line arrays, widerange mid drivers, & crossovers in the "critical range"

line arrays, widerange mid drivers, & crossovers in the "critical range"

I've been a single/widerange(assisted) fan for about 5-7 yrs now, currently owning Hammer Dynamics Super-12's and Cain & Cain IM-Bens w/ matching Bailey subwoofers (old pics of my setup)

[IMGDEAD]http://cgim.audiogon.com/i/vs/s/f/1122340527.jpg[/IMGDEAD] [IMGDEAD]http://cgim.audiogon.com/i/vs/s/f/1122340528.jpg[/IMGDEAD]

But, while doing research for widerange mid drivers for a line array began to question whether crossing over in the "critical range" has to be as detrimental as it is sometimes portrayed. Over the past few years, I have heard several high-end multi-way speakers who had x-o's right-smack-dab in the sacred 350-8kHz range, let alone 300-3,500Hz range, yet still seemed amazingly seamless and coherent.

Specifically, I've had several opportunities to hear:
ATC SCM150ASL (Active) monitors - Crossover Frequency : 380Hz & 3.5kHz - amazingly dynamic, powerful, and definitely some of the most accurate/transparent speakers I've ever heard.
ESP Concert Grand SI - Elliot Midwood of Acoustic Image always rolls these out at audio shows and it is nothing short of awe inspiring (for me).
Avalon Eidolon - Crossover Frequency : 300Hz & 3.5kHz - extremely smooth and seamless integration of multi-way drivers.

Now, granted, those three examples are all in the $25,000-~45,000 range, but I use them simply to show that it is possible to crossover in the "critical range" and still seemingly have coherence that we love(different, but just as engaging). I suppose this is more a matter of designer skill and implementation than sheer cost.

How does this relate to line arrays? Well, line sources are limited by the laws of "comb filtering" and lobing according to Jim Griffin's seminal paper on the subject. Center-to-center driver spacing really dictates the limitations of driver high frequency response and dispersion, no matter how good or expensive the units may be.

For midrange, a line of:
3" drivers will start to display effects at ~3.9k-4.5kHz and above(1 wavelength c-2-c) , depending on the size and shape of the driver frame.
4" at ~3,400Hz
5" at ~2,700Hz

So, I need a mid-driver capable of ~100-3.5k/4.5kHz , beyond that it really doesn't matter(Audience will strongly disagree). In fact, I have been tossing around the idea that since an x-o point is going to end up in the middle of the critical range, regardless, it might be better to trade off HF for extending the LF as close to the 80Hz omni-directional range where it can cross directly to four subwoofers like Audiokinesis' SWARM.

So, what are my (tentative) criterion? I'd like to use sealed enclosures, but could possibly be swayed towards open baffles because of room acoustics. Accuracy/transparency, dynamics, and tonality/timbre - nothing unique about that. I'd prefer a 3" driver to ease up stress of pushing the planar tweeter line too low. Low moving mass would be nice. And since I will be going for a true line of ~7 ft in height, upper limit of ~$30 per driver. (no, I won't run out and buy 60 of them on a whim. I will test a few of the finalists against each other. This is a long term project of a year or more)

Current candidates(no order of preference):
1. Tang Band TGW3-560SE 3" - these are a gamble because they are a close out of custom drivers for a speaker manufacturer who either didn't like them or abandoned the project. As of yet, I haven't found any reports from their actual use. Upside is the TB name and they do a great job of putting out FR curves that actually match their driver's performance. And the truncated driver frame is a huge bonus along with price.
2. Peerless 830987 - that's not a wizzer cone in the picture, just the inverted surround. Scaena actually uses a custom version of this driver in their $40k-80k line arrays.
3. Fountek FR88EX or FR89EX - ZaphAudio tested this driver against many, many, others and declared "This is without a doubt the highest performing 3" I've come across, at least on a few fronts."
4. Tang Band W3-1401SD - nice looking specs and FR curve, truncated frame, FS=75Hz.
5. Aurasound NS3-193-8A - lots of DIY info on this driver and seems to be a pretty reliable performer
6. Hi-Vi C3N-III 3" - a complete unknown, from what I can tell.

My questions to the community are:
1. What important specs I should be considering?
2. Which drivers above are the best performers or look most promising?
3. Are there any others I should be considering?
4. Are there standard, non-full/widerange, mid-woofers capable of ~100-3.5kHz?
 
Last edited:
I want to state my interest in the project. I'll be following this one.

I also want to throw what is perhaps a spanner into the works: The Omnes BB 3.01 three-inch driver. It's juuuust over $30. Made by Tang-Band for the European Omnes Audio distributor. If Tang-Band is Toyota, this is a Lexus – solid manufacturing base, somewhat of a luxury item, but it's still cheap :) I don't know how easily available it is in the US, so perhaps this is not a good suggestion, but I'm impressed enough with its smoothness and clarity that I must mention it.

It's here, a little down the page: OmnesAudio-Lautsprecher
 
About 2 yrs ago now I built a line array kit from Rick at Selah audio.
It incorporates 2 rows of 3" Hi-Vi drivers flanking a row of fountek ribbons.
I am very pleased with the results.....so much so that I had him design another pair using 12-4" Hi-Vi and the Dayton planers. I use these arrays
for my HT....the 3" for the fronts, the 4" for the rears. What is most remarkable about the 3"....called the Symmetricas....is the imaging and soundstage, absolutely the finest I have heard. If you would like to learn more about this design check out Selah at audiocircle.

For as nice as they sound, I use a different set up for my 2 ch listening.
A OB design by a differnt designer, using servo subs and a pro Co-ax driver.

Pics: The arrays

102_1579.jpg


The OB:

102_1830.jpg

102_1859.jpg
 
About 2 yrs ago now I built a line array kit from Rick at Selah audio. It incorporates 2 rows of 3" Hi-Vi drivers flanking a row of fountek ribbons...What is most remarkable about the 3"....called the Symmetricas....is the imaging and soundstage, absolutely the finest I have heard.
Aaaah, I remember the Symmetricas well, it always fascinated me. I actually traded a few posts with Rick on the old Line Array forum about the design right after he introduced it. I always wondered how the dual-line of mid drivers would affect imaging and soundstage as a result of horizontal C-to-C distancing. Rick is a truly talented designer and great guy, he always went out of his way to help DIY'ers.
For as nice as they sound, I use a different set up for my 2 ch listening.A OB design by a differnt designer, using servo subs and a pro Co-ax driver.
Brilliant design and beautiful looking speaker. There is something truly truly unique and special about the presentation of a large fullrange driver, isn't there? Is that a 12" or 15"?

I notice what looks to be a rear-firing tweeter, is there another one on the front?

Also, could you describe what is sonic difference that elevates those speakers over the Symmetricas, or even the large black planar/electrostats(?) in the background?
 
The Symmetricas as I said create a huge sound stage. I was the first to purchase the kit from Rick, at the time he hadnt even finished the x-over
so he couldnt even let me know his initial impressions. I'll never forget our first conversation once he had them up....he said even his most expensive offerings didnt throw a stage like these. Imaging is also spot on, vocals and instruments...
The only thing is they are a bit colored, brass instruments sound not quite as natural as they could, although strings seem seem very realistic. To be fair I have never had them in a properly treated room.....although I have one in the works, should be done about mid Dec... so it will be a new round of critical listening once its complete.

The black planers in the back are Carver Originals. This pair was made in 1987,are absolutely mint.....have all original boxes and packaging. Why do I have these?.....a question my wife asked when I bought them a couple yrs ago. I dont really know, I just always wanted a pair since I was young. They are a nice speaker....the bass is a bit muddy and not as tight as it could be, but thats the way Bob designed them to sound he liked it like that. These are the ones I listen to loud rock music on, they need gobs of power.....I always smile and tap my foot when these are on......they really dont compare to my arrays or OB though. The highs are nice...but just not the same.

The OBs are a design by Danny Richie. This is the first offering, they have evoled a bit since this model. The driver is a 12" coaxial, by P-audio therear firing tweeter is now gone on the newer models, instead he has opened up the back and redesigned the x-over to compensate. On the bottom are 2 servo subs driven with Rythmic amps. The soundstage is not as big as the arrays, but the imaging is excellent. The two things this set up really has going for it is... natural sound, both on vocals and instrumentals. I play both woodwinds and guitar, mandoline etc....these give a very realistic presentation. The other is the tightest bass I have ever heard. Because they are OB they dont load the room, the sound is fantastic. I have a Shefield recording of solo drum performances. These make it sound like the kit is in the same room....a VERY realistic presentation.

Danny Richie was very well received this yr at RMAF with stepped up version of this OB, and Rick Craig had some bookshelves there that were outstanding. Both Rick and Danny are great designers and I am proud to have each of their offerings in my collection.

Perry
 
...

But, while doing research for widerange mid drivers for a line array began to question whether crossing over in the "critical range" has to be as detrimental as it is sometimes portrayed. Over the past few years, I have heard several high-end multi-way speakers who had x-o's right-smack-dab in the sacred 350-8kHz range, let alone 300-3,500Hz range, yet still seemed amazingly seamless and coherent.

...

How does this relate to line arrays? Well, line sources are limited by the laws of "comb filtering" and lobing according to Jim Griffin's seminal paper on the subject. Center-to-center driver spacing really dictates the limitations of driver high frequency response and dispersion, no matter how good or expensive the units may be.

For midrange, a line of:
3" drivers will start to display effects at ~3.9k-4.5kHz and above(1 wavelength c-2-c) , depending on the size and shape of the driver frame.
4" at ~3,400Hz
5" at ~2,700Hz

...


If one cannot avoid interference effects due to driver
distance

-not only center to center but also "edge to edge" since
the effective cone diameter shrinks with frequency due
to non pistonic motion, making the 80% coverage
ratio proposed by Griffin impossible to achieve even for
tightly mounted fullrangers at higher frequencies -

maybe thinking about equidistant mounting being more
a problem than a solution in full- /widerange line arrays
would help ...


Interestingly the need for crossing to a tweeter
(or an array of tweeters), arises from driver
interferences and array dispersion issues, but not
necessarily from bandwith of the wide- fullrange
units being limited.

Instead of solving the problem at its root, further
problems are introduced by going multiway.

Surely kind of "good integration" could be achieved in
some multiways ... to avoid that "integrational effort" by
design seems by far more interesting to me.

Yes i see: Multiple drivers more than a wavelength apart,
cannot work, crazy guy ...

I invite you to examine the vertical dispersion of an
"dipol 08" like design in any common simulator and
compare it against equidistant line arrays due to
"useful vertical listening angles".

Assume the real driver edge to edge distances depending
on frequency, do not assume pistonic drivers, as this is
nothing but wishful thinking.

Not to talk about comparing the horizontal dispersion to
a 2-way (and two line) line array, as horizontal dispersion
often is neglected espacially when designers are too eager
to achieve a certain kind of dispersion in the vertical plane.

With "dipol 08" as an example, please note that the lower
3 drivers are continuously rolled off above above 2Khz.

Using an active solution, even the upper 3 drivers could be
reduced to one single driver in the brillance region.


Kind Regards
 
Last edited:
To be fair I have never had them in a properly treated room.....although I have one in the works, should be done about mid Dec... so it will be a new round of critical listening once its complete.
Yeah, I've come to realize that in untreated rooms, we hear more of acoustic room coloration's than the speaker's true performance. I've battled my living room for years and never really won.
I invite you to examine the..."dipol 08" like design
Thank you, Oliver. I am only a few pages into your well thought out and written website, but it is extremely interesting. Sadly, I haven't used my German since leaving high school, so I have to rely on Google translate. I will read and re-read your website tonight and have plenty of questions for discussion tomorrow.

I have been intrigued by dipole and omni-directional “Coherent Wave Transmission Line Drivers” like the original Ohm Walsh A's & F's as a result year long battle with room acoustics while trying to find an optimal setup in my living for the Cain & Cains. I never was able to win that battle without extensive room treatments. It was an extremely frustrating(and expensive) experience.

BTW, there is someone who has taken Lincoln Walsh's original design, improved it, and used present day materials and computer-aided manufacturing processes to remove most, if not all, of the major problems of the original driver. Dale Harder of HHR-ExoticSpeakers is just a library of knowledge about the Walsh line source from a decades worth of repair and innovation. He says his new Walsh-type drivers can handle more power, are far more durable, have greater output and less distortion/anomalies, etc.

Here's a spec sheet comparison of his new TLS-I & TLS-II speakers versus the original Ohm Walsh A & F
 
This is my answer to c2c distance , comb filtering and "power tapering", etc., and I've found that it works well for dedicated, critical listening situations. This is not a party speaker, not much good for PA either.

However, for solo listening, or even a couple of people, it is quite good and to quote Dr. David Hyre's remarks when he first heard them: "They're like the World's biggest headphones!"

That's not quite true, because you can move your head and the sound is still at the front of the room.

185781d1283063871t-puget-sound-speaker-


The focused array uses 12 1.5 inch "fullrange" drivers and is crossed over to an Adire Audio AV8 at 450 Hz using a series crossover. They recently took
1st place at "The Puget Sound! DIY Speaker Contest" in the 2-way class.

Best Regards,
TerryO
 
Last edited:
Hello TerryO,

i truly believe your speaker having won the
price "well earned".

Nevertheless up to now nobody could ever
consistently explain to me, which specifical
problem in line arrays is solved by focussing ...

Does it contribute to uniform dispersion ?
No IMO.

Can it mitigate comb filtering in a reasonable
wide region around a "sweet spot" ?
No IMO.

Comb filtering cannot be seen independently from
spatial dispersion and listening position.
Even for straight (and also equidistant) arrays you
will find listening positions which are quite balanced ...

The problem is IMO having a fairly balanced response
for a wider range of listening positions due to angle
and distance. Any asymmetry in listener-speaker
distance will show large effects with a focussed
line array.

My personal goal would be a "good natured" behaviour
due to change of listening position in
distance/height/L-R symmetry.

Narrowing the zone of stereophonic listening, cannot
be seen as an advantage in itself IMO.

Kind Regards
 
This unsmoothed FR is taken in 1.5m distance in front of
the dipol 08 array with the microphone placed on the
bottom, which is about 40 degrees off axis (below) vertically
but on axis horizontally, since the acoustic "center" can be
seen close to the second driver from above at about 1.2m
height.

It does not look beautiful surely, but i bet most would have
expected worse at that vertical angle and below the drivers
reproducing the highs, as this is the "LF side" of the array.

It is something like a "worst case FR", and this is not even
a relevant (and only hardly possible) listening position.

The corrsponding step response's long tail is due to a
fairly reverberant room (the FR is gated) but again i
have seen worse ones - even from single fullrangers
on axis ...
 

Attachments

  • Dipol08_Frequenzgang.JPG
    Dipol08_Frequenzgang.JPG
    191.1 KB · Views: 366
  • Dipol08_Sprungantwort.JPG
    Dipol08_Sprungantwort.JPG
    167.3 KB · Views: 334
Last edited:
Linearray,

Golomb Rulers and Babcock's work on intermodulation interference seems like a really interesting way to address comb filtering and other dispersion type problems. I read what I could understand and think I have been able to grasp the simplest parts of the subject.

Could you please explain how you came to decide on which proven ruler distances to use? Did you have to build an adjustable baffle and simply try different ruler spacing by frequency tests?

image005.jpg
 
Hi darkmoebius,


seems you have made a vertically compressed version of "dipol 08" ... :)


It must have been in the early 80's when i first read about the golomb
rulers in the german edition of "Scientific American".

I must have been around 18 when i built a quick setup using 4 cheap
fullrangers attached on an adjustable frame. Each fullranger had its own
small cabinet and baffle.

Left and right speaker as well had 4 drivers, positioned at distances
proportionally to a golomb ruler of length 4.

The setup sounded very homogeneously in my small room, and even though
the drivers were far apart (height of the assembly must have been around
1.2 meters) there was little subjective evidence for listening to multiple
drivers, as long as if you did not listen too close.

The sound was balanced and "naturally spaceous" which was difficult in that
small room of my youth, which had about 12 sqmtrs only.

Then the idea was sleeping for about 2 decades. I built lots of speakers
meanwhile, mostly 2-Ways, and i was experimenting with self built ribbons
and line sources made from these ribbons.

I noticed with the years, that dispersion characteristics and room
interaction has more influence on the sonic result, than the "type of
transducer". The difference between an ESL and a dynamic 2-way is mostly
due to dipolar radiation and directvity ...

I wanted a speaker which sounds the same, even when placed in different
rooms. A high goal and not fully achievable, but you can try to come close.

Combining the dipole and the line array approach using fullrangers came to
my mind ... and i have not been the only one as i can see.

Fullrangers having "bass capable" size cannot be mounted "acoustically close"
for higher frequencies as their center to center distance compares to wavelengths
radiated. As i pointed out before, additionally the active cone area shrinks with
frequency. That means mounting "as close as possible" does not solve the problem
of interference and comb filtering. Also a line source with >80% of its length
covered with "pistonic" motion is not possible ... as the 80% ratio is e.g. given
as a minimum coverage of area in the paper on "Practical Nearfield Line Arrays"
by Griffin.

On the other hand for low frequencies, no close mounting is needed in a
line array. The idea was to have the drivers at distances which "grow together"
with lowering frequency, thereby inherently compensating the rolloff of a
narrow baffle, which is capable of giving horizontal dispersion free from
side lobing.

Then i started some simulations - and my old "golomb" idea came to my mind
again. As i wrote in my german text, dipol 08 is "inspired" by golomb ruler
spacing, but it IS NOT a golomb ruler spacing.

The starting point was an equidistant spacing, capable of compensating the
baffle rolloff, that would be present with a single driver.

Next step was to introduce "golombification" to the upper part of the array,
thereby it's low frequency behaviour could be largely maintained, since the
distances for the 4 lower drivers stayed untouched and the upper distances
shrink progressively. There are other configurations possible, but a decision
among a limited set of patterns had to be made and this turned out to be a
good natured one.

Next step is to introduce power tapering and roll the equidistant part of
the array off towards higher frequencies. That yields in a combination
of "power tapering" and "tapering by distance weighting". Both effects
make the array shrink in its effective size towards higher frequencies,
as the sound sources are more "dense" in the upper part.

Next step was to maintain the dipole radiation up to the highs, which is not
possible when using fullrangers only, since their rear radiation drops above
some Khz, that is what the rear tweeter panel is good for. It is a "fullranger's
rapair kit for rear radiation".

The positions and distances of the drivers were mainly tested and optimized
using "edge". Accordingly a crossover was adapted, since crossover and
distance tapering must go together.

I took some time until i decided for six drivers, but the frequency response
curves at typical listening positions were better than with 5 drivers and
7 drivers did not bring improvement. Also the "3 and 3" approach was going
together well with a passive crossover concept.

Using an active concept, the array could be reduced to just one single driver
in the highs - but the upper trio works amazingly well, even in the brilliance
region, so i would say there is no need for it
(though still room for improvement ...).

The crossover was extensively tested and modified, when the real object was
available (as this is kind of usual i guess ..).

During computational optimization, the arrays bottom mirror image has been
taken into account as an integral part, since there is significant contribution
especially for the bass performance.


Kind Regards
 

Attachments

  • image005.jpg
    image005.jpg
    59.8 KB · Views: 303
  • image008.jpg
    image008.jpg
    35.6 KB · Views: 296
Last edited:
For those who like to play around using EDGE, here
are 2 files representing the approximate configuration
of "dipol 08" as a starting point.

The file with suffix "HT" includes only the 3 upper
drivers and no mirrored baffle. The other file includes
all 6 drivers and a mirrored baffle, representing the
bottom reflections from the array.

The baffle is rotated by 90 degrees in the sketch,
so the bottom line has to be imagined vertically
in the middle of the sketch, separating the double
height baffle (baffle and bottom mirrored image)
into 2 halves.

This way the image fits better into EDGE's sketch
window ... you cannot handle a tall baffle
including bottom reflections otherwise.

If you make changes, be shure to change the
mirrored baffle accordingly - mirorring is done
manually, it is not an EDGE built in feature ...

File "HT" has better approximation above 2-4Khz,
since the lower drivers are rolled off and floor
reflections are less pronounced.

---
EDGE from Tolvan data:

Home of the Edge

Of course you can also use other diffraction simulators
of your choice. Think of the results as being "tendencies",
not representing what we call "reality".
 

Attachments

  • dipol08_edge.zip
    1.4 KB · Views: 35
Last edited:
...

3. Are there any others I should be considering?
4. Are there standard, non-full/widerange, mid-woofers capable of ~100-3.5kHz?


Visaton AL 130

Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories

Has been tested years ago and shows a nearly perfect
spectral decay, you can even listen to it "widerange"
without getting "hurt" ...

But there are some nasty resonances around 10Khz or so, which
are restricting the driver for use as a woofer/midrange.


Maybe this one is also interesting, but i have no experience
with it myself:

http://www.visaton.de/de/chassis_zubehoer/tiefmittelton/ti100_8.html


The parameters to be preferred will largely depend on your decision,
whether to go closed box or open baffle ... as open baffle will call for
higher fs and higher Q.
 
Last edited:
As wide- or fullranger FR 125 S from CSS is still in the
race i would say - in an OB application it has the
advantage of properly fitting parameters - especially
when using a sub crossing at 80Hz - and it has high
linear excursion, which is needed in OB depending on the
amount of drivers used and your requirements due to
maximum output level.
 
As wide- or fullranger FR 125 S from CSS is still in the
race i would say - in an OB application it has the
advantage of properly fitting parameters - especially
when using a sub crossing at 80Hz - and it has high
linear excursion, which is needed in OB depending on the
amount of drivers used and your requirements due to
maximum output level.


The CSS FR 125 is an excellent driver, but would certainly need a high-pass filter if it is to sound it's best, based on my use of them. Having said that, I've yet to hear a line array using them that I liked, and I've certainly heard a number of them, however VMMV.

Best Regards,
TerryO
 
Hello,

my dipol 08 is crossed to a subwoofer at 80 Hz and
the highpass is a PLLXO with 6 db slope.

That reduces excursion and thereby increases dynamic
headroom. As many state, the FR125 excels in a fairly
balanced midrange to my ears also, which is a problem
for most fullrangers of comparable or larger size.

For most listeners it will sound somewhat "dull" in the
highs. With dipol 08 this is not a problem, because
a correction network lifts the highs anyway and the
dispersion is different to a single driver.

With that compensation even the highs are very good,
no sharpness and fine "sparkle" up to the top.

I would describe myself being "fussy" concerning the
highs, as i do not like a speaker with any sharpness,
missing resolution or sparkle, even if everything else is
fine ...

I thought about using a dipole ribbon to integrate into
the dipole 08 design, but after listening to the more
"matured" versions i decided to stay with fullrange, as
there is not much to gain from a tweeter in that design,
instead of - yes - spoiling homogeneity, which is
simply outstanding.

The quality of the highs was a surprise to myself, as i did
not think to get away without a tweeter at firstplace, if
i wanted good resolution and "sparkle".

I was willing to accept somewhat missing sparkle,
balancedness and resolution in the highs from the start
of the project, because i was aiming for bass to upper
midrange homogeneity and "uncoloredness", which i
missed in ANY speaker listened to so far.

Aiming for that consequently, i was also willing to pay
a "sonical price" for it, since that project was pretty
"adventurous" when idesiged it.

That there was no price to pay -in form of deficient highs-
was the biggest surprise to me personally.

And it made me think about subjective "tweeter quality"
afresh ... i think different sonical quality in tweeters is
mainly due to dispersion characterics.

The transducer quality itself (resonances/modes e.g.)
plays an important but inferior role compared to
dispersion, which should be "consistent" and avoid
abrupt changes with frequency. A higher directivity
(compared to a 1inch dome e.g.) is advantageous
in most homely environments, which are quite "lively".

---
At the time i purchased the CSS drivers in germany
serial tolerances and quality was an issue - i deliberately
sent some of them back to the dealer, since i also had to
match the left and the right line array. Hopefully this has
become better now ...

Even the individual drivers were selected, only the
"top performers" in the highs qualify for the
"upper trio" in the dipol 08 array and there is also a
selection where to play in the upper trio.

A driver with is little hot around 7Khz e.g. should not
play at an elevated angle but somewhat lower, so
there are some rules how i select them and match the
left and right speaker.


Best Regards
 
Last edited:
Last edited: