JBL M2 Audiolense Digital Crossovers w/Subs

[dallasjustice]

I first became seriously interested in purchasing the JBL M2 when I was with @amirm at CEDIA 2015. I heard the Harman demo room and was shocked at the sound quality I heard. The Harman demo at CEDIA is like waiting in line for the Pirates of the Caribbean ride at Disney World with my kids. Its something you have to do if you are there.

I've always been a bit of a subwoofer fanatic. My room is approximately 14'6" wide, 24' long and 9'3" tall. It is an almost perfect rectangle except the small entry way at the rear. When I built the room on to my home about 6 years ago, I had no idea what I was undertaking. The last 6 years has been a learning adventure. I started off trying to throw money at my system. I started buying expensive speakers, amps, DACs and even cables. At this stage of my journey, I was convinced not by measurement and science but by the dollar. This is how it goes: dealer sells you an expensive pair of speakers. You like them, but then you are convinced that if you spent X dollars on speakers, you also MUST spend X dollars on power cables and speaker cables and DACs and Amps and on and on.

Fortunately, I could still hear that the bass quality wasn't very good in my room. So I started searching for new information. I had previously learned about Bernt Rønningsbakk about 10 years ago online reading AVSforum. He had a new product called Audiolense. So I bought a measurement microphone, Lynx card and used the software with Bernt's help. I really didn't understand how it worked or what I was doing. I quickly gave up on it. Bernt was a gentleman and permitted me to sell my license to someone else who could better use it. At that time, I learned that crossovers could be digital instead of analog. I asked Bernt about it at the time but never really pursued it because I thought it was too complicated. I later met Nyal Mellor who began the deprogramming process on me. I thank him for that. Nyal re-introduced me to digital room correction (a misnomer I will refer to it herein as "DSP"). I bought a DEQX and used a subwoofer pair with a digital crossover. I really like the improvement in bass quality. I had previously spent a lot of time money on room treatments and system setup to optimize bass in my room. Even with the best technology is bass mode control (RPG modex plates), I could only get a little better results.

Later on, I met Dr. Uli Brueggeman who is the designer of software called Acourate. Uli is a meticulous German engineer and audiophile who possesses didactic skills and patience unmatched in my experience. Uli helped me setup active crossovers using both 2 subs and then 4 subs. At that time, I learned more about subwoofer integration after reading an AES paper from Todd Welti. I was also introduced to more subwoofer research in an AES study conducted by Fazenda. Although Welti and Fazenda approached the problem of room mode control in different ways, they both seemed to agree on mono subwoofer integration. IOW, each subwoofer gets a summed R/L signal. That approach worked well for me. I did notice tho, that I could never get "perfect" time domain integration with mono subs. However, mono subs always gave me better frequency response than 2 stereo subs. So there was a trade-off.

After exploring all sorts of mono sub arrays. I found the 2 mono "source-sink" array studied by Fazenda to be superior to even the 4 mono midwall sub array that Welti advocated for rectangular rooms. This "source sink" had the advantage of being able to eliminate otherwise untreatable length modes (1,0,0 and 2,0,0). The disadvantage to this array is that it can never be truly time aligned. (the rear sub must be in opposite electrical polarity and delayed to reach LP a little bit after the front sub).

I recently found a pair of JBL M2 from the family of a deceased East Texas dealer. I got a great price and I thought I'd try my hand at setting the M2 up using digital crossovers generated by Audiolense. @amirm and I previously discussed the Harman system for subwoofer integration with the M2. Although the Harman method is very advanced, I never thought I would be happy with it due to the controlled roll-out of this method. Basically, its dealer-only setup. I also reject any closed box DSP application. Harman only offers off-the shelf DSP inside either their SDEC or their Crown Amps. I know that any digital crossover or EQ in a box solution will be a digital compromise. (limited number of taps, no linear phase crossovers) I also learned that Harman's out of the box DSP sadly does NOT time align the D2 tweeter with the midwoofer. For me, that's too many compromises. I know I can implement a better linear phase crossover and perfectly time align the drivers using Audiolense. (the same could be done in Acourate but I chose to use Audiolense)

A note about time domain. I've read Sound Reproduction. I'm aware of the paucity of science that show any advantage to time aligned drivers compared to typical minimum phase implementation. I accept the lack of research in this area. BUT, I can speak from my own experience. I've time aligned many different speakers in my room and have always heard a big improvement in 3D image and image accuracy. I think some audiophiles have morphed Toole comments about the lack of research in this area into proof that driver time alignment cannot be perceived. That's simply not the case. There's just not enough research in this area and more needs to be done, IMO.

I'm not going into detail about the M2 vs. the 4367. I've owned both in the same room. Anyone could look at the Harman spin data and reach the conclusion that the M2 measures a little better on and off axis. This is why I chose the M2 for this project. The M2 may be the finest measuring loudspeaker available at this time. So I wanted to hear it setup in a way I thought I could achieve ultimate playback in my room. Yes, I did that.

I want to thank @mitchco for all of his help. For those of you who don't know Mitch, you should know him. Mitch is the leading DSP evangelist. He's written numerous articles and has now written a book on how to implement DSP. His book is called Accurate Sound Reproduction. Mitch's guidance and insights were critical to me getting the M2 setup. Thank you Mitch!

I'm going to post the most important REW measurements of the M2 setup in this post. I will follow with additional info and pics regarding the new 4 stereo sub array I'm using with the M2.

 

[dallasjustice]

The JBL M2 is 2 way speaker system without a passive crosssover. There is a protection circuit built into the speaker which is designed to protect the D2 compression driver from DC and excessively loud low freqeuncy noise. Harman recommends an 800hz crossover for the JBL M2. I use software called Audiolense to generate the crossovers. In addition to the M2 crossover, I setup a 4 subwoofer array. Each subwoofer either gets a R or L channel. There are no summed R/L subs in this array. The crossovers are custom digital crossovers created by Bernt in Audiolense. They resemble NT crossovers. All crossovers are symmetrical but they are not all the same width. You can see the M2 is crossover over at 800hz using a 2 octave crossover. The high frequency stereo sub pair are crossed over at 175hz also using a 2 octave crossover width. Finally, the low frequency stereo sub pair are crossed over at different slightly different frequencies using a 1 octave crossover. I can go into greater detail about why I chose the crossovers I did later on. Here they are:

And here is my speaker setup and crossover settings in Audiolense:

This is just a shot of Audiolense frequency response panel showing all driver measurements, filters applied, target curve and predicted result.

 

[dallasjustice]

A keen observer would have noticed a rather odd subwoofer crossover setup upon examining the above depicted speaker routing/crossover setup in audiolense. There are several rules broken here:

1. Never put a subwoofer behind the seated position. Here is a front view of my room and you can see 2 JL audio F112vs subs placed up against the wall behind the seated position.

2. Why are you sitting on a really thick rug? It does a great job absorbing the floor bounce. It totally eliminated the bounce. I use a thick rug with a few felt rug pads underneath.
3. You have four subs and 2 crossovers for the subs. What's going here? All subs are stereo. There are a high frequency pair of stereo subs and a low frequency pair of stereo sub. The high crossover is at 175hz 2 octave slope and the low frequency ccrossover is 50hz on the left side and 47hz on the right side with a 1 octave slope. Here is the picture of where the high frequency subs are located.

This is a flanking subwoofer setup. I've learned over the years the difference between EQing minimum phase room modes and non-minimum phase room problems. The non-minimum phase problems I'm referring to are called Allison Effect. Some folks call them "nulls." Allison effect nulls cannot be EQ'd. It's that simply. They never change unless you move a speaker, the seated position or (sometimes) treat a wall. In the case of sidewall allison effect, there's no good way to eliminate them. In my room I have modex plates setup on my sidewalls. They cannot touch the allison effect. The only way to reduce the sidewall allison effect using room treatements would be to use a shitload of thick fiberglass panels. This is a terrible idea because that would kill the high frequency spectral balance of the room. You'll notice from the RT60 above, my room is at 400ms in the critical midrange listening region. That's on the upper end of good. I'd say it a lively room, but not too lively. So what have I learned? If you have reasonable room treatments, there's really no problem EQing minimum phase room modes. However, room treatments can only help so much with non-minimum phase problems and they can never be EQ'd.

The human ear begins to locate the origin of sound wave at around 80-90hz. So the ear can begin to localize the direction of a subwoofer at around 80hz. That's why so many home theater receivers have 80hz crossovers for the subs. I've experimented with using a mono flanking sub just behind my left speaker the past. I placed the sub there because there's an antimode position for the 2,0,0 room mode (47hz). It eliminates that mode using only one sub and no DSP ( a neat trick!). But I noticed something else when I placed my sub in that position, it also helped to eliminate a ceiling allison effect reflection due to its location when I chose a 90hz (very high) crossoover for that mono sub. It eventually dawned on me that I could position 2 flanking subs against the sidewalls to help eliminate the sidewall allison effect. I would have to cross them over at a much higher frequency (175hz) but I could place them close to my R/L and run them in stereo.

I did a little research to see if anyone else had noticed the same advantage to using stereo flanking subs up against the sidewalls in a relatively small room. I found Wayne Parham's writings on the subject. He is a great horn speaker manufactuer. I've heard his Pi speakers before at the Lone Star audiofest here in Dallas. They are really cool speakers. Wayne explains why the flanking sub setup is so effective here:

https://audioroundtable.com/forum/index.php?t=msg&th=16726

I like Wayne's idea here. I also think he has a valid criticism of the Todd Welti subwoofer approach. All subwoofer arrays are designed to only address bass problems less than 100hz. I think that a big mistake. IMO, the bass heard 100hz-200hz is even more critical to music enjoyment than the bass less than 100hz.


So can I hear an image shift using flanking subs and a 175hz crossover? I tried my best. I listened to jazz upright bass ascedning and desceding notes and could never hear any image shift.

What are those little subs?
Rythmik F8. These are dual 8' driver subs. They are specifically designed to play at a higher frequency than almost all other subs. They are totally flat to beyond 200hz. That's remarkable since almost no sub is flat beyond 100hz. Rythmik subs have incredible output. Thes F8 subs have an 11' baffle; they look tiny. But they produce incredibly powerful and tight bass. I love them.

So what about the rear wall subs? You can hear them behind you at the LP, right? Why are they there?

Uh no, you can't hear them. The right side is crossed over at 47hz and the left at 50hz using very steep 1 octave crossovers.

The stereo rearwall subs totally eliminate the backwall allison effect reflection center at 35hz. The difference in bass quality it tremendous and there are no downsides to this configuration.

 

[dallasjustice]

This is the JBL M2 using the same crossover at 800hz. The green is the M2 using a flat target curve and no subwoofers. Both measurements are the same speaker in the same room. This time 1/12 per octave smoothing.

Here are some comparisons between the measurements in my room of the JBL 4367 without subwoofers versus the JBL m2 using a above described subwoofer array.

 

[fitero]

This is exactly the thread that I was waiting for! Thank you for sharing.

I just brought home a pair of M2's last night and hooked them to a JBL SDA-8300 amplifier and played them for a little bit. Even in that brief sample it was abundantly clear to me that I had made the correct choice selling the 4367's and later the Salon 2's and getting the M2's.

I'm very curious to learn how you implement the crossovers etc.

 

[dallasjustice]

This is exactly the thread that I was waiting for! Thank you for sharing.

I just brought home a pair of M2's last night and hooked them to a JBL SDA-8300 amplifier and played them for a little bit. Even in that brief sample it was abundantly clear to me that I had made the correct choice selling the 4367's and later the Salon 2's and getting the M2's.

I'm very curious to learn how you implement the crossovers etc.

I did my best to give some details but I didn't want to get too bogged down. Feel free to ask questions.
Michael.

 

[dallasjustice]

Btw, I wish I had some screenshots of Roon using my 4 way stereo setup. But, again, Roon's convolver doesn't work. I have another ticket open with them to fix it. They previously fixed the 6 channel, 3 way setup I had. But whatever they did, didn't do anything for the 8 channel setup. It seems like Roon convolution is still a work in progress, I hope.

 

[RayDunzl]

If you would...

Post a zipped .mdat from REW please.

left, right, both sweeps if you have that. (taken at the listening position)

.zip files are accepted here under 2mb

 

[RayDunzl]

PS:

 

[dallasjustice]

If you would...

Post a zipped .mdat from REW please.

left, right, both sweeps if you have that. (taken at the listening position)

.zip files are accepted here under 2mb

https://www.dropbox.com/s/2akb0sbfnx58871/REW M2 4 subs.zip?dl=0

 

[dallasjustice]

I got the Roon Convolver working. All is good.

 

[Brad]

I'm confused about your flanking sub setup. The link you provided describes the flanking subs as overlapping the mid-bass of the mains, but your crossovers separate the flanking subs and the mains.
If that's the case, how are your flanking subs providing such a smooth response?

Also, would you be able to post a response with L+R playing?

 

[dallasjustice]

I'm confused about your flanking sub setup. The link you provided describes the flanking subs as overlapping the mid-bass of the mains, but your crossovers separate the flanking subs and the mains.
If that's the case, how are your flanking subs providing such a smooth response?

Also, would you be able to post a response with L+R playing?

Yes, I’m using a crossover and the link I posted is Wayne’s description of how he uses a similar flanking sub array to mitigate early nulls in smaller rooms. I didn’t post Wayne’s comments as a model of how I set up my subs. I was simply relieved someone else had noticed the same problem I had. I don’t think it’s wise to use subs without crossovers. Certainly it’s possible to do so, but you’ll almost always get better results using a crossover.

I’m not sure what kind of measurement you want. Are you asking for one of @RayDunzl summed R/L measurements?

If so, I’m sorry that I don’t have any measurements like that. I can’t think of a good reason to do one. I know Ray has explained why he does them in his room. If you look at the phase response below 200hz, it’s flat. So there would never be any destructive interference from mono bass program material.

 

[RayDunzl]

The link you provided describes the flanking subs as overlapping the mid-bass of the mains

I don't see that in the crossover setup

Low Sub - Low pass - 0->50Hz
Mid sub - bandpass - 50->180Hz
M2 woofer - 180->800Hz
Horn - 800-> whatever the top end is Hz

Looking at this image: https://www.audiosciencereview.com/forum/index.php?attachments/crossovers-png.10716/

If that's the case, how are your flanking subs providing such a smooth response?

I have cheap subs flanking the stereo mains pair. The subs are set for cutoff around 100Hz. They are ported at 23Hz. So, they overlap the mains, which actually go lower than the subs. Here's a measurement of my L/R tonight (gray) at about 85dB at the listening position, overlaid with the M2 setup. (no tricks - using AcourateDRC and a filter generated a few months ago, which probably should be updated). No crossovers in my filter, just overall "room correction".

 

[RayDunzl]

I don’t think it’s wise to use subs without crossovers.

You're a little more enthusiastic in your chase than I am...

My goal adding subs turned out to be LF distortion reduction. Three "woofers" can be run -9B vs a single woofer (per side).

 

[Brad]

Yes, I wouldn't expect you to use subs without crossovers, but extend the bass driver of the M2 down to 80Hz or so.
I assumed you tried this, but I am curious as to why it didn't give superior results.

I am interested in the L+R measurements, as based on Ray's posts I performed my own. On some of my acourate generated filters I see deep nulls above 1kHz (caused by the tweeter). I would like to understand if that is a typical response due to interference, or not.

 

[RayDunzl]

I am interested in the L+R measurements, as based on Ray's posts I performed my own. On some of my acourate generated filters I see deep nulls above 1kHz (caused by the tweeter). I would like to understand if that is a typical response due to interference, or not.

My little JBL's show more "hash" in the high frequencies, assumedly reflections cancelling the direct wave, than the MartinLogans, which have very little interaction with the sidewalls and ceiling. I have nothing absorptive on the wall or ceiling, CD racks on both sides, though.

Here's DallasJustice M2 vs the ML here, no smoothing, 5kHz to 10kHz. (Different rooms)

Here's my ML vs my LSR 308 (same mic setup, on top of couch)

That's my best guess.

I don't fault the driver. Red - JBL LSR308 nearfield (1m) , probably on a stool, and out in the room (farther from the walls), vs MartinLogan at listening position (10 feet, same trace as previously)

Amir will say you can't hear it. Notches too narrow.

I think it messes with the imaging, makes it a little "phasey" sounding to me.

Swapping from In-Phase to Out-of Phase is much less a distinct difference when Audio Buddy asked "Are they in phase" and I had to go, I guess so, let's see. (swapped phase a few times, asked Buddy if he could tell if they were in or out of phase, he couldn't).

It only comes up as a problem (to me) when dead center paying attention to imaging. Likely fixed with treatments, or simply accepted (by others) because that's what you're used to hearing.

Martin Logans came in last place (like, nobody liked them) in a 4-speaker shootout at Harman. I don't let Amir let me forget that. (Here's another chance for me to remind him to remind me)

 

[RayDunzl]

@dallasjustice

You smoked me on the equivalent SPL distortion.

(might add a picture later, too disgusted right now)

 

[dallasjustice]

I see deep nulls above 1kHz (caused by the tweeter).

That’s normal. You can’t hear those. That’s why Acourate or Audiolense will use a psychoacoustic filter before inversion.

 

[dallasjustice]

Yes, I wouldn't expect you to use subs without crossovers, but extend the bass driver of the M2 down to 80Hz or so.
I assumed you tried this, but I am curious as to why it didn't give superior results.

Sorry, I didn’t understand your question and I probably didn’t explain the flanking sub idea in sufficient detail.

Any “small” room will have Allison effect “nulls” from the boundaries adjacent to the speaker setup at listening position. The primary problem areas are:
1. Sidewall
2. Floor bounce
3. Backwall
4. Frontwall (only if speakers aren’t pulled out far enough)
5. Ceiling.

In my room, the backwall boundary interference is eliminated using rear wall subs.

I don’t have front wall interference because my speakers are out far away from front wall.

My floor bounce is mitigated with the thick carpet/felt pad.

There are a couple of other odd boundary interferences I dealt with using a couple of RPG BAD panels.

That leaves the ceiling and sidewalls. The ceiling is less of an issue because I have an overhead bass trap covered with T-fusers.

So the flanking subs are there mostly to eliminate the sidewall boundary interference. In my room that would be around 160hz on both sides. So I crossover my M2 over at 175 with the flanking subs, which is high enough to bypass that frequency. The flanking subs are placed against the boundary so there’s no additional sidewall boundary interference.

There are still other interfence with the flanking subs. For example, the backwall. So I had to be very careful about where to set the crossover with the rear subs so I can avoid that interfence.

Setting up 4 stereo subs in a cascaded way like this can be the tricky. Subwoofery is an iterative process. The nice thing about Audiolense is that I can do my speaker setup, take my sweeps and then experiment with different crossover combinations. The simulations in AL are spot on. I’ve always confirmed with REW.