My hybrid MTM project

[pangea]

Previous postings from this thread can be found in the thread "K 145".

The last posting from "K 145", is duplicated here though.
-----------------------------------------

First of all I would like to wish everyone a HAPPY NEW YEAR, although the new year celebrations have been very somber here, since estimates say that more than 150 000 people have died in Asia, of which more than 3000 were Swedes.
None the less I have been occupied trying to learn some about my project.
I'm anxiously awaiting the return of Earl K and hoping the discussion will continue soon.

-Earl K
In case you're back already, there have emerged a few questions during the holidays, which I would be grateful if you would like to address first.

You wrote:

While Operating in 43XX mode :
- This concept of some form of overlap within the 4-way crossovers' slopes ( between bass and mids ) is something I believe you can do with your Behringer. This will help provide some sonic "glue" for the split between the 2215 and the 2123 . You'll need to experiment some to find the best combination .
For example ; the 2215h lowpass could be set for around 340 hz, while the 2123H hipass might be set around for 120hz ( you will need to port/tune) the 2123 to add a bit of midbass extension. I'd mix and match slopes , like 12db for the 2215 lowpass and 6db for the 2123 hipass .

I’ve been thinking, perhaps I should try to make the boxes sealed. I’ve read that sealed boxes have a much better transient response, so I was thinking of using a large (D, 30cm*L, 15cm) cardboard duct, for the 2123’s. For the bass-bin I was thinking of making it possible to go both sealed (80L) and vented (163L), by plugging the holes in the internal bracing.

Do you think a bit of shelving EQ, on both woofers would be able to extend the frequency range somewhat, as well as making it possible to lower the output, so that they won’t be as pronounced in the 300 - 600Hz region, like you said and at the same time keep the sonic “glue” intact?
Any comments on this?

I have also read somewhere that a first order x-over on the bass and mid-bass (giving a 90 degree phase shift), should be complemented by an equally first order x-over, or a third order (90 or 270 degree phase shift) x-over on the midrange, giving a total of 180 or 360 degree, or a full turn, phase alignment, which would give a perfect summation.
In my case, using the 2441, I think a third order x-over would be better/safer with a x-over at 1000Hz. Wouldn't you agree?
Then the task of time-aligning the drivers, would be much easier for the Behringer UltraDrive, at least I think.

While you were away for the holiday’s, I’ve also been busy trying to figure out the best/ideal x-over point between the 2441 and the 2123 and all things considered I think that 1000Hz/34cm distance between driver-center, seems best in my view.

I’ve also come across a program called XDir, which is said to calculate lobing-effects, though I’m not sure how it works, or how to find the best values. Perhaps you’re familiar with the program? http://www.tolvan.com/xdir/ .

BR
Roland
This is the latest design attempt of my MTM project and there are also a few pictures of the first pieces and a few of my tools.







EDIT: Here are a few of my tools, as well as one of my "co-workers", as always taking a brake!

[Earl K]

Hi Roland !

Just checking in, I'm leaving for Montreal shortly - so, I'll have to answer your questions after I return .

Briefly; You should take the time to discover for yourself whether or not "you" think that sealed boxes offer better transient response versus ported boxes. "Theory" which hasn't been personally experienced, is at best just that, it's only "Theory", and at worst this type of info can quickly become "dogma". So, that means building something that has ports that can be stuffed so that you can discover your own preference .

Regarding, Crossover Points and Slopes ;
I think that you are are putting the "cart before the horse" by trying to predict in advance what will work best . Since you own a very variable DSP based crossover - you can discover this later / and then tell us what worked best for your project .

Lobing Prediction Programs ;
The unrestrained use of this sort of program , coupled with any dogmatic belief in its' "theoretical predictions", will "severely challenge" your current belief that your asymmetrical MTM-hybrid is worth pursuing as a project .

ie ; "Invest in & Believe" - at your own peril

( Personally - I think "floor-bounce cancellations", as measured within your "listening-window" - warrant as much or more scruntiny )

Happy New Year ! Earl K

[pangea]

-lfh

Quote:
Originally Posted by pangea
About "Xdir", I'm sorry I used the word "supposed". It wasn't my intention to be critical or anything of the program, it only meant that I didn't know how to use it correctly.
So perhaps I should have said, "it is said to"..., instead.
I'm sure the program is very good.



No worries, I just wanted to say that it most likely does what it's supposed to do (as opposed to some crappy freeware out there...). And yes, it should still be useful for your purposes. E.g. first look at the frequency range where the twin basses operate but the tweeter is silent (e.g. do one plot per octave) and then study what happens around the crossover frequency where all drivers contribute. Vary one parameter at a time (spacing/level diff/phase diff), and you'll get a good grasp of how the array works, and how the filter properties come into play. It should also be interesting to compare a conventional design and an MTM around the crossover frequeny.

It seems that when the distance between each driver is 34cm and at the same time, the "phase lag" is zero degrees, there will always be a significant lobe heading for the "sweet spot".

Could this be interpreted in such a way, that I most likely won't experience any negative surprises?

I'm still not sure how to read the various graphs.

-Earl K

I know I'm trying to put the cart before the horse! You're absolutely right, but I'm a cautious guy, who doesn't like to make any errors. I've always been this way, trying to find the weaknesses and potential problems in advance and subsequently trying to avoid them.

You said earlier the 2123 will perform best in a sealed box, so I figured a 7L (10L minus 3L driver displacment) "tube" box would be best, as well as easiest to make. As for the bass bin I'm planning of making it possible to go either way (80L sealed or 163L vented).

Regarding slopes, I'm counting on the Behringers being able to find the phase problems and help me to set up all the other parameters.

About "floor-bounce cancellations", I'm hoping I've managed to minimize those problems, by stuffing my listening room with lot's of stuffed furniture and carpets and last but not least having treated the room with quite a few absorbers and traps.

And there is not a chance; I'll give up this project!!!

BR
Roland

[lfh]

It seems that when the distance between each driver is 34cm and at the same time, the "phase lag" is zero degrees, there will always be a significant lobe heading for the "sweet spot".

Could this be interpreted in such a way, that I most likely won't experience any negative surprises?

I'm still not sure how to read the various graphs.

Yep, a main lobe in the horizontal plane is the expected behaviour, and the pattern should be symmetrical above and below the "x-axis". No unpleasant surprises. The graph basically shows the variation in readings you would get when moving a mic along a half circle (at large distance from the array) in the vertical plane around the array center (and on axis in the other dimension) in an anechoic environment. The frequency response at any desired angle (e.g. 30 deg above/below horizontal plane) can be derived from a set of such polar plots calculated for a sequence of frequencies. (Disclaimer: Again, I can't run it conveniently on my Macs, so I don't know whether he does some normalization, but in principle this holds true.) Note that such a FR is due solely to the array geometry (and possibly source phase and level diffs), and the corresponding interference patterns. A more realistic simulation would (at least) take the driver directivity patterns into account (possibly modeled with the well known directivity patterns of simple propotypes such as a piston in a baffle), but the model used in the program still gives a good demonstration of the fundamental properties of a certain array.

EDIT: In this particular case - using a 10" and a 15" in an MTM setup - I'd want to use the piston model (or better yet; measured magnitude and phase data) for the simulation. Such a hack is on my to-do list, but ETA is like Q2... Perhaps you could convince SG to add this (piston model), along with a few text book filters, and means to enter driver acoustical centers?

EDIT 2: Then (after reading Earls post again), what would you do with the more accurate plots? New conclusion: Point sources are just fine. Time better spent would most likely be to build test boxes and experiment! "Probieren geht ueber studieren."

[pangea]

WOW, lot's of response, great!

Yep, a main lobe in the horizontal plane is the expected behaviour, and the pattern should be symmetrical above and below the "x-axis". No unpleasant surprises. The graph basically shows the variation in readings you would get when moving a mic along a half circle (at large distance from the array) in the vertical plane around the array center (and on axis in the other dimension) in an anechoic environment. The frequency response at any desired angle (e.g. 30 deg above/below horizontal plane) can be derived from a set of such polar plots calculated for a sequence of frequencies.

It wasn't until now, that I realized this. I thought a nice even curve looking like a half circle was the target at all times.:o

Note that such a FR is due solely to the array geometry (and possibly source phase and level diffs), and the corresponding interference patterns. A more realistic simulation would (at least) take the driver directivity patterns into account (possibly modeled with the well known directivity patterns of simple propotypes such as a piston in a baffle), but the model used in the program still gives a good demonstration of the fundamental properties of a certain array.

Isn't it possible that the Behringer UltraDrive will take care of this, or isn't it the same thing as time aligning the drivers?

EDIT: In this particular case - using a 10" and a 15" in an MTM setup - I'd want to use the piston model (or better yet; measured magnitude and phase data) for the simulation. Such a hack is on my to-do list, but ETA is like Q2... Perhaps you could convince SG to add this (piston model), along with a few text book filters, and means to enter driver acoustical centers?

I'm not sure I understand these abbreviations and also, wouldn't these additions to the program complicate it quite a bit?

EDIT 2: Then (after reading Earls post again), what would you do with the more accurate plots? New conclusion: Point sources are just fine. Time better spent would most likely be to build test boxes and experiment! "Probieren geht ueber studieren."

"Aber natuerlich". Though being able to predict or knowing even, makes it possible to avoid spending a lot of time and money on building boxes, when the odds are stacked heavily against any solution.

Well, it depends. Sealed boxes behave like 2:nd order high pass filters whereas ported boxes behave like 4:th order ones. Thus, a sealed box has the theoretical possibility to perform better in this respect than a ported one. However, performance depends on the "filter" tuning/alignment. A closed box with a high Q-value will "ring"/"boom", and a ported box with a shallow roll off will have a "tight" character. That said, a 2:nd order HP characteristic with a Q around 0.6 (or even lower, when factoring in room gain at lower frequencies) is pretty much as good as it gets, but a ported box tuned with transient performance in mind yields very good results as well.

I have already been using JBL's "SpeakerShop Enclosure Module 1.0" and stopped at a net volume of 7L sealed box (10L minus 3L driver displacement) and a Q of 0,591, giving a Fc=166,9 and F3=205,7 and no filling.

Often the latter is the better overall compromise, when weighing in other aspects such as the desired cutoff frequency, the acoustical output capability etc. Yet another thing to keep in mind is that the "best" box type to a large extent depends on the driver at hand. Typical JBL drivers often work best in ported enclosures (if you still want some LF, that is).

I don't think I've ever seen the 2123 in a vented box and I was hoping I'll be able to EQ the 2123 to some extent, giving it a wider FR and at the same time lowering the output somewhat.

You're not doing a middle-of-the-road design here... Do you have some references of similar "hybrid MTMs" (IIRC Earl runs such stuff - I must read the entire K 145 thread) and have you looked into D'Appolitos designs - noteably the crossover networks? How will you tackle the low end where the 2123s runs out of steam but the 2215s keep pumping? I think some careful thinking is needed here when designing the two boxes. (Sure, you're "cheating" with your trick EQ and all, but some master plan on how to design the disparate bass boxes for the desired combined LF response (10"+15"+EQ+room) is called for IMO.) Finally; why is the tweeter not above the mid horn?

Mainly I'm acting on a very strong hunch and then I wanted to use the things at hand! I've been reading some D'Appolito papers and I've seen a few asymmetric solutions. So when Earl K suggested that I should build just that, I knew I had to.
About the overlapping, I'm hoping it will help creating a sonic "glue", like Earl K is suggesting. To me it seems very reasonable, given the need to look at the combined sound level and a lot of fiddling with x-over points, slopes and EQ'ing.
Hmmmm, "master plan", I don't think it ever came to that...:o

The tweeter has been moved around a few times. at one time there wasn't enough room between the 2441 and the 2123. Then again Earl suggested a placement, which resembled the JBL 43XX design and therefore should be quite a safe position.
I take it you would like to place it above? If so, could you please elaborate on why? Hopefully it would help me understand the principle a little better.

Bullock on boxes is a great read on this subject.

He says a properly tuned low Q woofer will sound every bit as good as a seal design. He then goes on to discuss the finer points of reflex tuning, accounting for Rx, QL and actually T/L measurements.

I know myself you can alter the quality of the bass in sublte but audible degrees by optimising the woofer for a given enclosure Vb, tuning frequency Fb and Rx the total Dcr behind the speaker.

Ian

Thanks Ian, where can I find this read? Is it to be found on this forum?

BR
Roland

[lfh]

It wasn't until now, that I realized this. I thought a nice even curve looking like a half circle was the target at all times.:o

Isn't it possible that the Behringer UltraDrive will take care of this, or isn't it the same thing as time aligning the drivers?

(Just some further elaboration.) "No surprise" since the pattern (basic properties) really can be predicted without any SW once one gets used to think about interference. The target is not necessarily a half circle. Rather one strives to control in what directions to radiate - and where not to.

As to the FR part, I was a bit unclear: "Such" refered to off axis responses derived from the polars. You should be able to get a flat on axis response, but there will always be ripple due to the twin mids/basses when you move off axis. BTW, see the Project May blog for interesting info about that.

I'm not sure I understand these abbreviations and also, wouldn't these additions to the program complicate it quite a bit?

Sorry about the technobabble. IMO it would not be so complicated (technically rather straightforward), but amount of work depends on the programming language and libraries used (read: how much "nuts & bolts" you need to implement yourself). I'll post here once I'm done (using a language that lets me be really lazy ).

I have already been using JBL's "SpeakerShop Enclosure Module 1.0"
/.../
I don't think I've ever seen the 2123 in a vented box

What's this SpeakerShop thing and what happens if you select vented box?

I take it you would like to place it above? If so, could you please elaborate on why? Hopefully it would help me understand the principle a little better.

Actually you have the perfect tool to explore this - Xdir The question is in what plane (horizontal or vertical) you want this pattern (thinking of the overlap region where both drivers contribute).

EDIT: However, lobing is only one aspect. Even with brickwall filters (no overlap and hence no interference) a strict line array would be the way to go according to conventional wisdom for imaging reasons. Why spread out the drivers in another dimension (unless called for by size constraints)?

[4313B]

You should take the time to discover for yourself whether or not "you" think that sealed boxes offer better transient response versus ported boxes.

Absolutely!

[Ian Mackenzie]

Bullock on boxes is a great read on this subject.

He says a properly tuned low Q woofer will sound every bit as good as a seal design. He then goes on to discuss the finer points of reflex tuning, accounting for Rx, QL and actually T/L measurements.

I know myself you can alter the quality of the bass in sublte but audible degrees by optimising the woofer for a given enclosure Vb, tuning frequency Fb and Rx the total Dcr behind the speaker.

Ian

[lfh]

I’ve read that sealed boxes have a much better transient response [than ported], ...

Well, it depends. Sealed boxes behave like 2:nd order high pass filters whereas ported boxes behave like 4:th order ones. Thus, a sealed box has the theoretical possibility to perform better in this respect than a ported one. However, performance depends on the "filter" tuning/alignment. A closed box with a high Q-value will "ring"/"boom", and a ported box with a shallow roll off will have a "tight" character. That said, a 2:nd order HP characteristic with a Q around 0.6 (or even lower, when factoring in room gain at lower frequencies) is pretty much as good as it gets, but a ported box tuned with transient performance in mind yields very good results as well. (Not that JBL always used to do this, though: Work through the old JBL recommended tunings with a good box design program some rainy day ). Often the latter is the better overall compromise, when weighing in other aspects such as the desired cutoff frequency, the acoustical output capability etc. Yet another thing to keep in mind is that the "best" box type to a large extent depends on the driver at hand. Typical JBL drivers often work best in ported enclosures (if you still want some LF, that is). Finally I agree with Earl K: Let your ears be the judge and steer clear of dogms ("must use 6 dB filters, must not use feedback, must..." yada yada). Audio design is the fine art of making compromises and many (most?) audiophiles seem to have forgotten that...

[lfh]

You're not doing a middle-of-the-road design here... Do you have some references of similar "hybrid MTMs" (IIRC Earl runs such stuff - I must read the entire K 145 thread) and have you looked into D'Appolitos designs - noteably the crossover networks? How will you tackle the low end where the 2123s runs out of steam but the 2215s keep pumping? I think some careful thinking is needed here when designing the two boxes. (Sure, you're "cheating" with your trick EQ and all, but some master plan on how to design the disparate bass boxes for the desired combined LF response (10"+15"+EQ+room) is called for IMO.) Finally; why is the tweeter not above the mid horn?