Magico A5 spinorama (CEA2034)

[BobR]

The measurements in the original post are much more accurate and detailed, though. I don't see why we should even need to look at measurements like this when we already have the best (most accurate) data.

Correct, but since they are limited in range, they do not show how close these are to the ideal pistonic tweeter.

 

[Beave]

Oh believe me it's important. But don't take my word for it - try adding a 1dB high shelf filter to 10khz-20khz. It makes a huge difference in voicing. In fact, I adjust in 0.5dB increments.

It's amazing the logical stretches people will make just to defend the beautiful directivity.

You either have extraordinary high frequency hearing and/or you listen to some really weird music with a lot of content above 10kHz.

 

[preload]

You either have extraordinary high frequency hearing and/or you listen to some really weird music with a lot of content above 10kHz.

Try it!!

 

[richard12511]

Oh believe me it's important. But don't take my word for it - try adding a 1dB high shelf filter to 10khz-20khz. It makes a huge difference in voicing. In fact, I adjust in 0.5dB increments.

It's amazing the logical stretches people will make just to defend the beautiful directivity.

Already tried that experiment hundreds of times , fine tuning my various systems, so I know it matters. After many hours going back and forth trying various combinations of high frequency shelving, I landed on a high shelf at 10,000Hz , -~1db, .5Q. No one's saying it doesn't matter(as far as I'm aware). We're saying it matters less than the other octaves. I 1dB cut from 200-400Hz or 2-4kHz for example is much more audible. Would you agree?

As someone who fully embraces EQ(even full range EQ if it sounds better), directivity is by far the most important factor for me. I do think FR matters more, but FR is easy to fix with EQ. Directivity you're stuck with, so better get it right on the production side.

*Edit: a little bit higher

 

[richard12511]

Correct, but since they are limited in range, they do not show how close these are to the ideal pistonic tweeter.

That's true, but I don't think anyone's really saying that the treble is rolled off because the tweeter is bad. I think it's probably a great tweeter. To me, based on what @ctrl has shown, it's clear that the reduced treble is an intentional crossover design choice, probably because the rolled off treble does better in listening tests.

 

[preload]

Already tried that experiment hundreds of times , fine tuning my various systems, so I know it matters. After many hours going back and forth trying various combinations of high frequency shelving, I landed on a high shelf at 10,000Hz , -~1db, .5Q. No one's saying it doesn't matter(as far as I'm aware). We're saying it matters less than the other octaves. I 1dB cut from 200-400Hz or 2-4kHz for example is much more audible. Would you agree?

No. @Beave wrote that "the top octave is probably the least important octave for frequency response errors." I don't agree with that. Certainly, a 1 dB cut from 20-40Hz is going to be less noticeable than a 1dB cut from 10khz-20khz. I can't imagine anyone making such minute adjustments in that range.

As someone who fully embraces EQ(even full range EQ if it sounds better), directivity is by far the most important factor for me. I do think FR matters more, but FR is easy to fix with EQ. Directivity you're stuck with, so better get it right on the production side.

*Edit: a little bit higher

I agree with that to a certain extent. If you want to make some minor adjustments and correct for little response bumps, it's nice to have constant directivity, but it still won't be perfect because of possible cancellation/augmentation delayed reflected sound. But all bets are off when you your FR goes down the roller coaster after 7khz. A pair of Radio Shack speakers with paper tweeter cones don't even do that.

 

[Ata]

Olive score is 6.0 and w/sub it would be 7.8. That's a good speaker.
Score could improve up to 6.7 with an EQ but it doesn't change the tonal balance much and I do not think it is required.

EQ for Magico A5 computed from Misc data
Generated from http://github.com/pierreaubert/spinorama/generate_peqs.py v0.8
Dated: 2021-06-01-18:52:19

Preamp: -3.5 dB

Filter  1: ON PK Fc   837 Hz Gain -0.77 dB Q 10.26
Filter  2: ON PK Fc 12863 Hz Gain +3.07 dB Q 1.15
Filter  3: ON PK Fc  6780 Hz Gain -1.13 dB Q 4.44
Filter  4: ON PK Fc 11752 Hz Gain -1.08 dB Q 9.87
Filter  5: ON PK Fc  1727 Hz Gain -0.72 dB Q 2.80
Filter  6: ON PK Fc  1166 Hz Gain +0.73 dB Q 12.00
Filter  7: ON PK Fc  5279 Hz Gain +0.37 dB Q 12.00
Filter  8: ON PK Fc   449 Hz Gain +0.80 dB Q 3.13
Filter  9: ON PK Fc  4586 Hz Gain -0.30 dB Q 12.00
Filter 10: ON PK Fc   979 Hz Gain +0.24 dB Q 12.00
Filter 11: ON PK Fc 12552 Hz Gain -0.26 dB Q 12.00
Filter 12: ON PK Fc  2079 Hz Gain -0.44 dB Q 12.00
Filter 13: ON PK Fc  2317 Hz Gain -0.31 dB Q 12.00
Filter 14: ON PK Fc  3148 Hz Gain +0.39 dB Q 5.59
Filter 15: ON PK Fc  1347 Hz Gain -0.72 dB Q 12.00
Filter 16: ON PK Fc  1226 Hz Gain +0.28 dB Q 12.00
Filter 17: ON PK Fc   887 Hz Gain -0.20 dB Q 12.00
Filter 18: ON PK Fc  1312 Hz Gain -0.20 dB Q 12.00
Filter 19: ON PK Fc 12863 Hz Gain +0.55 dB Q 0.92
Filter 20: ON PK Fc 11762 Hz Gain -0.65 dB Q 4.02

View attachment 133183

View attachment 133184View attachment 133185

Hi Pierre,

Do these adjustments use the AutoEq tool or a different algorithm for PEQs?

What I am looking for is a tool that can generate optimal FIR filter from a single measurement, or a complete spinorama. The reasons are:
- I can easily run many tap FIRs on my PC and find it easier to work with than dialling in PEQs;
- I expect the FIR can achieve near perfect FR at MLP, assuming of course, that it mostly cuts rather than boosts throughout the frequency range;
- Bonus points if this can be done live (periodic pink noise gen on computer-> DAC -> amp -> speakers -> mic -> FIR tool -> FIR filter -> FIR corrected periodic pink noise -> ...) to automatically achieve near perfect room correction with live data.

Does this makes sense? On paper it sounds like a great idea, I have asked elsewhere on this forum but without responses. Thanks!

 

[Beave]

No. @Beave wrote that "the top octave is probably the least important octave for frequency response errors." I don't agree with that. Certainly, a 1 dB cut from 20-40Hz is going to be less noticeable than a 1dB cut from 10khz-20khz. I can't imagine anyone making such minute adjustments in that range.

OK, you got me. I'll concede that the octave from 20-40Hz is arguably less important than the top octave. Happy now?

But a 1dB cut from 20-40Hz will or will not be audibly significant depending on speakers (can they even do that octave?) and program material (anything even there?).

As for your suggestions about trying filters for the top octave, the majority of the difference you'll hear is at the 10kHz region, not the 20kHz region. Try a filter that doesn't change 10kHz but slopes gradually up or down from 10kHz to 20kHz. It'll be much less audible.

 

[richard12511]

No. @Beave wrote that "the top octave is probably the least important octave for frequency response errors." I don't agree with that. Certainly, a 1 dB cut from 20-40Hz is going to be less noticeable than a 1dB cut from 10khz-20khz. I can't imagine anyone making such minute adjustments in that range.

I don't think it's a huge stretch to say, tbh. Probably a matter of personal preference, but 20-40Hz is more important to me than 10-20kHz, as it has a huge impact on the overall tactile impact for a lot of pop and electronic music(which is a lot of what I listen to). I've spent a ton of time trying to optimize the slope in that area, and very slight slope differences cause different physical experiences that really affect my overall preference. 10-20Hz, yeah that's less important, even for me, same with 5-10Hz . I don't think @Beave really had the infrasonic bass in mind when he made his statement, so I can forgive him the slight exaggeration . I'd probably rate 10-20kHz as the second least important octave for me personally(ignoring 1.25-2.5, 2.5 - 5, 5-10).

 

[preload]

OK, you got me. I'll concede that the octave from 20-40Hz is arguably less important than the top octave. Happy now?

Thank you, and no.

But a 1dB cut from 20-40Hz will or will not be audibly significant depending on speakers (can they even do that octave?) and program material (anything even there?).

I agree and that's kind of my point. Who (other than Richard12511) is going to notice a 1dB difference between 20-40Hz?

As for your suggestions about trying filters for the top octave, the majority of the difference you'll hear is at the 10kHz region, not the 20kHz region. Try a filter that doesn't change 10kHz but slopes gradually up or down from 10kHz to 20kHz. It'll be much less audible.

I have tried both shelving and peaking filters when adjusting a pair of Genelecs to my listening room, including ones that slope up gradually with frequency. Even small adjustments are audible. 1db that affects that entire top octave is "very noticeable."

I don't think it's a huge stretch to say, tbh. Probably a matter of personal preference, but 20-40Hz is more important to me than 10-20kHz, as it has a huge impact on the overall tactile impact for a lot of pop and electronic music(which is a lot of what I listen to). I've spent a ton of time trying to optimize the slope in that area, and very slight slope differences cause different physical experiences that really affect my overall preference.

Really? You can hear 0.5-1dB differences in slope from 20-40Hz? Amazing.

, so I can forgive him the slight exaggeration . I'd probably rate 10-20kHz as the second least important octave for me personally(ignoring 1.25-2.5, 2.5 - 5, 5-10).

The bottom line is that a -6db downslope starting at 7kHz is not trivial.

 

[Beave]

The bottom line is that a -6db downslope starting at 7kHz is not trivial.

What about a 5dB downslope starting at 10kHz, as shown in the Stereophile plot?

The Klippel response showing rolloff starting at 7kHz looks more worrisome to me than the Stereophile response showing rolloff starting around 10kHz.

Which one is the "correct" measurement? Are the differences due to measurement techniques, mics, calibration, or unit-to-unit variations in the speakers?

 

[ctrl]

Which one is the "correct" measurement? Are the differences due to measurement techniques, mics, calibration, or unit-to-unit variations in the speakers?

For the CTA-2034-A standard, the "first" frequency response shown is the axis frequency response.

Stereophile measurement states:

...and we do not know if the exact same reference point was taken for both measurements.

But the tendency of the declining axis frequency response is clearly visible in both measurements.

Correct, but since they are limited in range, they do not show how close these are to the ideal pistonic tweeter.

With every post you try to sell us that the early roll-off of the axis frequency response occurs because the tweeter shows ideal pistonic behavior.

Without facts this is pure marketing for Magicoaudio, please prove your statement with facts.

In Post#64 I have shown, with the simulation of an ideal 28mm tweeter, that your statement is not true - there is no "natural" early frequency response roll-off on axis.

We can also have an animated look (6kHz-30kHz) at the overall radiation pattern of an ideal (in this case 1'') tweeter (without waveguide) with high resonant frequency (fs=1.2kHz).
There, too, it can be clearly seen that only above 14kHz the frequency response decreases on axis - the color for the 0° direction (0 at x-axis) does not change for a fixed distance (y-axis). The frequency is displayed on the top right.

FR deg0-80 ideal 1'' tweeter, fs=1.2kHz

 

[preload]

What about a 5dB downslope starting at 10kHz, as shown in the Stereophile plot?

The Klippel response showing rolloff starting at 7kHz looks more worrisome to me than the Stereophile response showing rolloff starting around 10kHz.

Exactly. There's a hf rolloff there.

Which one is the "correct" measurement? Are the differences due to measurement techniques, mics, calibration, or unit-to-unit variations in the speakers?

For starters, the difference is that the cea2034 listening window includes slight vertical off axis measurements as well as the 30def horizontal off axis. The stereophile listening window is 30 dega horizontal only.

 

[richard12511]

I agree and that's kind of my point. Who (other than Richard12511) is going to notice a 1dB difference between 20-40Hz?

Maybe I'll try recording it(umik-1?), but there's certainly a noticeable difference between my 2 target curves on some songs. It ruins some songs for me. I'd be shocked if others couldn't feel it, but maybe I'm wrong. At loud volumes, I feel like I feel it easily.

I have tried both shelving and peaking filters when adjusting a pair of Genelecs to my listening room, including ones that slope up gradually with frequency. Even small adjustments are audible. 1db that affects that entire top octave is "very noticeable."

Oh it's definitely audible(more audible than the 20-40 bass difference), it just doesn't affect my overall enjoyment very much. Mostly what I've found with my top adjustments(usually land between -0 and -3.5 at 20kHz) is that I'm trading small amounts of enjoyment for small amounts of endurance, but I could live without it. Bass on the other hand can be make or break for me.

Interesting that you've tried sloping up the 8351s treble. To me they're already hair too bright out of the box. Just goes to show how different preferences can be.

Really? You can hear 0.5-1dB differences in slope from 20-40Hz? Amazing.

At loud volumes and certain material I can feel the difference. I have two different sub 50Hz filters I run, .5db/oct, and 1.2db/oct, and I regularly flip back and forth(2 separate convolution files). I've also found that there can be a fine line between pieces of the house vibrating or not vibrating, and sometimes one slope will excite that, but not the other. I think it makes a real difference for some music, and none at all for others. I hear it less so than I feel it, but the difference in joy is greater for me than that top octave difference, as I just care about that range so much more. The 1.2 bass slope ruins some songs, but awakens others. By comparison, I fiddle with the top shelf filter much less. Could be a personal preference thing. 20-100Hz is maybe the most important 3rd of the graph for me, but I'm also kinda a bass head.

The bottom line is that a -6db downslope starting at 7kHz is not trivial.

I'm with you here, and I wouldn't buy this speaker because of that. It's just too much of a roll off in my eyes. One of the biggest I've ever seen.The Revel F328Be looks closer to what I'd like. My defense of this speaker has been about the reason for that downslope. My opinion is that it's an intentional deviation from neutral, not a failure of engineering(similar to the 800 series B&Ws). I think Magico(like B&W) has the engineering to get basically whatever response they want. They chose this response because they think thinks it sounds better(maybe internal listening data?), even though I don't agree with it.

I guess inserted myself in you and @Beave's conversation because I thought you kinda jumped on him unnecessarily for what was a fairly off the cuff statement that was almost(but not quite) correct. I see it mentioned on here quite a bit that many don't really pay attention to errors above 10kHz all that much. Generally, I don't either, though I'd still prefer flat. I didn't think Beave was going out of his way to praise Magico(but maybe he has history I don't know of) unfairly, which is what seemed to upset you. I also think he was mainly thinking in context of this speaker(50-20,000Hz), and I can forgive him for forgetting that bottom octave.

I'm curious, though, other than 20-40Hz octave, which octaves would you consider less important than the top octave? Are there any octaves above 100Hz that matter less to you than the 10-20kHz octave?

 

[Beave]

For what it's worth (not much at all), I have no affiliation with Magico, have never seen nor heard a Magico speaker, would never buy such an expensive speaker, and agree that the Klippel measurement showing a droop after 7kHz is a bit concerning. I'm not praising or defending their performance.

It was an off the cuff statement that I feel is kinda true, but yeah, you can call me out for not being precise in my language.

When I made the statement, I was looking at the Stereophile measurements, not the Klippel plot posted on page 1. To me, the Stereophile measurements show a response far less concerning: flat to about 10kHz, then a smooth drop to 20kHz.

 

[preload]

Maybe I'll try recording it(umik-1?), but there's certainly a noticeable difference between my 2 target curves on some songs. It ruins some songs for me. I'd be shocked if others couldn't feel it, but maybe I'm wrong. At loud volumes, I feel like I feel it easily.

At loud volumes and certain material I can feel the difference. I have two different sub 50Hz filters I run, .5db/oct, and 1.2db/oct, and I regularly flip back and forth(2 separate convolution files). I've also found that there can be a fine line between pieces of the house vibrating or not vibrating, and sometimes one slope will excite that, but not the other. I think it makes a real difference for some music, and none at all for others. I hear it less so than I feel it, but the difference in joy is greater for me than that top octave difference, as I just care about that range so much more. The 1.2 bass slope ruins some songs, but awakens others. By comparison, I fiddle with the top shelf filter much less. Could be a personal preference thing. 20-100Hz is maybe the most important 3rd of the graph for me, but I'm also kinda a bass head.

In theory, a 1dB difference with 1 oct bandwidth in the 20-40hz range should not be audible under blind listening conditions, so again, you must have amazing hearing, though this threshold varies with overall volume.

 

[preload]

For what it's worth (not much at all), I have no affiliation with Magico, have never seen nor heard a Magico speaker, would never buy such an expensive speaker, and agree that the Klippel measurement showing a droop after 7kHz is a bit concerning. I'm not praising or defending their performance.

I'm willing to bet they sound pretty good, as Richard suggested as well. Rather my opinion is that measurements are not completely reliable predictors of perceived SQ, and there is no consensus method of interpreting them, as responses to this thread have clearly illustrated. And certainly the incredible similarity in stereophile FR measurement to the $1,200 Kef LSX supports the notion that loudspeaker measurements are far from predictive. (see my earlier post).

When I made the statement, I was looking at the Stereophile measurements, not the Klippel plot posted on page 1. To me, the Stereophile measurements show a response far less concerning: flat to about 10kHz, then a smooth drop to 20kHz.

Not sure how you can simply ignore the klippel cea2034, which is at least a standard way of measuring the listening window, which considers a little vertical off axis as well as horizontal 30 degs. It's not the same methodology used by stereophile. Both should be concerning.

 

[Beave]

I'm willing to bet they sound pretty good, as Richard suggested as well. Rather my opinion is that measurements are not completely reliable predictors of perceived SQ, and there is no consensus method of interpreting them, as responses to this thread have clearly illustrated. And certainly the incredible similarity in stereophile FR measurement to the $1,200 Kef LSX supports the notion that loudspeaker measurements are far from predictive. (see my earlier post).

That "incredible similarity" is on-axis only, doesn't account for that possibly nasty port resonance on the KEF, doesn't include off-axis response, doesn't include bass extension, doesn't include distortion, and doesn't include *anything* about compression at loud levels.

So yea, I agree that on-axis only plots aren't very predictive.

Not sure how you can simply ignore the klippel cea2034, which is at least a standard way of measuring the listening window, which considers a little vertical off axis as well as horizontal 30 degs. It's not the same methodology used by stereophile. Both should be concerning.

I don't think I did ignore the klippel plot. I'm also not convinced that the differences in rolloff between the two sets of measurements are simply a result of listening window averaging.

 

[HooStat]

Someone else asked this above, but I am curious as to what people thing of a pair of speakers with 3 x 9" drivers for bass not really producing much in the way of deep bass. Is this a situation where Magico has opted for maximizing volume and minimizing distortion over extension? They claim it can produce 115 db at 50 Hz (not sure what distortion level, of course). The in-room bass response was pretty nice in the Soundstage review though. It is based on an average of 9 positions across a 20' arc in the reviewer's room. Obviously room-specific but it does show that "in-room" can be quite different from anechoic.

 

[richard12511]

^ If I ever become billionaireD) I'd probably have a pair of these in one of the rooms in my mansion. I'd also love to see the NFS CEA-2034 of these. I'm sure Magico already has it.