't trivial, that much is certain!

 

[KaiserSoze]

"And by the way, this is all stuff that most all speaker designers have understood since about when they first began contemplating the design of loudspeakers"

And thats why they are in business for 50 years now

"I appreciate your tone of civility, which sets an example for me to follow. "

But i do think you lost that at attitude quite fast. i have a feeling you know me and i know you but i surely never had the honor to meet a Kaiser/Emperor so who are you ?

I apologize for having not maintained a proper tone of civility. I am entirely certain that you and I have never met.

The point is that there must be a reason why the vast majority of speakers built in the past fifty years have avoided placing tweeters side-by-side with either a midrange or a woofer. I should not have ended on the sentence that you quoted. It was unnecessary and was not appropriate. I will try harder to behave like a civilized person.

 

[MRC01]

Large dipoles are hard to measure because what you measure near field isn't what you hear, and what you hear depends on the room. However, my experience owning Magnepan 3.6/R is that actual in-room measurements can be quite good. At the listener position using REW and calibrated mics, mine measure within 3 dB of flat from 35 Hz to 20 kHz, which happens to be the spec. Distortion measures -40 dB (1%) at 60 Hz, -50 dB in the mids, -60 dB in the treble. And CSD is pretty smooth. Group delay is quite flat.

No question, these big dipoles do have their drawbacks. They're unwieldy, they take over the room because they must be away from walls and corners, they require a big powerful amp, and they don't get as loud as conventional speakers. But, the bigger ones seem to measure much better than the LRS that Amir measured here, and shouldn't be painted with the same brush.

 

[KaiserSoze]

That's actually backwards. Power is conserved, Torque is not. By this I mean power at the engine crankshaft equals power at the dyno drum (minus frictional losses). Torque at the dyno drum is R times torque at the engine crankshaft, if the engine crankshaft spins R times as fast as the dyno drum (R is the overall gear ratio from crankshaft to drum).

For example, consider a 300 HP engine that makes 250 ft.lbs. of torque, where 2nd gear is 3:1 and the differential is 3:1. When you run this on the dyno in 2nd gear, the dyno drum receives 300 HP of power, and 250 * 3 * 3 = 2,250 ft.lbs. of torque. So what the dyno drum directly senses in power is the same as at the engine crankshaft. But what the dyno directly senses in torque is 2,250 ft.lbs. and the dyno has no idea how much torque the engine makes at the crankshaft, unless you tell it engine RPM so it can calculate the gear ratio and divide it out.

Thus, the dyno can directly measure the engine's power output. But the dyno cannot sense the torque at the engine crankshaft, which is what you want to know. It can only sense torque applied at the drum. These are not the same, because Torque is not conserved. To give engine crankshaft torque, the dyno must know the engine RPM so it can compute R (the overall gear ratio), and divide it out.

If that's not convincing, try this practical experiment: run the car on the dyno without connecting the dyno's engine RPM sensor. The dyno will still tell you the engine's power output, but it can't tell you the engine's torque output.

All of this is absolutely correct and absolutely pertinent. With one minor quibble. From a very fundamental philosophical perspective on measuring stuff, it does not make substantive sense to think that when the measure of some quantity has been taken, that there would be some specific circumstances where this will have been done "actually", and other specific circumstances where this will not have been done "actually". If the measure of some quantity has been taken, it has been taken in a way that is every bit as "actual" as any other way to do it. If it was done, it was actually done. You may think I'm being silly, but I can promise you that if you were to approach any PhD in physics with this question, and ask whether some ways of taking a measurement are less actual than other ways of taking a measurement, the response you are most likely to get is an outburst of laughter.

 

[Vasr]

... it does not make substantive sense to think that when the measure of some quantity has been taken, that there would be some specific circumstances where this will have been done "actually", and other specific circumstances where this will not have been done "actually".

I thought the difference being talked about was in direct (actual) measured vs computed (from other measurements). The latter is subject to both margin of error in measurement as in the former but also the correctness and precision of the computed formula itself. Sometimes, the formulas may be just approximations which can differ from direct measurements if possible to do so.

In audio measurements, you can measure in an anechoic chamber, have a model of room acoustics and compute the "room measurement" without actually measuring in a room. It is only as good or as bad as the model used. This can be done even if it is possible to do the "actual" measurement in a room.

Unless I am misunderstanding the context.

 

[Francis Vaughan]

There has always been a problem when discussing Klippel measurements because there is a lot of reused terminology that confuses people who come to the system expecting a simple scale up of what they already know. It isn’t helped by Klippel.
what is important to realise is that the measurements taken in a space around the speaker are not just summed or averaged into some final result. The system doesn’t just take measurements at one distance. It takes enough measurements to be able to solve the equations governing the actual propagation of sound in the air. This isn’t trivial. It has to solve for things like reflections off the surface of the speaker. But isn’t the end there is a very complete model that describes the propagating sound in 3D. Once you have this it is possible to pick any location in 3D space and calculate the sound at that point. This calculation fully takes into account the emission from every part of the speaker. For boxes it takes into account radiation from the box panels, ports, the lot, and does so so that the location in space and extent in space is correctly folded into the calculation. This is why it is a pseudo anechoic measurement. It can provide the same information as allowing you the put the speaker in an anechoic chamber and measuring at as many locations as you wish.
Where there has been misunderstanding in the past has included interpretation of some of the values in the results. For instance, the pressure levels are normalised to a 1 metre measurement, but are calculated at a different distance. This does not mean the sound field at 1 metre looks like the results. It is just a scaling of the numbers. The shape of the response is correct at the distance it was calculated for, implicitly taking into account the entire size and shape of the speaker and its components.
But more than this, it can calculate the effect of reflections off surfaces, which you can’t with just anechoic measurements. Not unless you put the reflector in the chamber and measured. This is because the calculated model yields the propagating sound field, not point frequency response numbers.
What is maybe missing from an understanding of dipole speakers is how much effort the Klippel modelling does when asked to predict in room response. This is a potentially huge computational task, even once the actual measurements are done and processed. It isn’t just calculating floor wall and ceiling bounce. The information is there in the measured model. It would be very useful to know the available parameters to the room model and how Klippel uses them in calculating the predicted in room response.

 

[Francis Vaughan]

You may think I'm being silly, but I can promise you that if you were to approach any PhD in physics with this question, and ask whether some ways of taking a measurement are less actual than other ways of taking a measurement, the response you are most likely to get is an outburst of laughter.

For classical physics absolutely.
What you have described it not too far off one of the core philosophical problems in quantum mechanics- that of the measurement problem. Don’t anyone think that this is relevant to this topic. Ever.

 

[thewas]

That's actually backwards. Power is conserved, Torque is not. By this I mean power at the engine crankshaft equals power at the dyno drum (minus frictional losses).

You say it "minus frictional losses" so its not a conservative system where power is conserved, unless you count the all the losses (friction=>heat) which cannot be measured easily though.
On the other hand Newtons third law says that actio is always equal to reactio, thus when one body applies a force (or torque) to another then the other always applies the same force (or torque) to the first one with opposite direction and therefore that principle is used at dynos to calculate the power.
The engine or wheels accelerate the dyno which has a known moment of inertia (the equivalent to mass for rotation) and from Newtons second law F=m*a and in rotational domain torque=moment of inertia*rotational acceleration the torque can be directly calculated which is done to calculate also the power as known.
Now about the mechanical losses (frictions) at the transmission. tires etc-, these are measured when the engine doesn't create a torque and thus the frictions slow down the dyno from which their (negative) torque can be calculated again from the same equation.

 

[Vasr]

What is maybe missing from an understanding of dipole speakers is how much effort the Klippel modelling does when asked to predict in room response. This is a potentially huge computational task, even once the actual measurements are done and processed. It isn’t just calculating floor wall and ceiling bounce. The information is there in the measured model. It would be very useful to know the available parameters to the room model and how Klippel uses them in calculating the predicted in room response.

Agreed.

I would suggest further that people not conflate predicted (i.e., computed) in-room response with actual room measurement in a discussion to keep context clear, however close they may believe it to be. To add to the sufficiency of modeling, there is the issue of measurement in different context than the scenario in which the speakers are used (for example, raised off the floor for measurement). Whether the model compensates for this to predict in-room response when the speakers are on the floor and if it does for what kind of speakers, I don't have a clue. Experts in this area might know.

 

[Shazb0t]

I apologize for having not maintained a proper tone of civility. I am entirely certain that you and I have never met.

The point is that there must be a reason why the vast majority of speakers built in the past fifty years have avoided placing tweeters side-by-side with either a midrange or a woofer. I should not have ended on the sentence that you quoted. It was unnecessary and was not appropriate. I will try harder to behave like a civilized person.

You've been more than fair to him. All the fake expert word salad clogging up this thread and he's been dead wrong from the very start. He even knows that he's wrong now and still won't fully acknowledge it.

 

[Vasr]

But, the bigger ones seem to measure much better than the LRS that Amir measured here, and shouldn't be painted with the same brush.

The REW measurements for the 25 year old SMGa (comparable panel size) I posted earlier didn't have the extension of the LRS (100hz vs 50hz) but it didn't have the exaggerated low-mid peak like the LRS as measured here either, so is it less bass deficient than the LRS as interpreted here (without that peak, it doesn't have as steep a fall off on either side of the peak back to the relatively flat line)? Would I hear better bass on it? Absolutely not because the bass extension just isn't as deep as the LRS which seems to go down to about 50hz before falling off. But the interpretation here says I might. This is why I have a disagreement with that interpretation of the measurement here.

I suspect the LRS would likely sound better than the old SMGa (but be a bit more finicky with the amp requirements). The SMGa sounds (even after my manual gluing restoration) far better to me than the subjective listening test report here would suggest. So, I am not sure how much salt to use in reading the conclusions from the review.

 

[MRC01]

You say it "minus frictional losses" so its not a conservative system where power is conserved, unless you count the heat losses which cannot be measured easily though. ...

Saying that power is conserved does not imply that the system is 100% efficient. We say energy is conserved, even though energy is lost through friction, heat, etc. Power is the same at every point in the drivetrain, except for these losses, which are small. By contrast, torque is not conserved because it is multiplied (or divided) by gearing. Yet when you do this, RPM * torque is constant, and that is power. That is, if you gear it down N:1 you get N times the torque at 1/N the RPM, and their product (power) remains constant.

... On the other hand Newtons third law says that actio is equal reactio, thus when one body applies a force (or torque) to another then the other always applies the same force (or torque) to the first one with opposite direction and therefore that principle is used at dynos to calculate the power. ...

The dyno doesn't even have to know the torque at the drum. It can compute power from the mass of the drum and its rotational speed. If you know the drum was spinning at rate W1, then T seconds later it was spinning at rate W2, then you know how much energy it gained in T seconds, which is power. The dyno doesn't know or care and can't even tell whether that energy came from an engine with 100 ft.lbs. at 7500 RPM through a 3:1 gear ratio, or from an engine with 300 ft.lbs. at 2500 RPM through a 1:1 ratio. Either way, that (or infinitely many other different combinations of torque, RPM and gearing) is 143 HP and it has the exact same impact on the drum.

 

[thewas]

We say energy is conserved, even though energy is lost through friction, heat, etc.

Its what I wrote above, but officially a system is called conservative when there are no frictional losses.

Power is the same at every point in the drivetrain, except for these losses, which are small.

Nope, the losses at car drivetrains etc are huge and not neglible (up to 20%), thats why also a no-load run is performed as I wrote above to determine them.

By contrast, torque is not conserved because it is multiplied (or divided) by gearing. Yet when you do this, RPM * torque is constant, and that is power. That is, if you gear it down N:1 you get N times the torque at 1/N the RPM, and their product (power) remains constant.

Of course you need to consider the gear ratio, but that is known while the energy losses are neither known or easy to measure in a garage.

If you know the drum was spinning at rate W1, then T seconds later it was spinning at rate W2, then you know how much energy it gained in T seconds, which is power.

Which is nothing different than above 3rd Newton torque equation multiplied by rotational velocity, those are all interchangeable descriptions.

 

[Robbo99999]

I'd say it's probably better to keep on topic, but I understand people want to solidify people's impressions of their intelligence & general scientific knowledge/ability/background.....but best to illustrate this by intelligent replies related to "audio science" rather than on the minutia of an analogy (car dynos) that was used to illustrate one point many many many pages back.

 

[Objectivist01]

And so many positive reviews !!!!!

 

[Juhazi]

The mathematics is not limited to describing a point source. It only uses the spherical coordinate system (in this case). You can take a look at this paper to see an example of the application of this method to characterize the sound radiation pattern from a vibrating rectangular plate and the comparison to the theoretical prediction.
https://www.xlrtechs.com/dbkeele.com/PDF/Keele (2006-10 AES Preprint) - Full Sphere Measurements Using HELS Method.pdf

Well well, NFS needs a multiplexer for tall speakers/line arrays with multiple/tall drivers having same band... I have mentioned this before, but Amir responded that it applies for PA tower. Who is right here I'm left unanswered!

https://www.klippel.de/products/rd-system/modules/nfs-near-field-scanner.html
The sound pressure output of loudspeaker systems with multiple transducers (line sources, sound bars, 3-way systems, etc.) can be determined by measuring each transducer separately using a multiplexer. After the holographic processing the sound pressure of the individual source can be superimposed to determine the total sound pressure output of the audio device.

 

[Joppe Peelen]

I apologize for having not maintained a proper tone of civility. I am entirely certain that you and I have never met.

The point is that there must be a reason why the vast majority of speakers built in the past fifty years have avoided placing tweeters side-by-side with either a midrange or a woofer. I should not have ended on the sentence that you quoted. It was unnecessary and was not appropriate. I will try harder to behave like a civilized person.

Ok no problem all is fine You are right indeed ,i dont know of any normal speaker layed out like that! Thought about the desk monitors, but also they are the way you describe ! Except every line source. Since it is physically not possible to make a line source otherwise as far as i know. I admit a midrange in between would be nicer ofcourse in this case.

 

[Francis Vaughan]

Well well, NFS needs a multiplexer

If the speaker is too tall to manage, yes you need the multiplexing. You changed the word “can” to “needs”. That is not a reasonable thing to do. It totally changes the meaning. No it does not need one. It can use one if needed. Thinking the Klippel system cannot measure a 3 way system is totally misunderstanding how it works.

 

[Frank Dernie]

I want to experience as is and not with artificial "imaging" added to it all. It gets tiring.

But isn't the sound of wide dispersion speakers in domestic rooms preferred, and need to have even off axis response, for exactly this same reason?
Any speaker using the room reflections to give a more spatial sound is adding euphony, whether it is panels or wide dispersion box speakers?
I find narrower directivity more accurate on my own recordings, I would imagine I would be one of the minority who would find the JBL M2 to be better than the Revel in the comparison I read about.