JBL M2 Clone

[MAB]

I recently measured a JBL D2430K compression driver and M2 lens. My goal is to build an M2 clone. The speakers are going to go into individual boxes allowing me to stack the woofer and D2 into an M2, or even include a midrange. I got a pair of B&C 5NDL38 for cheap, and if they don't work I just go with the M2 stack. Something like this, sealed rather than ported:

Of course I'm going to call it the M3.
Before I get ahead of myself, I am going to build some prototype boxes to see how this might work.

I spent some time tapping threads to mount the D2 driver to the lens and the lens to the frame.

The lens is huge, even compared to the 2216Nd woofer:

This frame for the lens is rough scraps. I'm planning on a hardwood frame with a removable back plate. The passive crossover is from the schematic of the M2 Crossover at rePhase, note the value is misprinted in JBL's M2 tech bulletin but correctly reported in the rePhase document.

I built the boxes.

I used leftover scrap wood for most of the build, but they turned out well. My plan is to build hardwood-trimmed ply boxes for the final rev. I'm now having second thoughts and might just paint these black...

So I now have a pile of modular elements that I can stack.

I went right to the full stack, like a hero:

I loaded some settings from the D2 measurements I had made, got the B&C midrange equalized flat. I EQ'ed the woofer, which wasn't too hard. It sounded horrible. The midrange is odd, with combing as I move my head around. One thought I have is the midrange is too close to the D2 horn, which overhangs it and I am getting diffraction. I've been too busy to diagnose the issues. I also thought it might be good to try to reproduce the M2 first, then maybe figure out how to integrate a midrange, or not...

I strapped the woofer and tweeter together and made a primitive turntable out of the shop-stool.

With 5 degree indexing:

I loaded the M2 PEQ and crossover filters into a MiniDSP Flex. I made horizontal and vertical measurements at 5, 10, and 15 degree increments across the speaker's axis.
Vertical mounting was sketchy, I am super-glad I didn't make the boxes more stout!

Also, I think I had large issues with the size of the speaker relative to the measurement distance. The mic dramatically changes distance relative to the drives as the speaker is rotated. I need to move outside and get the mic farther away I think.

Adding nearfield I and using VituixCAD I get the following:

What's puzzling is the magnitude of the DI mismatch between woofer and tweeter.

I'm wondering if the mic being so close to the speaker is amplifying the DI mismatch. Or maybe I need to do more work on the crossover. In any case, I think I am at a good stopping point. I'm listening to them now, and am really happy despite the DI measurement!

Here are the spinorama:

 

[Nisse10000]

Nice build! But yes, it looks like your mic is too close to the speaker. Aim for at least 3x the baffel width. Did you measure distortion too?

 

[Martin]

Where did you source the lens?

 

[MAB]

Nice build! But yes, it looks like your mic is too close to the speaker. Aim for at least 3x the baffel width. Did you measure distortion too?

Thanks!
Hi, yes I think I need to move outside, get a reasonable mic distance.
I did capture distortion, for the same reasons you mention I think the low frequency data isn't useable, but I get the same relative distortion that Erin observed at high frequencies. Specifically the same 2nd HD peak at 1.6kHz:

 

[MAB]

Where did you source the lens?

The Speaker Exchange. Here is a link to the M2 components they sell:

You searched for jbl m2 - Speaker Exchange

reconingspeakers.com

Only the ports are unavailable. They are apparently special, and some people have 3D printed them from JBL's white paper and measurements of the actual ports. For me sealed with PEQ bass extension instead makes sense I think...

 

[Kwesi]

Mmh, that mismatch looks strange. do you have measurements of the separate drivers/channels to see how they behave and add within the xover range?

 

[MAB]

Mmh, that mismatch looks strange. do you have measurements of the separate drivers/channels to see how they behave and add within the xover range?

I published the tweeter in another thread.
My mic was WAY too close to get a good measurement.
I am going to redo the measurements outdoors at 2 meters. I'm not too worried about the measurements, its a 50cm wide by 90 cm tall speaker measured in my basement at 1 baffle-width distance. I need to get better measurements. I will certainly publish them. And the individual driver responses. And my attempt to integrate a midrange.
Now that I have things up and running I am enjoying them in their current state.

 

[Robh3606]

They look great! I did a passive "M2" build using a 4" CD instead of the 2430. Love the way it came out. I don't know how you are going to integrate the mid-range without a big DI mismatch.

Rob

 

[MAB]

They look great! I did a passive "M2" build using a 4" CD instead of the 2430. Love the way it came out. I don't know how you are going to integrate the mid-range without a big DI mismatch.

Rob

Odd, I thought the directivity mismatch between the large woofer and the D2 at 800 Hz was one of the big compromises JBL made in this speaker. My thought was a 6-8" would be a better match, I got the 5" mids for so cheap I couldn't resist trying (especially since it took a while to get the woofers).

Here are the raw traces from the three drivers, with a dotted line snapped at the crossover frequency (~800 Hz).

The 5NDL38 is on the low side in the crossover region, 2216Nd a bit on the high side. I am still thinking that a 6-8" midrange would be a better solution than extending the 15" woofer to 800Hz where it gets ragged. Perhaps I am not thinking clearly!

 

[MAB]

A couple more thoughts on the performance of the 2216Nd, along with some measurements.
In the measurement of my woofer, you can see a dip in the response around 500Hz. This is attended by a spray of harmonic distortion components.

I observe the effects of this in my version of the M2 speaker.
It seems to show up in Erin's review of a JBL M2 production unit as well:

Looking at JBL's PEQ, they have no EQ directly on the woofer at ~500 Hz. They do EQ the input at 500Hz with a 1.1 dB peak, but Erin and my data shows a 2-3dB notch in the woofer's response that gets deeper off-axis. It seems the woofer turns into a flying potato-chip around 500Hz! I don't think the distortion is the problem. just a result... I think the audible issue is the dip in on- and off-axis response. Which is why I'm motivated to see if a midrange will address this, even if the ones I have aren't ideal.

 

[tmuikku]

Hi, I think you are right.
Quickly reasoning from the near field response and harmonic distortion plot it looks like that the harmonic distortion is kind of normal, it seems to stay the same through the midrange except at the 500-600Hz region, but since frequency response drops there and the HD is plotted as percentage it would show a peak because it is relative level. The distortion products are actually at higher frequency and not affected by the acoustic attenuation at the 500-600Hz region. It might be unintuitive at first and its due to the plot is made so the harmonic distortion is plotted at the root frequency and not at the frequency they were registered by the mic. I'm not sure which is more common way to plot distortion so that the harmonics are plotted at frequency they are measured, or mapped back to the root frequency. Both are regularly seen on the forums.

So I suspect the HD is generated in electric domain, from driver motor and other electronics, and the frequency response dip happens in acoustic domain.

Acoustic output could be lower for example due to different parts of the cone being in opposite phase, or it might be just that the radiating area gets smaller, part of the cone is not moving at all. Some cone resonance.

Or, instead of cone resonance it could be resonance inside the box reducing the acoustic output by destructive interference. This could be the case if you have quite closely dimensioned and damped box like the original, since the same phenomenon seems to be on the erin's measurements with original speaker.

You could straighten out the frequency response dip electronically with EQ, but you would boost also the distortion then and distortion plot would remain elevated. Since the distortion is quite low in general you could try if it sounds better compensated with EQ, it is likely that distortion is not as audible as ~6db dip in frequency response in this case. Perhaps listened at loud level it is, and engineer at JBL left it as is.

So, like you suspect distortion seems to be fine in general, while there is some destructive interference in acoustic domain, either from the driver or the box, this is the problem here. You could measure naked driver to test which one is it. Could be both. Could be something else. If you want to fix the issue prepare to swap the driver or make new box or modify the old one. If its something else, put your geek glasses on and get bottom of it

In general, big two way speaker like this you can pretty much expect issues like this even on a top shelf product like M2 and its not because its somehow bad system, its probably as good set of a compromises as it gets for the business they attend to. Bottom reason for this kind of issues is that sound wavelength changes so drastically between low and high frequency so only two ways is a stretch. If you want to get low lows from a woofer the midrange gets compromised. A loudspeaker should not compromise midrange but lows, because its always possible to add more low frequency sources to the system to restore while fixing midrange needs the whole loudspeaker replaced.

This would make it three way system, perhaps active, perhaps not their goal for this particular product but it would fix issues like this. But you as DIY person can make M3 as you have already plans for. I'm not sure if the small midrange works there though; use 15" mid like the original but in smaller (sealed) enclosure, perhaps another driver to cleanup the midrange. Leave the original box / driver for bass only duty if you wish.

 

[MAB]

Hi, I think you are right.
Quickly reasoning from the near field response and harmonic distortion plot it looks like that the harmonic distortion is kind of normal, it seems to stay the same through the midrange except at the 500-600Hz region, but since frequency response drops there and the HD is plotted as percentage it would show a peak because it is relative level. The distortion products are actually at higher frequency and not affected by the acoustic attenuation at the 500-600Hz region. It might be unintuitive at first and its due to the plot is made so the harmonic distortion is plotted at the root frequency and not at the frequency they were registered by the mic. I'm not sure which is more common way to plot distortion so that the harmonics are plotted at frequency they are measured, or mapped back to the root frequency. Both are regularly seen on the forums.

So I suspect the HD is generated in electric domain, from driver motor and other electronics, and the frequency response dip happens in acoustic domain.

Acoustic output could be lower for example due to different parts of the cone being in opposite phase, or it might be just that the radiating area gets smaller, part of the cone is not moving at all. Some cone resonance.

Or, instead of cone resonance it could be resonance inside the box reducing the acoustic output by destructive interference. This could be the case if you have quite closely dimensioned and damped box like the original, since the same phenomenon seems to be on the erin's measurements with original speaker.

You could straighten out the frequency response dip electronically with EQ, but you would boost also the distortion then and distortion plot would remain elevated. Since the distortion is quite low in general you could try if it sounds better compensated with EQ, it is likely that distortion is not as audible as ~6db dip in frequency response in this case. Perhaps listened at loud level it is, and engineer at JBL left it as is.

So, like you suspect distortion seems to be fine in general, while there is some destructive interference in acoustic domain, either from the driver or the box, this is the problem here. You could measure naked driver to test which one is it. Could be both. Could be something else. If you want to fix the issue prepare to swap the driver or make new box or modify the old one. If its something else, put your geek glasses on and get bottom of it

In general, big two way speaker like this you can pretty much expect issues like this even on a top shelf product like M2 and its not because its somehow bad system, its probably as good set of a compromises as it gets for the business they attend to. Bottom reason for this kind of issues is that sound wavelength changes so drastically between low and high frequency so only two ways is a stretch. If you want to get low lows from a woofer the midrange gets compromised. A loudspeaker should not compromise midrange but lows, because its always possible to add more low frequency sources to the system to restore while fixing midrange needs the whole loudspeaker replaced.

This would make it three way system, perhaps active, perhaps not their goal for this particular product but it would fix issues like this. But you as DIY person can make M3 as you have already plans for. I'm not sure if the small midrange works there though; use 15" mid like the original but in smaller (sealed) enclosure, perhaps another driver to cleanup the midrange. Leave the original box / driver for bass only duty if you wish.

I'm guessing it is the driver. My woofer measurements are in a sealed box of different dimensions than the production M2 that Erin measured, yet show what appears to be the same resonance. And it's hard for me to imagine a resonance in the box that excites so many harmonic modes, or the box's resonances creating such a dip in the system's acoustical output (especially the JBL box since my boxes are simple test-enclosures and JBL's M2 is a reasonably braced product). More like a typical cone breakup. While cone breakup can be mitigated, better to avoid if possible.

BTW, I love the idea of two of those 2216's in a tower:

 

[fluid]

Odd, I thought the directivity mismatch between the large woofer and the D2 at 800 Hz was one of the big compromises JBL made in this speaker. My thought was a 6-8" would be a better match

There is a good match of DI between the 15" woofer and the M2 waveguide at the crossover frequency, in Erin's measurements this is a DI of about 8dB. This is normal for devices that are around the same size, to control horizontal directivity to a similar frequency. The M2 waveguide is a little unusual in that it widens around 2K before narrowing back down.

With the smaller driver as a mid you could match the patterns better at around 1.5 to 1.8K. It is a bit odd to use a waveguide this big that high but it would let you smoothly increase the DI from low to high frequencies instead of there being a knee in the DI as happens with a single large woofer.

It is normal with surround edge resonances for there to be a dip in the frequency response together with an increase in distortion. Most 15" woofers have this in the 300 Hz range but it depends on the cone profile among other things. It is strange that the reduction in output happens across a range of angles.

 

[MAB]

There is a good match of DI between the 15" woofer and the M2 waveguide at the crossover frequency, in Erin's measurements this is a DI of about 8dB. This is normal for devices that are around the same size, to control horizontal directivity to a similar frequency. The M2 waveguide is a little unusual in that it widens around 2K before narrowing back down.

With the smaller driver as a mid you could match the patterns better at around 1.5 to 1.8K. It is a bit odd to use a waveguide this big that high but it would let you smoothly increase the DI from low to high frequencies instead of there being a knee in the DI as happens with a single large woofer.

It is normal with surround edge resonances for there to be a dip in the frequency response together with an increase in distortion. Most 15" woofers have this in the 300 Hz range but it depends on the cone profile among other things. It is strange that the reduction in output happens across a range of angles.

Thanks Fluid, you're always helpful. And make sense.
Yes, my measurements say 8dB at 800Hz too, with both drivers matched at that point. Agreed the 5" is way too small. But sounds like I need to temper my enthusiasm that an 8" midrange will magically close the gap to the D2 at 800Hz.

I agree the surround edges edges are the source of breakup. But I had thought that this was often coupled with odd off-axis behavior. I guess I am confused why you would expect that the cone breakup would have this directivity? I would have thought that the cone breakup would reduce output not just on-axis.

 

[fluid]

Thanks Fluid, you're always helpful. And make sense.

You are welcome

But sounds like I need to temper my enthusiasm that an 8" midrange will magically close the gap to the D2 at 800Hz.

The 15" driver is the right size to match the DI at 800Hz, the 8" driver will be different and like the 5" driver could be crossed over higher for a better match but most PA drivers struggle to get very high in frequency before they break up.

I agree the surround edges edges are the source of breakup. But I had thought that this was often coupled with odd off-axis behavior. I guess I am confused why you would expect that the cone breakup would have this directivity? I would have thought that the cone breakup would reduce output not just on-axis.

Usually all the axes converge at the same point centred on the dip, as there is out of phase motion from the surround causing a nulling effect. Satori paper woofers have quite a pronounced dip

https://hificompass.com/en/speakers/measurements/satori/satori-mr16p-8

https://manualzz.com/doc/37151866/eds-9990013-jbl-2216nd

JBL's own measurements show a similar thing but it's harder to see from the scale and the different diffraction response.

Cone breakup isn't till much higher with worst breakup at about 3K, there is a rapid fall off from there in output so the higher frequency breakup problems don't really show in the distortion plot.

 

[hex168]

The linked discussion is about a different woofer, but it contains something you can try to see if it affects the postulated surround resonance:

https://www.avsforum.com/threads/diysg-htm-12v2-review.3204391/page-3

See post #51 through #64 or so.

 

[MAB]

You are welcome

The 15" driver is the right size to match the DI at 800Hz, the 8" driver will be different and like the 5" driver could be crossed over higher for a better match but most PA drivers struggle to get very high in frequency before they break up.

Usually all the axes converge at the same point centred on the dip, as there is out of phase motion from the surround causing a nulling effect. Satori paper woofers have quite a pronounced dip

I see. I am mixing up concepts, confusing cone breakup modes with resonance in the surround. Like you said the Satori MR16P cleanly demostrates, with a blip in the impedance too...

Looking at datasheets with my newfound understanding, is the resonance frequency is related to the surround's circumference?
Here is an SB Acoustics 12" with the dip at 600Hz:

http://sbacoustics.com/wp-content/uploads/2020/02/12in-SB34NRXL75-8.pdf

JBL's own measurements show a similar thing but it's harder to see from the scale and the different diffraction response.

Cone breakup isn't till much higher with worst breakup at about 3K, there is a rapid fall off from there in output so the higher frequency breakup problems don't really show in the distortion plot.

Yeah, I got my head on straight now (mostly). Thanks for the education. You may have saved me from futile shopping for expensive 8" midranges.

 

[fluid]

Looking at datasheets with my newfound understanding, is the resonance frequency is related to the surround's circumference?

Yes, there are other things can can modify it but the size of the driver is the biggest determinant of the frequency where it happens, bigger driver lower frequency.

 

[Mr. Widget]

I hope you pull this off... I also hope you go the hardwood route or at least veneer the boxes. I positively love this aesthetic!

 

[MAB]

Yes, there are other things can can modify it but the size of the driver is the biggest determinant of the frequency where it happens, bigger driver lower frequency.

OK, I understand. I've seen that dip over and over and must admit I was unaware of what it was. Very informative.