• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Looking for insights into Magico's amazing off-axis measurements

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
525
Likes
1,035
Location
Princeton, Texas
#1
The off-axis response of early generation Magico speakers was nothing special. Here are Stereophile's normalized horizontal off-axis curves for the Magico V3, and the radiation pattern mis-match between the 6" midwoofer and 1" tweeter in the crossover region is pretty obvious. We see the midwoofer beaming in the 2 kHz ballpark, and then the tweeter's WIDE off-axis response at the bottom end of its range, peaking at about 3.5 kHz:

MagicoV3.offaxis.jpg


We also see the ring radiator tweeter beaming pretty badly, which they tend to do.

But then something happened. Alon Wolff made some dramatic improvements, as you will see. Here are the normalized horizontal off-axis curves for the much more recent Magico M2, which uses a 6" midwoofer just like the V3, and a 1" soft dome tweeter:

MagicoM2.offaxis.jpg


GONE is that severe radiation pattern discontinuity in the crossover region!! Now ONE reason for the visible improvement is that the scale is changed - the M2's off-axis curves only go out to 45 degrees, whereas the V3's off-axis curves went all the way out to 90 degrees. BUT STILL if you pick out the 45 degree curve on the V3, that radiation pattern discontinuity is obvious.

Let's look at another recent Magico, the S5 which likewise uses a 6" midwoofer and a 1" dome tweeter:

MagicoS5.offaxis.jpg


Once again the scale is different, this time it only goes out to 30 degrees. BUT there isn't even a HINT of any radiation pattern discontinuity between the 6" midwoofer and 1" dome tweeter!! And again if you look a V3's 30 degree curve, the discontinuity is obvious.

Anybody have any guesses or ideas or theories as to how Magico is now able to transition from a 6" midwoofer to a 1" dome tweeter in the M2 and S5 WITHOUT having any radiation pattern discontinuity, such as was seen in the earlier V3?

Thanks!
 

Matias

Major Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
1,620
Likes
2,274
Location
São Paulo, Brazil
#2

Beave

Active Member
Joined
May 10, 2020
Messages
205
Likes
307
#3
Keep in mind that the off-axis plots are normalized to the on-axis response.

In the case of the V3, the on-axis response shows a dip above the crossover region, at around 3-4kHz. So the off-axis response, if it's relatively well-controlled, will show a bump there due to the plots being normalized. The non-normalized off-axis response wouldn't look as bad as what is shown.

The M2 is flatter in this region on-axis, so the off-axis normalized response is closer to what the off-axis non-normalized response would be.
 

richard12511

Major Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
1,928
Likes
2,607
#5
OP
Duke

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
525
Likes
1,035
Location
Princeton, Texas
Thread Starter #6
Keep in mind that the off-axis plots are normalized to the on-axis response.
Right, so any deviations from "flat" in the normalized curve are DISCREPANCIES between the on-axis response and the off-axis response. Such discrepancies are obvious on the V3 (first set of curves), but not on the M2 or S5.

In the case of the V3, the on-axis response shows a dip above the crossover region, at around 3-4kHz. So the off-axis response, if it's relatively well-controlled, will show a bump there due to the plots being normalized.
My guess is that in the V3, Alon Wolff compensated for the tweeter's off-axis flare (relative to the midwoofer's narrow pattern in the crossover region) by introducing an on-axis dip at the bottom end of the tweeter's passband.

The M2 is flatter in this region on-axis, so the off-axis normalized response is closer to what the off-axis non-normalized response would be.
I'd be interested in any thoughts you might have as to why the M2's off-axis response is so flat. Assuming the crossover is in the 2-3 kHz ballpark, which seems reasonable, normally we'd expect a 6" midwoofer to have a much narrower pattern than a 1" dome tweeter (as the V3's curves indicate), but that doesn't seem to be the case with the M2.

I think Magicos measure reasonably well (evidence here: https://speakerdata2034.blogspot.com/?m=1), but I also think a lot of speakers would look that good at only 30 degrees off axis.
Maybe so, but on the M2 we have data out to 45 degrees. I don't recall seeing such smooth off-axis behavior out to 45 degrees from a 6" woofer and non-waveguided 1" tweeter combination. It's arguably even better than the 45-degree curve of the Revel Salon 2, which uses a SMALLER diameter midrange (4") and a waveguided tweeter:

RevelSalon2.offaxis.jpg
 
Last edited:

Beave

Active Member
Joined
May 10, 2020
Messages
205
Likes
307
#7
Total guess - they lowered the crossover when they changed from the ring radiator tweeter to the Be dome.
 
OP
Duke

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
525
Likes
1,035
Location
Princeton, Texas
Thread Starter #8
Total guess - they lowered the crossover when they changed from the ring radiator tweeter to the Be dome.
Could be. Hmmm. That would explain the midwoofer and tweeter pattern widths being similar out to 30 or 45 degrees. Seems to me that a low crossover would be very demanding of the tweeter as far as power handling goes (both the M2 and S5 are rated for use with amps up to 1000 watts). However that may be an engineering problem Magico has solved.

Thank you for your "total guess" - I don't have a better one!!
 
Last edited:

Ilkless

Addicted to Fun and Learning
Joined
Jan 26, 2019
Messages
986
Likes
1,803
Location
Singapore
#9
I think it's the cumulative effect of several incremental improvements: baffle geometry, a very slight waveguide milled seamlessly into the baffle and the implementation of their elliptical crossover.

Magico is proud of their elliptical crossover topology (Joseph is another one known for it). This topology gives very aggressive slopes - 30+-50dB/octave is not uncommon. Cross low, with cost-no-object drivers that can do so, optimise the baffle geometry so that diffraction and baffle step are not hindering the FR smoothness and directivity match, and mill the slightest waveguide to rein in the flaring in the far off-axis.

My own speakers are based on an existing kit, but only keep the CtC constant. The box volume, padding resistor and most crucially baffle geometry are different. My sims with the modified (wider, shorter) baffle show a difference in baffle diffraction that I'd imagine can be effectively harnessed in a favourable manner, by a company with good R&D capabilities. Technically, the Magico A1, based on what's described, is not strictly elliptical but a Butterworth with a tuned notch filter (which has sometimes been called elliptical but is not in the purest sense). It is unclear whether the larger speakers do use a truly elliptical topology or the other approach. Nonetheless, what's clear is the pursuit of extremely aggressive slopes.
 
Last edited:

kceenav

New Member
Joined
Sep 9, 2020
Messages
2
Likes
3
#10
Regarding the elliptical crossover:

I may be wrong - in this case - but isn't it a characteristic of these types that they are 'aggressive' once they reach their maximum slope, but are in fact rather mild in the crossover region? Meaning that there is a broad area of overlap between both drivers?

That would explain (imo) why there is hardly any narrowing in beam width around the point of crossover. Because the tweeter still contributes more than is usual to the output, even half an octave below the point of nominal hand-over.
 

Ilkless

Addicted to Fun and Learning
Joined
Jan 26, 2019
Messages
986
Likes
1,803
Location
Singapore
#12
As stated, keep in mind it’s normalized, so unless the on-axis is very flat, the off-axis could look worse than it is. Also, keep in mind the difference in how far off-axis each goes, from +/-90 to +/-45 to +/-30.
Normalisation in most cases helps isolate the directivity pattern from the actual magnitude response (unless there are severe anomalies that only appear on-axis such as cancellations). I find it useful to that extent. Some like Geddes find it doesn't (obviously because of on-axis holes in his OS waveguide implementation), but normalisation helps us filter away speakers with flaws that are correctable (eg. ADAM T7V - great directivity after normalisation, ragged rising native response, which implies easy EQ) vs those that aren't.

Regarding the elliptical crossover:

I may be wrong - in this case - but isn't it a characteristic of these types that they are 'aggressive' once they reach their maximum slope, but are in fact rather mild in the crossover region? Meaning that there is a broad area of overlap between both drivers?

That would explain (imo) why there is hardly any narrowing in beam width around the point of crossover. Because the tweeter still contributes more than is usual to the output, even half an octave below the point of nominal hand-over.
I agree, especially because they are presumably crossing low to yield such a pattern. Smooth blended transition towards the crossover, then a steep slope to suppress excursion right after.

1605094276359.png


Magico show as much.
 

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
2,901
Likes
6,706
Location
Land O’ Lakes, Florida
#13
Normalisation in most cases helps isolate the directivity pattern from the actual magnitude response (unless there are severe anomalies that only appear on-axis such as cancellations). I find it useful to that extent. Some like Geddes find it doesn't (obviously because of on-axis holes in his OS waveguide implementation), but normalisation helps us filter away speakers with flaws that are correctable (eg. ADAM T7V - great directivity after normalisation, ragged rising native response, which implies easy EQ) vs those that aren't.
Correct; it is typical to have cabinet diffractions though. For most cases, normalizing to the listening window may give a better picture in terms of directivity.
 

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,426
Likes
8,441
#14
Based on the impedance trace and vertical polar response of the M2, I don’t think they’re crossing particularly low (appears to be c. 2kHz).

1605094414317.jpeg

1605094390699.jpeg


So we really are looking at a woofer here that maintains wide dispersion up to 2-ish kHz (at least out to 45 degrees).
 
OP
Duke

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
525
Likes
1,035
Location
Princeton, Texas
Thread Starter #15
I think it's the cumulative effect of several incremental improvements: baffle geometry, a very slight waveguide milled seamlessly into the baffle and the implementation of their elliptical crossover.

Magico is proud of their elliptical crossover topology (Joseph is another one known for it). This topology gives very aggressive slopes - 30+-50dB/octave is not uncommon. Cross low, with cost-no-object drivers that can do so, optimise the baffle geometry so that diffraction and baffle step are not hindering the FR smoothness and directivity match, and mill the slightest waveguide to rein in the flaring in the far off-axis.

... what's clear is the pursuit of extremely aggressive slopes.
I think you're right. I hadn't appreciated the implications of their elliptical crossover topology as far as being able to significantly lower the crossover frequency, but I now think that's a major piece of the puzzle. And without needing high efficiency for the tweeter, Magico could make design tradeoffs in favor of bandwidth and power handling (mechanical as well as thermal).

(I also use accelerating-slope crossovers, but not ellipticals, as I'm trying to avoid peaks in the impedance curves.)

Based on the impedance trace and vertical polar response of the M2, I don’t think they’re crossing particularly low (appears to be c. 2kHz).
Could be, but on the other hand given that the crossover topology is "elliptical", then it might be lower than that 2 kHz peak would otherwise lead us to expect.

An elliptical crossover involves notch filters to accelerate the rolloff, one for the lowpass section and one for the highpass section. The lowpass section's notch would be above the crossover frequency, and the highpass section's notch would be below the crossover frequency. [speculation] So perhaps the 2 kHz impedance peak is from the lowpass section's notch filter, and the 350 Hz bumpage is from the highpass section's notch filter, implying that the crossover frequency is about midway in between them, ballpark 800 Hz perhaps. [/speculation]

So we really are looking at a woofer here that maintains wide dispersion up to 2-ish kHz (at least out to 45 degrees).
"Very wide dispersion midwoofer" was my initial thought as well. In order for a midwoofer cone of that size to have wide dispersion up that high it would need to be flexing, but the claims about the midwoofer cone's construction all seem to point to rigidity being the goal, and its behavior is supposedly pistonic from what I understand. If it WAS in breakup, that would be incredibly well-controlled breakup, which tends to be inconsistent with how a rigidity-optimized cone behaves when it does go into breakup.

I'm NOT saying your theory is wrong - it may well be right, or at least be a piece of the puzzle.
 

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,426
Likes
8,441
#16
An elliptical crossover involves notch filters to accelerate the rolloff, one for the lowpass section and one for the highpass section. The lowpass section's notch would be above the crossover frequency, and the highpass section's notch would be below the crossover frequency. [speculation] So perhaps the 2 kHz impedance peak is from the lowpass section's notch filter, and the 350 Hz bumpage is from the highpass section's notch filter, implying that the crossover frequency is about midway in between them, ballpark 800 Hz perhaps. [/speculation]
The thing is, if you look closely at the vertical polars, it appears very likely IMO that the XO is at c. 2kHz, as just above that frequency is where we see a suckout begin to form vertically off-axis at -15°:

1605112919212.png


If that suck-out weren't a result of the crossover, we would expect to see it in the horizontal polars as well (but of course we don't).

"Very wide dispersion midwoofer" was my initial thought as well. In order for a midwoofer cone of that size to have wide dispersion up that high it would need to be flexing, but the claims about the midwoofer cone's construction all seem to point to rigidity being the goal, and its behavior is supposedly pistonic from what I understand. If it WAS in breakup, that would be incredibly well-controlled breakup, which tends to be inconsistent with how a rigidity-optimized cone behaves when it does go into breakup.
That's quite true. My best speculation is that the thickness/mass of the diaphragm is nonuniform. I have no direct evidence of this, though.
 

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,426
Likes
8,441
#17
@Duke perhaps another small piece of the puzzle? Their "6-inch" midrange driver is apparently only 5" (although I admit it does look quite large in comparison to the tweeter!):

1605113751625.png
 
OP
Duke

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
525
Likes
1,035
Location
Princeton, Texas
Thread Starter #18
The thing is, if you look closely at the vertical polars, it appears very likely IMO that the XO is at c. 2kHz, as just above that frequency is where we see a suckout begin to form vertically off-axis at -15°
Excellent detective work! I hadn't noticed that. I think a deep notch in the woofer's response at 2 kHz, as with my speculative notch filter placement, would have precluded the possibility of a suckout like that. Thanks!

My best speculation is that the thickness/mass of the diaphragm is nonuniform. I have no direct evidence of this, though.
I think you're right! Here are some statements about Magico's midrange cone. From John Atkinson's write-up of the M2:

"The cones of the M2's 6" midrange driver and 7" woofers use multiple layers of woven carbon-fiber, incorporating graphene, a form of carbon in which the atoms are assembled in a sheet just a few atoms thick that is said to be 100 times stronger than steel. The resulting cone is both light and stiff, pushing breakup modes well above each unit's operating passband."

And from Magico's M2 page:

"The newly designed Magico six-inch midrange and seven-inch bass driver cones found in the M2 are both manufactured using Multi-Wall carbon XG Nanographene and an ultra-stiff proprietary new carbon weave which is 20% lighter and 300% stiffer than previous cone material we used in Magico loudspeaker offerings."

So these statements are emphasizing STIFFNESS, with John Atkinson stating that the breakup modes are "well above each unit's passband". However they don't actually use the word "pistonic".

It LOOKS to me like in this new generation of midrange cones "the thickness/mass of the diaphragm is nonuniform", EXACTLY as you said. From Atkinson's write-up of the S5:

"It's desirable that a speaker cone be of varying thickness: thickest at the center and the boundary with the voice-coil former, thinnest at the junction with the surround. However, Magico used to use a sandwich core of constant thickness, because the foam material would fracture if the thickness varied. For their new generation of midrange units they developed a process in which the foam is carefully injected between the front and back carbon-fiber, to permit the overall thickness to vary in the desired manner."

And finally, from Magico's "Drivers" technology page:

"The result is a cone that displays bending strength, self-damping and attenuation of ringing that are all an order of magnitude or more higher than that of conventionally manufactured cones."

In my opinion the word "bending" implies "non-pistonic" behavior, which I would have called "breakup". And in my opinion "self-damping" and "attenuation of ringing" imply WELL CONTROLLED breakup.

[I found this online definition of "cone breakup": "At low frequencies a cone moves as a whole. This is the 'pistonic' area of operation. At higher frequencies the cone starts to flex, leading to resonances. This is what is referred to as 'breakup'."]

So maybe this is what's going on: The new cone goes into extremely well-behaved breakup and flexes just the right amount to significantly widen the radiation pattern up into the crossover region, with any "nasty" breakup peaks being well above the driver's passband. And the marketing department chose to emphasize the strength and stiffness of the cone because admitting to "cone breakup" within the driver's passband would hurt sales.

And IF this is what's going on, then in my opinion the marketing department's selective wording IN NO WAY DIMINISHES the amazing accomplishment of Magico's engineering department.

I have ZERO affiliation with Magico... but mucho admiration for their engineering department.
 
Last edited:

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,426
Likes
8,441
#19
Excellent detective work! I hadn't noticed that. I think a deep notch in the woofer's response at 2 kHz, as with my speculative notch filter placement, would have precluded the possibility of a suckout like that. Thanks!



I think you're right! Here are some statements about Magico's midrange cone. From John Atkinson's write-up of the M2:

"The cones of the M2's 6" midrange driver and 7" woofers use multiple layers of woven carbon-fiber, incorporating graphene, a form of carbon in which the atoms are assembled in a sheet just a few atoms thick that is said to be 100 times stronger than steel. The resulting cone is both light and stiff, pushing breakup modes well above each unit's operating passband."

And from Magico's M2 page:

"The newly designed Magico six-inch midrange and seven-inch bass driver cones found in the M2 are both manufactured using Multi-Wall carbon XG Nanographene and an ultra-stiff proprietary new carbon weave which is 20% lighter and 300% stiffer than previous cone material we used in Magico loudspeaker offerings."

So these statements are emphasizing STIFFNESS, with John Atkinson stating that the breakup modes are "well above each unit's passband". However they don't actually use the word "pistonic".

It LOOKS to me like in this new generation of midrange cones "the thickness/mass of the diaphragm is nonuniform", EXACTLY as you said. From Atkinson's write-up of the S5:

"It's desirable that a speaker cone be of varying thickness: thickest at the center and the boundary with the voice-coil former, thinnest at the junction with the surround. However, Magico used to use a sandwich core of constant thickness, because the foam material would fracture if the thickness varied. For their new generation of midrange units they developed a process in which the foam is carefully injected between the front and back carbon-fiber, to permit the overall thickness to vary in the desired manner."

And finally, from Magico's "Drivers" technology page:

"The result is a cone that displays bending strength, self-damping and attenuation of ringing that are all an order of magnitude or more higher than that of conventionally manufactured cones."

In my opinion the word "bending" implies "non-pistonic" behavior, which I would have called "breakup". And in my opinion "self-damping" and "attenuation of ringing" imply WELL CONTROLLED breakup.

[I found this online definition of "cone breakup": "At low frequencies a cone moves as a whole. This is the 'pistonic' area of operation. At higher frequencies the cone starts to flex, leading to resonances. This is what is referred to as 'breakup'."]

So maybe this is what's going on: The new cone goes into extremely well-behaved breakup and flexes just the right amount to significantly widen the radiation pattern up into the crossover region, with any "nasty" breakup peaks being well above the driver's passband. And the marketing department choose to emphasize the strength and stiffness of the cone because admitting to "cone breakup" within the driver's passband would hurt sales.

And IF this is what's going on, then in my opinion the marketing department's selective wording IN NO WAY DIMINISHES the amazing accomplishment of Magico's engineering department.

I have ZERO affiliation with Magico... but mucho admiration for their engineering department.
Well done! I would say we can close this case :)

EDIT: and PS, what a superb driver that midrange appears to be!
 
OP
Duke

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
525
Likes
1,035
Location
Princeton, Texas
Thread Starter #20
Well done! I would say we can close this case :)

EDIT: and PS, what a superb driver that midrange appears to be!
Yes, and I think the elliptical crossover topology is also playing an important role, but probably more in the area of power handling. 2 kHz is a very challenging crossover point for a 1" dome in a speaker rated for use with thousand-watt amps.
 
Top Bottom