Post by: Yoel Farkas on July 21, 2016, 09:52:20 AM
I currently own and use the Danlay Sound labs J1's, and they are fantastic. I would like to understand the scientific matter between the 2 approaches.
Post by: John Chiara on July 21, 2016, 10:36:17 AM
What are the ups and downs between the "Multi cellular array" technology. like EAW Annya-Anna, Martin Audio MLA, d&b Audio J, Y series. Versos "Single source" systems like Danley sound Labs?
I currently own and use the Danlay Sound labs J1's, and they are fantastic. I would like to understand the scientific matter between the 2 approaches.
Since you are one of a few J1 owners, maybe we can set up a comparison demo somehow. I bet a LOT of us would be interested!
Post by: Helge A Bentsen on July 21, 2016, 11:12:17 AM
like EAW Annya-Anna, Martin Audio MLA, d&b Audio J, Y series.
While it would be fun to attend a J1 demo, those systems you have listed here wary greatly in technological finesse and isn't exactly an apple to apple-comparison.
Sent from my iPhone using Tapatalk
Post by: thirtha chengappa on July 21, 2016, 11:45:26 AM
Heard the VHD 2.0 (2 boxes per side). They managed to give a stiff competition in terms of SPL with several line array brands. The clarity stood out and had a solid lead over the rest.
Post by: Doug Fowler on July 21, 2016, 11:46:09 AM
What are the ups and downs between the "Multi cellular array" technology. like EAW Annya-Anna, Martin Audio MLA, d&b Audio J, Y series. Versos "Single source" systems like Danley sound Labs?
I currently own and use the Danlay Sound labs J1's, and they are fantastic. I would like to understand the scientific matter between the 2 approaches.
I'll let this continue as a discussion about steering and beam forming, but let's leave the Danley products out of it.
I too have a fair amount of experience with J1, but enough of the "whatever" vs point source debate.
Post by: Yoel Farkas on July 21, 2016, 11:58:51 AM
I'll let this continue as a discussion about steering and beam forming, but let's leave the Danley products out of it.
I too have a fair amount of experience with J1, but enough of the "whatever" vs point source debate.
You are right.
my point is not to review the danley products or to compare it to others. my point is as the subject title "Multi Cellular Array" vs "Single Source" (the word array does not really come in here since it is a single source and it is usually not arrayed)
the reason i mentioned Danley is because they are the only one i'm aware of that makes large scale single source systems. KV2 could also be considered. but i don't want to discuss brands . all i want is understand the different technologies.
"Single source" is more simple as it forms a single beam from a single horn. (not to say it is simple to make it a "single source") but "Multi Cellular arrays" have a lot of components with different time offsets so does they overcome the combfiltering interference issue that Line array has?
Post by: Yoel Farkas on July 21, 2016, 12:23:17 PM
Since you are one of a few J1 owners, maybe we can set up a comparison demo somehow. I bet a LOT of us would be interested!I would wish to. our J1's are sitting in storage for the next few months. i would wish to take it out on a Sunday to a location that have hoist and compere it.
but it is only one site of the engagement, we need some one with a "Multi Cellular array" system to compare it to.
Post by: Yoel Farkas on July 21, 2016, 12:36:39 PM
I'll let this continue as a discussion about steering and beam forming, but let's leave the Danley products out of it.Well i know there was a lot of discussions about Danley vs "Whatever". the discussions was about "Line Array" vs "single source". my point here is about a (fairly new) technology of "multi Cellular Array" or "steerable array"
I too have a fair amount of experience with J1, but enough of the "whatever" vs point source debate.
i doesn't want to repeat an old discussion if it was discussed before let me know.
Post by: Tom Danley on July 21, 2016, 01:01:24 PM
I'll let this continue as a discussion about steering and beam forming, but let's leave the Danley products out of it.
I too have a fair amount of experience with J1, but enough of the "whatever" vs point source debate.
Hi Doug,
You didn’t leave much wiggle room to answer Yoel’s question or discuss two completely different approaches to beam forming, or the theory behind each.
While I have made loudspeakers since the 70’s, my interest in how sources behave began working with acoustic levitation for space flight hardware in the 80’s, up until then, I never gave much thought to how sound radiates and tended to think of a pa system as something like a bank of lights. That changed working on levitation systems which required a “beam” of very intense sound and where side lobes were very harmful to levitation stability.
One can form a beam two simple ways, one is a horn, with a CD horn producing a near constant beam width over a range of frequencies and the other is with an array of sources as describe in Huygens theory and originally applied to light.
In the latter, the assumption is that there is both constructive and destructive interference “in front” of the sources as the individual radiations combine and cancel depending on their phase in what’s called Vector addition.
While each source is often too large compared to the wavelength to coherently combine or add everywhere, there is a region in front where they do add constructively. This can be used to produce the desired beam and by altering the phase angle between sources, that beam can be steered. This was applied extensively with sound in the early sonar arrays my old boss and acoustician developed back at Mullard labs in WW2.
The unspoken part of that is that there is more than one beam produced and with loudspeaker which cover a vast span of wavelengths, what you get “out front” is also strongly dependant on both the wavelength and size of the array of sources. The other part is while you get a beam, you also have sources more than ¼ wavelength apart and this dictates that they produce an interference pattern, a pattern of lobes and nulls spatially distributed around the array and visible when a high resolution polar plot is made. This spatially dependent interference pattern and not loudness is what limits the usable “throw” of the array approach.
The big sonic difference between this and a CD horn is seen IF one examines the Time aspect and not the steady state view which is normally the only way arrays are examined. With an array of sources, each one has a “vertical beam width” and this beam width is strongly dependant on frequency and below that point, the sources radiate independently and increasingly broadly and follow the Huygens theorem forming the new radiation.
The CD horn on the other hand confines the radiation from a single driver to the angle defined by the horn wall and dimensions, following Don Keele’s “pattern loss frequency” rule of thumb which in inches is 1X10^6 / horn wall angle / horn mouth dimension. That dictates that a 12 inch tall horn with a 10 degree horn wall angle losses pattern control around 8300Hz for example and the angle doubles each octave you go below that frequency, also the pattern loss frequency halves each time the horn wall angle or horn mouth dimension is doubled.
The differences primarily are that above pattern loss F, the horn not producing an interference pattern, generally radiates much less energy above, below and behind the horn than an array for the same beam width.
In large rooms like stadiums etc, this is a good thing and needed to maximize intelligibility and musical articulation and can be seen in the Hopkins Stryker equation where N is the number of sources and Q being the directivity of the source (the difference between the energy radiated in the desired pattern vs the rest radiated outside the pattern)
http://www.acousticworx.com/sound%20system%20design%20hopkins%20stryker%20formula.html
Secondly, out front in the time domain, the full range horn is a single source and produces one arrival in time, a single finger snap arrives as a single sonic event while the array of separately radiating sources delivers one arrival for each source, each separated in time and at a level depending on frequency, beam width at that frequency and distance to each individual source.
In this regard, for a given identical beam width, the single source is much more like the input signal than the array which is inherently dispersive in the time domain due to the extended size of the radiation area. An Energy vs Time measurement shows this very clearly.
Since we hear not only amplitude but also much of music and voice especially IS TIME VARIANT this part is a key but often forgotten element in this kind of discussion.
Hopefully this was helpful and more scientific than marketing.
Best,
Tom Danley
Post by: Lee Buckalew on July 21, 2016, 04:18:18 PM
What are the ups and downs between the "Multi cellular array" technology. like EAW Annya-Anna, Martin Audio MLA, d&b Audio J, Y series. Versos "Single source" systems like Danley sound Labs?
I currently own and use the Danlay Sound labs J1's, and they are fantastic. I would like to understand the scientific matter between the 2 approaches.
These are each utilizing very different technologies, more than just 2 different approaches here. This is not something that lends itself to an internet discussion. There are some very good technological white papers published with the AES regarding at least some of these. If you are a member of the AES you should be able to access the archives. For MLA technology look up papers by Ambrose Thompson.
I am not sure who has published papers on EAW ANNYA and ANNA. D&B is doing very little in the technology department compared to either EAW or Martin and are really still just a line array with some FIR applied (up to 6 max per cabinet if my information is correct) across the full range cabinet as I am understanding it.
EAW and Martin have very different approaches to each as well.
Lee
Post by: Roland Clarke on July 21, 2016, 07:22:18 PM
These are each utilizing very different technologies, more than just 2 different approaches here. This is not something that lends itself to an internet discussion. There are some very good technological white papers published with the AES regarding at least some of these. If you are a member of the AES you should be able to access the archives. For MLA technology look up papers by Ambrose Thompson.
I am not sure who has published papers on EAW ANNYA and ANNA. D&B is doing very little in the technology department compared to either EAW or Martin and are really still just a line array with some FIR applied (up to 6 max per cabinet if my information is correct) across the full range cabinet as I am understanding it.
EAW and Martin have very different approaches to each as well.
Lee
They might use very different technologies, but they are both owned by Loud technologies.
Post by: Lee Buckalew on July 21, 2016, 07:46:43 PM
They might use very different technologies, but they are both owned by Loud technologies.
Quite true although I do not see how that has any bearing on anything. The R&D teams are independent, manufacturing is independent, etc.
Lee
Post by: Steve Eudaly on July 21, 2016, 09:07:58 PM
D&B is doing very little in the technology department compared to either EAW or Martin and are really still just a line array with some FIR applied (up to 6 max per cabinet if my information is correct) across the full range cabinet as I am understanding it.
d&b's ArrayProcessing is pretty interesting. Each box gets its own amp channel and each is processed differently from its neighbors in the array based on the system and venue design work done in ArrayCalc.
I attended a demo recently and we ran pink noise through each box in the line to show the difference in level and tonality when using ArrayProcessing. I was quite surprised at how much difference there was. It certainly did help with consistency throughout the venue.
Post by: Lee Buckalew on July 21, 2016, 10:08:57 PM
d&b's ArrayProcessing is pretty interesting. Each box gets its own amp channel and each is processed differently from its neighbors in the array based on the system and venue design work done in ArrayCalc.
I attended a demo recently and we ran pink noise through each box in the line to show the difference in level and tonality when using ArrayProcessing. I was quite surprised at how much difference there was. It certainly did help with consistency throughout the venue.
Yes, D&B is making some significant improvements to their line arrays but they are doing 1/1000 or less (literally) of the processing that MLA does. D&B has a maximum of 6 filters (I believe that it is 6) spread across the full range cabinet, not divided by pass band and per driver as MLA has (6000 + or - filters). The accuracy of the models and the resolution provided are also a night and day difference.
They (D&B) are making improvements in the line array interactions of their systems but they still have a line array with line array interference patterns. MLA is creating a single source of sound by creating a complex directional point source. It does not behave as a line array.
Lee
Post by: Roland Clarke on July 22, 2016, 05:12:13 AM
Yes, D&B is making some significant improvements to their line arrays but they are doing 1/1000 or less (literally) of the processing that MLA does. D&B has a maximum of 6 filters (I believe that it is 6) spread across the full range cabinet, not divided by pass band and per driver as MLA has (6000 + or - filters). The accuracy of the models and the resolution provided are also a night and day difference.
They (D&B) are making improvements in the line array interactions of their systems but they still have a line array with line array interference patterns. MLA is creating a single source of sound by creating a complex directional point source. It does not behave as a line array.
Lee
This is true and I heard a very good deployment of MLA last year that sounded very good. The reality is that even with 6000 filters you are still going to get comb filtering in any system that uses multiple drivers that are separated by space. Too my knowledge, Danley Sound Labs are the only manufacturer of a point source, full range speaker. KV2, although not a completely point source system, also have a lot of good subjective reports about their sound quality. I have heard quite a few line array systems from various manufacturers where subjectively the sound was very good. As is constantly mentioned, deployment and venue have a huge effect on the final sound, along with the performers.
Post by: David Sturzenbecher on July 22, 2016, 06:47:04 AM
Too my knowledge, Danley Sound Labs are the only manufacturer of a point source, full range speaker.
Depends what your definition of "full range" is. Fulcrum Acoustics is another fine company that makes single source solutions.
Sent from my iPhone using Tapatalk
Post by: Helge A Bentsen on July 22, 2016, 07:09:09 AM
Post by: Keith Broughton on July 22, 2016, 08:42:51 AM
Depends what your definition of "full range" is. Fulcrum Acoustics is another fine company that makes single source solutions.I had a look on Fulcrums page and I don't see products that are horn loaded and exhibit good pattern control down to the sub 200 range.
Sent from my iPhone using Tapatalk
Co- axials are not quite like the other products mentioned.
Post by: Lee Buckalew on July 22, 2016, 08:48:24 AM
This is true and I heard a very good deployment of MLA last year that sounded very good. The reality is that even with 6000 filters you are still going to get comb filtering in any system that uses multiple drivers that are separated by space. Too my knowledge, Danley Sound Labs are the only manufacturer of a point source, full range speaker. KV2, although not a completely point source system, also have a lot of good subjective reports about their sound quality. I have heard quite a few line array systems from various manufacturers where subjectively the sound was very good. As is constantly mentioned, deployment and venue have a huge effect on the final sound, along with the performers.
Your statement contradicts itself. Danley utilizes "multiple drivers that are separated by space". Therefore, according to your statement, they must produce comb filtering but they claim that they do not. I have never measured them to check (and I am not suggesting that they do not meet this claim) but here on this forum they are given the benefit of the doubt and they are believed when they say they do not.
On the defined listening plane MLA is eliminating comb filtering by creating correct vector summation of all driver interactions. This is true from the smallest recommended deployment to the largest and does not change within recommended cabinet quantity or coverage angle changes. All other systems, not just line arrays, that I am aware of (with the possible exception of ANNYA/ANNA which I do not know well enough to say) either inherently create comb filtering within each cabinet and/or create comb filtering as additional cabinets are added in order to create the desired coverage. There are no other systems that have the modeled to measured accuracy and resolution that MLA has. This does not mean that everyone will prefer it, that it is "the best", or that any of the MLA options are the best solution for every situation.
I routinely come to the defense of MLA here on this forum because many who are talking about it here are accepting of Danley's claims without understanding what Tom is really doing and without measuring what is actually going on yet, at the same time they argue that MLA is simply a processed line array and it still creates the comb filtering interactions expected of a multiple driver, multiple box system even though they are not understanding what is really being done nor have they measured it. Danley is doing some very impressive, amazing things with audio, Martin's MLA systems likewise are doing some very impressive, amazing things with audio. Let's seek to understand what each is doing, how they are doing it, what the values and benefits, drawbacks and limitations are.
Many systems can sound very good, even great. The differences in listener perception for "it sounds really good" at the highest level often come into issues that are outside of the sound system and/or its deployment.
Lee
Post by: Ivan Beaver on July 22, 2016, 09:27:29 AM
Yes-it is impossible to have more than 1 driver occupy the same physical space.
Your statement contradicts itself. Danley utilizes "multiple drivers that are separated by space". Therefore, according to your statement, they must produce comb filtering but they claim that they do not.
HOWEVER-it is possible to have multiple drivers enter into the same physical space-ie a horn, so from an audience perspective they ARE coming from the same physical space.
The "trick" is to get them do so across a wide freq range.
Simply put in pink noise into any system and walk around or get up close and move your head between the drivers/cabinets.
If you hear "swooshy swooshy" then you HAVE combfiltering.
That is EXACTLY how a guitar phase pedal works. You simply have 1 signal delayed a little bit from the dry signal and you have notches in the response. The "phaser" is also changing the delay time, which gives you the "Swirling" effect.
But it is NO different than 2 loudspeakers in different physical spaces.
BTW, a "different physical space" is defined as arrival times that are more than 1/4 wavelength at the highest freq of interest for a particular passband. So subs can be further apart and still be within 1/4 wavelength
Combfiltering is quite easy to measure. Just take a freq response measurement and make sure you have enough resolution.
Or take an ETC measurement. You will easily see the multiple arrivals.
Many people think that combfiltering is this big bad horrible sounding thing. It is not.
When 2 people are talking in a normal room, there is lots of combfiltering going on. Nobody dies or runs from the room.
HOWEVER-if the same 2 people were talking in a room that did not have reflections, the QUALITY of the sound would be better.
Many people hear combfiltering and "think" it is simply a loss of high freq or mid freq if far enough apart.
It is simply a series of notches (that look like the teeth of a comb when viewed on a LINEAR scale.
You CANNOT-not with all the DSP in the world fix time issues with eq. You also cannot-no matter how much delay you use, fix a time issue in multiple places with drivers that are physically spaced.
Post by: Keith Broughton on July 22, 2016, 10:16:28 AM
So please explain how the coupled J3's J4's work. how does they couple and not creating comb filtering in their acoustical crossover region?I don't think anyone would say there is no comb filtering in the array of speakers in the photo. Laws of physics and such.
However, a horn of that type which exhibits well behaved frequency respone at the edge of the coverage angle can provide a more tollerable sounding interaction.
Post by: Lee Buckalew on July 22, 2016, 10:24:27 AM
Yes-it is impossible to have more than 1 driver occupy the same physical space.
HOWEVER-it is possible to have multiple drivers enter into the same physical space-ie a horn, so from an audience perspective they ARE coming from the same physical space.
The "trick" is to get them do so across a wide freq range.
Simply put in pink noise into any system and walk around or get up close and move your head between the drivers/cabinets.
If you hear "swooshy swooshy" then you HAVE combfiltering.
I completely understand this Ivan. There is no "swooshy swooshy" with MLA.
Have you tested and measured them?
As I am sure you are aware comb filtering is not only the "swooshy swooshy" sound that you refer to although that is the most directly identifiable to many people. There is also a significant amount of HF comb filtering created in individual horn throats. Most audio people think this is simply a horn throat distortion and leave it at that. It is actually comb filtering due at least in part to the air impedance change that occurs at the junction of the horn throat to horn and the reflections caused by this. Dave Gunness has addressed this interaction in many (all, I don't know for sure) of the best tunings for his Fulcrum speakers. He is not addressing it by correcting it after it happens, he is correcting it in a way that cancels the reflection in the horn throat through DSP. There can also be path length differences which lead to comb filtering due to reflections within the horn, path length differences within the horn manifold (if so equipped), etc.
You CANNOT-not with all the DSP in the world fix time issues with eq. You also cannot-no matter how much delay you use, fix a time issue in multiple places with drivers that are physically spaced.
Many would have said the same thing about multiple drivers on a single horn. It had been attempted many times, never fully worked, it was not possible, etc. It seems that Danley has figured a way to make it work. Not by changing physics but by a more thorough understanding of them and by asking the right questions to solve the right problems.
"Fixing" something with DSP is not always about changing an output with EQ, delay, etc.
You understand and have stated on this forum that utilizing a filter out of the pass-band of the device to which it is applied can be done in order to change the in pass-band phase response as you desire it. In much the same way changes to phase angle, eq, and amplitude can all be utilized to make multiple precisely known and precisely created physical driver interactions behave as one at a specific known location in space (each measurement point along the listening plane) in much the same way that physically correcting for these interactions on a common horn, once thought to be approachable but perhaps not actually achievable has been done by Danley to create summation at the horn exit.
I also want to be very clear that MLA is not simply applying DSP as a solution. It is first an integrated system of precisely designed and precisely known drivers, driver interactions, cabinet interactions, etc. Processing is then applied to make these precision individual systems/cabinets behave as a single complex directional point source.
Have you directly listened to and/or measured MLA?
You can not stick your head in between drivers and listen for problems because the system is inherently creating summation at the listening plane and not at the cabinets. If you measure the results at the listening plane, which is the target, you will see/experience a lack of comb filtering across the full frequency range of that system.
Still, it is certainly not the right, correct, or proper solution for every situation. Nothing is. When we ask the proper questions we can apply the proper logic in order to achieve an actual solution. If we ask the wrong questions we can direct answers to achieve our desired response but that may not be a solution to the actual problem at hand.
Not saying you are doing this, merely pointing out to those following the thread that often the wrong questions are asked, this in turn leads down a rabbit trail that achieves a result but not necessarily a solution to the problem at hand.
Lee
Post by: David Sturzenbecher on July 22, 2016, 11:23:40 AM
I don't think anyone would say there is no comb filtering in the array of speakers in the photo. Laws of physics and such.
However, a horn of that type which exhibits well behaved frequency respone at the edge of the coverage angle can provide a less more tollerable sounding interaction.
Any idea where this post/picture went?
Post by: Yoel Farkas on July 22, 2016, 11:55:52 AM
i see it is not just the picture it is the whole reply. maybe something wrong with it.
Any idea where this post/picture went?
i understand that 4 single source speakers will be better of any system that has 100 sources.
but my question still apply: what is the approach of coupling those speakers. if the theory is that is have to be a "Single source" then they would have to use one single source like their J5's, and if the J5's are not enough they would have to build something like the Matterhorn. is there a way that those speakers coupled will sum as a single source or it just the best they could do at the time for this coverage requirement? what is the theory behind it?
Post by: Scott Carneval on July 22, 2016, 12:06:17 PM
i see it is not just the picture it is the whole reply. maybe something wrong with it.
Probably deleted as we're 'not allowed to discuss Danley in this thread' lol
Post by: Keith Broughton on July 22, 2016, 12:17:07 PM
There is no way to couple the speakers shown in the (missing) photo as a single, coherent source.
but my question still apply: what is the approach of coupling those speakers. is there a way that those speakers coupled will sum as a single source or it just the best they could do at the time for this coverage requirement? what is the theory behind it?
Your "best they could do" is the most accurate answer.
It's a compromise, as usual.
Post by: Yoel Farkas on July 22, 2016, 12:19:13 PM
Probably deleted as we're 'not allowed to discuss Danley in this thread' loli understand that people are fed-up discussing Danley products over and over. one of the reasons it comes up regularly is that because they are here and they answer questions. others don't answer questions here.
My question was not direct to Danley products. it was about the approach of Single source systems. if someone feels threaten by this post/question, it will not help to delete the post. because science doesn't change and the question is still here. if not now it will come up another time from another member.
Post by: John Chiara on July 22, 2016, 12:24:37 PM
i understand that people are fed-up discussing Danley products over and over. one of the reasons it comes up regularly is that because they are here and they answer questions. others don't answer questions here.
This.....
Post by: Tim McCulloch on July 22, 2016, 12:24:49 PM
Probably deleted as we're 'not allowed to discuss Danley in this thread' lol
Tom and the folks at DSL have managed to do something that inventors and manufacturers really want: a proprietary and patented way of doing something that is an improvement over existing, similar things. That gives them a virtually unlimited ability to be in an internet comparison as there is nobody else doing what they do the way they do it.
While I have great respect for Tom and have worked casually with Ivan several years ago on Paul Bell's NYC subwoofer shootout, I think this has turned any mention of D.S.L. products into a marketing platform. It's somewhat akin to the iPod and iPad. When released there was no immediate competition, but lots of fan-boy action. In some respects that is the case for Danley products.
In the hands of owner-operators I'm fairly sure that there isn't much out there that can equal what D.S.L. does for either size or price point, but I've heard (and measured and evaluated) a couple of installations that used Danley products exclusively and they don't sound good. That's out of Tom, Mike and Ivan's hands - and I don't assess blame to them or DSL for the result - I offer these observations to point out that whatever 'secret sauce' is used, the positive results of that can be instantly and permanently negated by a problematic design or FUBAR installation provisions by the general contractor, or errors by the installing dealer... or misapplication by end users.
I'm certainly interested in hearing less fan-boy and more input regarding other technologies that are raising the bar of both sound quality and touring practicality.
Post by: Yoel Farkas on July 22, 2016, 12:31:39 PM
In the hands of owner-operators I'm fairly sure that there isn't much out there that can equal what D.S.L. does for either size or price point, but I've heard (and measured and evaluated) a couple of installations that used Danley products exclusively and they don't sound good. That's out of Tom,my question was about the theory of coupling single source speakers. as they where designed and installed by DSL. i believe they sound good. my question is how. what are theory behind it.
Post by: Lee Buckalew on July 22, 2016, 12:46:22 PM
... a proprietary and patented way of doing something that is an improvement over existing, similar things. That gives them a virtually unlimited ability to be in an internet comparison as there is nobody else doing what they do the way they do it.
MLA is also in this category. The processing and/or its application is patented.
I'm certainly interested in hearing less fan-boy and more input regarding other technologies that are raising the bar of both sound quality and touring practicality.
I could not agree more!
Actual users with data and quantifiable results and also information on the various pluses and minuses.
How does it rig, how does it travel, what aspects work well/don't work well in given situations, reliability, service, etc., etc.
Lee
Post by: Doug Fowler on July 22, 2016, 02:15:25 PM
Probably deleted as we're 'not allowed to discuss Danley in this thread' lol
That's why I deleted it. I don't want another "Danley vs the world" thread via thread drift.
Post by: eric lenasbunt on July 22, 2016, 05:36:52 PM
In the hands of owner-operators I'm fairly sure that there isn't much out there that can equal what D.S.L. does for either size or price point, but I've heard (and measured and evaluated) a couple of installations that used Danley products exclusively and they don't sound good. That's out of Tom, Mike and Ivan's hands - and I don't assess blame to them or DSL for the result - I offer these observations to point out that whatever 'secret sauce' is used, the positive results of that can be instantly and permanently negated by a problematic design or FUBAR installation provisions by the general contractor, or errors by the installing dealer... or misapplication by end users.
Tim is spot on as always. Besides poor installation or implementation there is also just the basic problem that some mix engineers suck at their job. We install in a lot of churches with volunteer sound teams. While some are great, others obviously aren't absorbing the training and in some cases have made our very flat, VERY good sounding installs sound quite horrible. There are a few churches around that I am proud of our install work but not proud to tell people it is our install, as the operators make it sound horrible.
Even the nicest car can be crashed into a tree by a driver who doesn't know what they are doing.
Post by: David Sturzenbecher on July 22, 2016, 05:43:49 PM
That's why I deleted it. I don't want another "Danley vs the world" thread via thread drift.
The picture was essentially a Danley line array. Not sure what else you would call 4 boxes stacked in a vertical line at 0 degree splay...What do you call it??? Just google "LSU Danley Jericho" in an image search and it will come up on the first page. It's odd that the post with the picture gets deleted, but not Ivan's post. Without context it's meaningless, and not germane to the topic. This thread is surely beyond simple elementary electro-acoustics.
Post by: Ivan Beaver on July 22, 2016, 09:53:34 PM
The picture was essentially a Danley line array. Not sure what else you would call 4 boxes stacked in a vertical line at 0 degree splay...What do you call it??? Just google "LSU Danley Jericho" in an image search and it will come up on the first page. It's odd that the post with the picture gets deleted, but not Ivan's post. Without context it's meaningless, and not germane to the topic. This thread is surely beyond simple elementary electro-acoustics.Yes it is 4 J3-64 cabinets stacked on top of each other.
HOWEVER-it is important to understand some of the theory and design intent behind it.
Most large stadiums use a single J1 for the far seats.
But because LSU is known for being loud (they set off the seismic detectors in the earthquake lab during a play at one game) the consultant wanted "a bit more" in the system. (there is a different solution available now that was not available when the install was done)
It is not the physical distance between the elements that causes comb filtering. But rather the distance in time between the elements that the listener hears.
Since the closest listener for these cabinets is around 700' away, the signal arrival time difference are very small.
At that distance, the air absorption pretty much kills anything above 4Khz-unless it is REALLY loud (that is where the "tweeter cabinets" come into play).
So where the differences in arrival would be causing interference in freq response, the air is killing those freq anyway. So it is actually minimal interaction. Up closer it would be a different story-but there are no listeners there.
It is important to consider how a particular system is used and the intended outcome.
I could go into more details, but should probably stop now.
Post by: Lee Buckalew on July 23, 2016, 12:09:52 AM
Yes it is 4 J3-64 cabinets stacked on top of each other.
HOWEVER-it is important to understand some of the theory and design intent behind it.
Most large stadiums use a single J1 for the far seats.
But because LSU is known for being loud (they set off the seismic detectors in the earthquake lab during a play at one game) the consultant wanted "a bit more" in the system. (there is a different solution available now that was not available when the install was done)
It is not the physical distance between the elements that causes comb filtering. But rather the distance in time between the elements that the listener hears.
Since the closest listener for these cabinets is around 700' away, the signal arrival time difference are very small.
At that distance, the air absorption pretty much kills anything above 4Khz-unless it is REALLY loud (that is where the "tweeter cabinets" come into play).
So where the differences in arrival would be causing interference in freq response, the air is killing those freq anyway. So it is actually minimal interaction. Up closer it would be a different story-but there are no listeners there.
It is important to consider how a particular system is used and the intended outcome.
I could go into more details, but should probably stop now.
For those following along who do not routinely deal in comb-filters and time offsets here is an equivalent for a comb-filter with the first null at 4kHz.
1.685" distance offset = 0.125 ms offset = 1st null at 4kHz and 1st peak at 8kHz for the comb-filter.
Lee
Post by: Steve Anderson on July 23, 2016, 02:00:33 AM
For those following along who do not routinely deal in comb-filters and time offsets here is an equivalent for a comb-filter with the first null at 4kHz.And at 700' on axis the distance delta between middle and end of array is about 0.55" so first null would be at 12.4kHz… which is well and truly above the HF air loss freq Ivan referred to.
1.685" distance offset = 0.125 ms offset = 1st null at 4kHz and 1st peak at 8kHz for the comb-filter.
Lee
Not to say that entire audience is on axis at that distance, but at least for the ones that are, this "chosen compromise" design decision stands up to scrutiny.
Post by: David Sturzenbecher on July 23, 2016, 03:09:03 AM
And at 700' on axis the distance delta between middle and end of array is about 0.55" so first null would be at 12.4kHz… which is well and truly above the HF air loss freq Ivan referred to.
Not to say that entire audience is on axis at that distance, but at least for the ones that are, this "chosen compromise" design decision stands up to scrutiny.
This is also the same "design" that goes into the HF and MF section of every line array on the planet, worth its salt of course.
Sent from my iPhone using Tapatalk
Post by: Lee Buckalew on July 23, 2016, 09:04:14 AM
And at 700' on axis the distance delta between middle and end of array is about 0.55" so first null would be at 12.4kHz… which is well and truly above the HF air loss freq Ivan referred to.
Not to say that entire audience is on axis at that distance, but at least for the ones that are, this "chosen compromise" design decision stands up to scrutiny.
I was merely providing the distance and time delay required for a first null for the 4kHz frequency that Ivan mentioned. I made no suggestion that the design would not "stand up to scrutiny".
Since this thread is ultimately about how cabinets interact let's look at this specific compromise, understanding that it is representative of a common compromise that is necessary in many designs, not just for the now removed photo but for all separately mounted speaker cabinets and most individual cabinets with separate components.
In this case it is one compromise to utilize a secondary enclosure to gain additional HF because it creates a distance between the primary enclosure and the secondary. If air absorption is creating an acoustic crossover at 4kHz from the main cabinet and the additional HF cabinet is augmenting that beggining at 4kHz then those cabinets would need to be within 1.685" vector delta centerline of HF to centerline of HF in order to avoid comb-filtering.
If the vector delta is great enough they would also require very steep crossover slopes or the effective beggining of the comb-filter would be at a lower frequency.
A second common compromise is utilizing separate enclosures to cover a wider horizontal area. When two cabinets are used, or in the case of MLA two arrays, to cover adjacent areas there will be an area of overlap that does, at some frequencies and dependant on the vector delta between band-pass center lines, produce comb-filtering. That is usually preferable to leaving a hole in the coverage.
Physical design changes in cabinets or in array layouts can only accommodate the physics involved up to a certain point before other factors become more important. Things like size, weight, SPL capability, etc.
Most of us use lip fills in a large deployment. They all produce comb-filtering. There is no way to arrange multiple speakers along the face of the stage so that you transition from one coverage area to the next with no overlap. The same holds true for out fills. This is one of those areas that can be minimized but not eliminated once we begin to choose our compromises for real world coverage.
Lee
Post by: Ivan Beaver on July 23, 2016, 09:30:45 AM
This is also the same "design" that goes into the HF and MF section of every line array on the planet, worth its salt of course.Except that with line arrays, the audiences are MUCH closer to the cabinets, making the difference in distances MUCH larger-lowering the null point
Sent from my iPhone using Tapatalk
In this case the cabinets are well above peoples heads and the horns are larger and wider coverage angles-so the pattern control extends MUCH lower.
The horns on line arrays are very small and narrow.
So they lose pattern control up high.
A 12" tall horn that has a 10 degree coverage pattern (typical for many line arrays) will lose pattern control around 8Khz
So below that freq the horn will simply be "spraying" sound everwhere.
As the pattern gets narrower-the horn MUST get larger to have the same control to the same freq.
Or for a given size-a wider coverage horn will control to a lower freq
So when you are closer- you will hear the different arrivals.
This is VERY EASILY measured when looking at and ETC or impulse response of a line array. Especially when looking at the higher freq.
Post by: Ivan Beaver on July 23, 2016, 09:39:00 AM
Maximum comb filtering happens when both signals are the same level
In this case it is one compromise to utilize a secondary enclosure to gain additional HF because it creates a distance between the primary enclosure and the secondary. If air absorption is creating an acoustic crossover at 4kHz from the main cabinet and the additional HF cabinet is augmenting that beggining at 4kHz then those cabinets would need to be within 1.685" vector delta centerline of HF to centerline of HF in order to avoid comb-filtering.
When one level goes down, the amount of interference goes down.
In this case- the HF cabinet is MUCH louder than the full range cabinet.
Since there is nobody within 700' or so of the cabinets-this level difference is not important at those closer distances.
Is this approach perfect-no. But no line array could possibly get the HF out to those seats at that distance as well.
But that is EXACTLY why there is now a much louder approach, that is a single cabinet (using 108 drivers) to "reach out and touch" those seats, that is being used in the large stadiums.
The latest one is being installed at FSU as we speak
Post by: Lee Buckalew on July 23, 2016, 11:46:33 AM
Maximum comb filtering happens when both signals are the same level
When one level goes down, the amount of interference goes down.
In this case- the HF cabinet is MUCH louder than the full range cabinet.
Since there is nobody within 700' or so of the cabinets-this level difference is not important at those closer distances.
Is this approach perfect-no. But no line array could possibly get the HF out to those seats at that distance as well.
But that is EXACTLY why there is now a much louder approach, that is a single cabinet (using 108 drivers) to "reach out and touch" those seats, that is being used in the large stadiums.
The latest one is being installed at FSU as we speak
Understood Ivan, not picking apart a specific design, merely answering some of Yoel's questions regarding what are the compromises involved.
The same issue however does not present with a level difference in the horizontal plane when butting two coverage patters to extend width of coverage, nor does a level difference present itself at the crossover region between a full range cabinet and HF supplemental devices.
Usually the distance involved, as has been pointed out in this thread, where the supplemental HF is needed creates a vector delta that eliminates the comb-filter or moves it's starting point to a high enough frequency that air absorption stops it from being heard although, if it is present, even though not heard, it will, as has also been previously pointed out, create nulls and peaks that create non-coherent propagation and limit "throw".
If the comb-filtering can be eliminated then the coherence of the acoustic signal is maintained and greater "throw" is achieved.
That is what you guys are doing with larger and larger coherent horn devices and that is what MLA is doing with a combination of driver, cabinet, and processing designs/implementation. I do not have access to it right now but Left Right hangs of full size MLA have been utilized for coverage of between 700' - 900' with minimal variation up to 12.5kHz. Memory says less than +/-3dB but I certainly may be remembering incorrectly.
Lee
Post by: Mac Kerr on July 23, 2016, 12:05:50 PM
It is not the physical distance between the elements that causes comb filtering. But rather the distance in time between the elements that the listener hears.
Since the closest listener for these cabinets is around 700' away, the signal arrival time difference are very small.
Very much like the listeners far from a line array who are hearing multiple elements of high frequency vs close listeners who are hearing fewer.
Mac
Post by: Ivan Beaver on July 23, 2016, 12:22:50 PM
I do not have access to it right now but Left Right hangs of full size MLA have been utilized for coverage of between 700' - 900' with minimal variation up to 12.5kHz. Memory says less than 3dB but I certainly may be remembering incorrectly.I have a VERY HARD time believing that you are getting out to 12K with anywhere near a 3dB variation at 900'.
Lee
When you consider that the normal difference between 10M and 300M is around 29dB. When you add air absorption (Yes it does vary with humidity and temp), that become 74dB-give or take good bit.
That is a HUGE HUGE HUGE difference to make up with DSP.
In the particular case in question, the coverage pattern of the "tweeters" was 30* wide x 10* tall. NO attempt was made to try to cover the near seats-in fact we wanted to miss them as much as possible-due to the levels being in excess of 160dB coming out of the "tweeters".
There were 256 1" compression drivers pointed just towards the rear seats on large horns.
The main cabinets got to around 4Khz and pretty much died at those seats.
The "tweeters" got us up to 8Khz. There was really nothing going beyond that.
Post by: Ivan Beaver on July 23, 2016, 12:26:18 PM
Very much like the listeners far from a line array who are hearing multiple elements of high frequency vs close listeners who are hearing fewer.
Mac
It depends on the freq involved. Yes-in the top octave, the small horns (relative to the coverage pattern) will control the sound and keep the levels down for people out of the pattern, but as the freq goes down, they will lose pattern control and people will hear the multiple arrivals at close seats.
What this results in is a "smearing" of the sound, that will make transient things not seem as real. ESPECIALLY percussive instruments and picked instruments.
Blown or bowed instruments will not suffer as much.
The "detail" is what goes missing when you have multiple arrivals
Post by: Lee Buckalew on July 23, 2016, 12:49:53 PM
I have a VERY HARD time believing that you are getting out to 12K with anywhere near a 3dB variation at 900'.
When you consider that the normal difference between 10M and 300M is around 29dB. When you add air absorption (Yes it does vary with humidity and temp), that become 74dB-give or take good bit.
That is a HUGE HUGE HUGE difference to make up with DSP.
In the particular case in question, the coverage pattern of the "tweeters" was 30* wide x 10* tall. NO attempt was made to try to cover the near seats-in fact we wanted to miss them as much as possible-due to the levels being in excess of 160dB coming out of the "tweeters".
There were 256 1" compression drivers pointed just towards the rear seats on large horns.
The main cabinets got to around 4Khz and pretty much died at those seats.
The "tweeters" got us up to 8Khz. There was really nothing going beyond that.
Like I said, I don't have access right now to the info.
I do know that there have been many concerts done with less than the full compliment of 24 boxes per side (typically 18 or 19 MLA and a single MLA-D) with coverage out to 250 meters and drop off created at 300 meters to avoid noise complaints.
I will try to find the info that I have on the other coverage. It was between 700' and 900' if I am remembering correctly although my memory is not what it used to be. :-)
Lee
Post by: Mac Kerr on July 23, 2016, 01:37:33 PM
It depends on the freq involved. Yes-in the top octave, the small horns (relative to the coverage pattern) will control the sound and keep the levels down for people out of the pattern, but as the freq goes down, they will lose pattern control and people will hear the multiple arrivals at close seats.
And as the frequency goes down the wavelength goes up, making the phase change critical point work for greater spacing between drivers.
Mac
Post by: Ivan Beaver on July 23, 2016, 01:43:06 PM
Like I said, I don't have access right now to the info.It is the freq above 4K that is affected the most by air absorption.
I do know that there have been many concerts done with less than the full compliment of 24 boxes per side (typically 18 or 19 MLA and a single MLA-D) with coverage out to 250 meters and drop off created at 300 meters to avoid noise complaints.
I will try to find the info that I have on the other coverage. It was between 700' and 900' if I am remembering correctly although my memory is not what it used to be. :-)
Lee
Sure-there is plenty of sound below that. But once you get above and actually MEASURE-you QUICKLY realize that there is a "problem".
I remember measuring my first stadium (BYU).
We had a mic at the far seats-about 700' in this case- and the response was rolling off on the top end.
So we boost a little at 10K. Nothing happened. We boosted some more-nothing happened. I did not want to add more than 12dB of boost. The response was so low on the screen (as compared to the lower part of the response) we could not even see it.
This was using a TEF 20.
That night at the hotel I did some research on air absorption and realized there was NO amount of boosting that was going to get the HF response up at those seats.
That is when I said to my self-we need a "real" super tweeter.
And even then all we picked up was an octave.
But considering that is higher than most systems go anyway-I guess it is fine.
When you actually MEASURE at those distances-you realize how "tall the mountain" is you are looking at.
But casual listening will not tell you that.
Simply saying "I have highs out that far" does not mean anything unless you attach some REAL numbers to it.
Of course the "excuse" that many people like to use is "Well your brain is used to the HF rolling off at a distance-so it is OK">
OK-Let's "buy" that argument for a second.
BUT LOOK at the VIDEO screen! They make it VERY LARGE so as to move you close to it. In REALITY, a body on the screen would be very small and you could not see it at that distance.
So you would expect the sound to be "far away" as well.
But since they have moved the image closer to you-should not the sound be the same?
Should not the objective of any sound system be to provide the same experience for everybody? Or as close as we can?
That INCLUDES full freq response as best as possible-along with the same SPL.
YES- there will be exceptions and variances, that "should" be the goal.
I have a good number of "stories" that I can't tell (or maybe I could if I left off manufacturers names-I'm not sure on that one) about ACTUAL "like" experiences of line arrays and long distances vs a single source approach.
If people are interested in that-then I need to get moderators approval first. But they would be interesting reading.
Post by: Lee Buckalew on July 23, 2016, 01:53:46 PM
I have a good number of "stories" that I can't tell (or maybe I could if I left off manufacturers names-I'm not sure on that one) about ACTUAL "like" experiences of line arrays and long distances vs a single source approach.
I am not talking about a line array, I am talking about MLA.
Lee
Post by: Ivan Beaver on July 23, 2016, 02:08:57 PM
I am not talking about a line array, I am talking about MLA.Sorry-but I don't have any stories about any comparisons with MLA because I am not aware of any that have been done.
Lee
Post by: Lee Buckalew on July 23, 2016, 11:04:06 PM
Like I said, I don't have access right now to the info.
I do know that there have been many concerts done with less than the full compliment of 24 boxes per side (typically 18 or 19 MLA and a single MLA-D) with coverage out to 250 meters and drop off created at 300 meters to avoid noise complaints.
I will try to find the info that I have on the other coverage. It was between 700' and 900' if I am remembering correctly although my memory is not what it used to be. :-)
Lee
Just following up on this. Since I was going by memory before.
I can't find the write up that I was thinking of (seems to me it was in Japan and was 2 - 24 box hangs at something over 700', I just can't come up with it) but I have found others referencing 200 and 250 meter coverage for specific events with MLA and MLD. The 200 meter mark seems to be the typical maximum before most people want to add delays.
There are no specific published measurements in the write ups so I did my own Display file with very little playing to smooth things out or try different array heights, starting distances, etc.
In a quickly created file I can keep response to approximately +/- 6dB from 100Hz to 10kHz, 20' in front of the array to 650' in front of the array (approximately 200 meters). This is only with standard configuration (splay and EEQ calculations) and not any system equalization factored into zones of the array. This is a single array of 23 MLA and 1 MLD.
Display predictions vs real world results have been shown to be accurate to +/- 0.5 dB/SPL.
There are 72 - 1" HF devices in a 24 box MLA array.
There could be MASSIVE differences in the result depending on humidity. High humidity areas could get away with this while arid areas could not.
Lee
Post by: Robert Lunceford on July 24, 2016, 01:05:46 AM
Quote from: Lee Buckalew link=topic=160077.msg1470975#msg147097292593
I do know that there have been many concerts done with less than the full compliment of 24 boxes per side ....
Something else to consider,
MLA nominal power consumption is 900 watts X 48 boxes = 43,200 watts
MLA weight is 193 lbs X 48 boxes = 9,264 lbs
Post by: Ivan Beaver on July 24, 2016, 05:55:54 AM
Just following up on this. Since I was going by memory before.IS that with air absorption turned on?
I can't find the write up that I was thinking of (seems to me it was in Japan and was 2 - 24 box hangs at something over 700', I just can't come up with it) but I have found others referencing 200 and 250 meter coverage for specific events with MLA and MLD. The 200 meter mark seems to be the typical maximum before most people want to add delays.
There are no specific published measurements in the write ups so I did my own Display file with very little playing to smooth things out or try different array heights, starting distances, etc.
In a quickly created file I can keep response to approximately +/- 6dB from 100Hz to 10kHz, 20' in front of the array to 650' in front of the array (approximately 200 meters). This is only with standard configuration (splay and EEQ calculations) and not any system equalization factored into zones of the array. This is a single array of 23 MLA and 1 MLD.
Display predictions vs real world results have been shown to be accurate to +/- 0.5 dB/SPL.
There are 72 - 1" HF devices in a 24 box MLA array.
There could be MASSIVE differences in the result depending on humidity. High humidity areas could get away with this while arid areas could not.
Lee
It makes a HUGE difference.
The numbers I gave are based on actual measurements in a venue-NOT predictions.
Every bodies predictions are much better with air absorption turned off
Post by: Lee Buckalew on July 24, 2016, 07:04:08 AM
IS that with air absorption turned on?
It makes a HUGE difference.
The numbers I gave are based on actual measurements in a venue-NOT predictions.
Every bodies predictions are much better with air absorption turned off
Yes, air absorption is on.
Lee
Post by: Lee Buckalew on July 24, 2016, 07:23:39 AM
5 date=1469
Something else to consider,
MLA nominal power consumption is 900 watts X 48 boxes = 43,200 watts
MLA weight is 193 lbs X 48 boxes = 9,264 lbs
Yes, although I would not use the nominal power rating, since peak is 6000 watt capability. They are 100, 120, or 208/240 volt capable.
JBL VerTec VT4889ADPDA is 205 lbs per cabinet with a similar power draw and universal voltage capability. As a line array it lacks the coherence capability for extremely long "throws" of HF.
Meyer LEO is 265 lbs per cabinet and capable of 208/240 only (no 100, 120 volt ability). Also a standard line array.
All of the other manufacturers fall right around this same weight, size, power, and per cabinet output specification. Mid-sized boxes is where the size, weight, power, and output per cabinet specs start to vary more significantly.
MLA falls right into the weight and power consumption, per box, of similar sized concert PA cabinets. It has higher output than others of similar size due to its greater coherence.
Other threads here have hashed out the portability, reconfigurability, etc. concerns of the touring and portable PA folks.
Lee
Post by: Hayden J. Nebus on July 25, 2016, 04:10:25 PM
D&B is doing very little in the technology department compared to either EAW or Martin and are really still just a line array with some FIR applied (up to 6 max per cabinet if my information is correct) across the full range cabinet as I am understanding it.
Lee
I will vouch that there's a serious amount of maths happening with d&b array processing. The crux of their approach looks to be a whole lot of vector summing.
I've literally just gotten the chance yesterday to point a mic at a d&b rig with array processing. 8x Y series on 30D amps, set to front/back -0dB SPL drop (smaller venue) , with Glory@11. Here's what I've gleaned on how it works. Just the facts (and some postulations) .
It appears to me that each cabinet gets its own linear magnitude (phase only) FIR filter, a parametric-allpass, if you will. AP adds 6.5ms throughput latency on the 30D. latency = N-1 / 2*fs, so it's likely to be 625 tap filters at 48kHz. There's enough ripple to the filters that 1/6 octave phase smoothing was required to make the trace human-legible, so IMO 1250 taps @ 96kHz, which I would expect to look a lot smoother, isn't likely. On this rig, d&b applies the AP filters from roughly 1k-20k.
Array calc reckons multiple data points of raw response curves along the vertical axis of the array. In "glory", each cabinet's mid-hi band phase response is adjusted such that the resultant frequency response is as similar as possible at each of the reckoned data points.
The result is interesting. The magnitude response is very consistent everywhere. A coherence-weighted average of multiple transfer functions looks a lot like just another data point. As a result, the rig is very responsive to EQ, and it responds to EQ the same way everywhere, much unlike most vertical arrays.
The resulting phase response at each seat falls somewhere within +/-45° of "normal" at the nyquist frequency, which nets a 90° distribution. I did not get to measure each box on/off axis to see what the AP is actually doing per box, but rather looked at the whole rig from multiple locations.
I can post some traces later, or open a new thread on the measurement forum if Doug or Mac would prefer.
Post by: Lee Buckalew on July 25, 2016, 06:32:52 PM
I will vouch that there's a serious amount of maths happening with d&b array processing. The crux of their approach looks to be a whole lot of vector summing.
I've literally just gotten the chance yesterday to point a mic at a d&b rig with array processing. 8x Y series on 30D amps, set to front/back -0dB SPL drop (smaller venue) , with Glory@11. Here's what I've gleaned on how it works. Just the facts (and some postulations) .
It appears to me that each cabinet gets its own linear magnitude (phase only) FIR filter, a parametric-allpass, if you will. AP adds 6.5ms throughput latency on the 30D. latency = N-1 / 2*fs, so it's likely to be 625 tap filters at 48kHz. There's enough ripple to the filters that 1/6 octave phase smoothing was required to make the trace human-legible, so IMO 1250 taps @ 96kHz, which I would expect to look a lot smoother, isn't likely. On this rig, d&b applies the AP filters from roughly 1k-20k.
Array calc reckons multiple data points of raw response curves along the vertical axis of the array. In "glory", each cabinet's mid-hi band phase response is adjusted such that the resultant frequency response is as similar as possible at each of the reckoned data points.
The result is interesting. The magnitude response is very consistent everywhere. A coherence-weighted average of multiple transfer functions looks a lot like just another data point. As a result, the rig is very responsive to EQ, and it responds to EQ the same way everywhere, much unlike most vertical arrays.
The resulting phase response at each seat falls somewhere within +/-45° of "normal" at the nyquist frequency, which nets a 90° distribution. I did not get to measure each box on/off axis to see what the AP is actually doing per box, but rather looked at the whole rig from multiple locations.
I can post some traces later, or open a new thread on the measurement forum if Doug or Mac would prefer.
First, I know, I write too much. :-)
Second, I may seem to be arguing but, for those who know me, they will attest that I am not. I am merely explaining differences and also hearing and understanding different approaches and different technologies.
I will be very interested to see/hear how D&B's ArrayCalc compares ultimately. Thus far it has been utilized for a demo of D&B in a room which has also demoed K2, KARA, Nexo, Lyon, and MLA-C (I may be forgetting 1 or 2). Measurement and listening comparisons have been interesting.
I was not meaning to suggest that there is not "a lot" of math being done to produce the results that D&B is now getting, just that there is significantly more math occurring in Display.
The raw number of filters able to be utilized by MLA is many times that of D&B's ArrayCalc. Given 10 years of refinement to MLA technology and 3 complete speaker lines all utilizing MLA's patented numerical optimization process I expect MLA to be the front-runner that everyone is chasing and comparing themselves to. Understanding what is being done with it and how and why it was developed has really helped me to have a greater understanding of applied acoustical principals.
My understanding is that D&B are doing something similar to what Clair is doing and what VUE is doing in terms of applying array processing. As I understand it it is full range cabinet processing, not driver by driver, and I think it is limited to 6 filter points per cabinet. This certainly helps to significantly smooth overall response.
On the other hand, MLA is providing driver by driver processing with up to 6000 filter points per array. This should equate to 250 filter points per cabinet since each box contains all of the processing that it utilizes. This is 41 filter points (and change) per drive section, per cabinet but I do not know how the filter point distribution is within the 6 independently powered and processed driver sections.
Martin Display analyzes multiple data points as well, I do not know how many data points are being calculated in D&B's calculations nor how many virtual test points they are using but, I believe that Martin Display is utilizing 1,800,000 per virtual measurement microphone for a full array of full size MLA (1,200,000 and 1,800,000 are both in my mind and I do not recall which is correct). So a 100 meter deep audience area would have 540,000,000 calculations for just the audience plane (if I am doing my math correctly), add to that the height of the walls, length of the ceiling, height of the backstage wall, etc. and you get an idea for how much math is being done here.
In Display virtual measurement mics are placed every 1/3 meter in the listening plane of the audience area and every 1 meter in the non-audience area, on-axis. Calculations are based on a combination of positional, flanked, and isolated cabinet data. This is because it was found that utilizing isolated cabinet data alone (which is what was used by all that I am aware of when MLA was in development) resulted in errors of as much as 8dB. The requirement for accuracy for modeling was found to be 1dB (+/- 0.5dB) which could not be obtained by utilizing isolated data. MLA utilizes balloon data generated with a combination of positional, flanked, and isolated data and confirmed to be accurate to +/- 0.5dB with a resolution of 1 degree through 360 degrees both vertically and horizontally.
This is utilized to create for an 8 box array for example, 5 unique balloons plus three mirror image balloons.
When all array angle calculations and elemental EQ are complete in Display the resulting balloon data is utilized to create a single complex directional point source balloon. This CDPS can be utilized in EASE for room modeling where EASE is the preferred method.
Here is a link to an indoor measured hang of 12 boxes MLA from 2009. http://forums.prosoundweb.com/index.php/topic,159272.msg1463372.html#msg1463372
Since it is now a patented approach it will be interesting to see what other developments happen to avoid infringement yet allow for the flexibility and resolution that MLA has introduced.
Lee
Post by: Mark Wilkinson on July 25, 2016, 07:54:29 PM
The resulting phase response at each seat falls somewhere within +/-45° of "normal" at the nyquist frequency, which nets a 90° distribution. I did not get to measure each box on/off axis to see what the AP is actually doing per box, but rather looked at the whole rig from multiple locations.
Hi, need help......my understanding is that nyquist frequency is 1/2 sampling rate...so 24Khz for a 48Khz dsp....?
What freq are you referring to 'within +/- 45 degree of "normal" '?
Thx, mark
Post by: Roland Clarke on July 26, 2016, 08:01:35 AM
First, I know, I write too much. :-)
Second, I may seem to be arguing but, for those who know me, they will attest that I am not. I am merely explaining differences and also hearing and understanding different approaches and different technologies.
I will be very interested to see/hear how D&B's ArrayCalc compares ultimately. Thus far it has been utilized for a demo of D&B in a room which has also demoed K2, KARA, Nexo, Lyon, and MLA-C (I may be forgetting 1 or 2). Measurement and listening comparisons have been interesting.
I was not meaning to suggest that there is not "a lot" of math being done to produce the results that D&B is now getting, just that there is significantly more math occurring in Display.
The raw number of filters able to be utilized by MLA is many times that of D&B's ArrayCalc. Given 10 years of refinement to MLA technology and 3 complete speaker lines all utilizing MLA's patented numerical optimization process I expect MLA to be the front-runner that everyone is chasing and comparing themselves to. Understanding what is being done with it and how and why it was developed has really helped me to have a greater understanding of applied acoustical principals.
My understanding is that D&B are doing something similar to what Clair is doing and what VUE is doing in terms of applying array processing. As I understand it it is full range cabinet processing, not driver by driver, and I think it is limited to 6 filter points per cabinet. This certainly helps to significantly smooth overall response.
On the other hand, MLA is providing driver by driver processing with up to 6000 filter points per array. This should equate to 250 filter points per cabinet since each box contains all of the processing that it utilizes. This is 41 filter points (and change) per drive section, per cabinet but I do not know how the filter point distribution is within the 6 independently powered and processed driver sections.
Martin Display analyzes multiple data points as well, I do not know how many data points are being calculated in D&B's calculations nor how many virtual test points they are using but, I believe that Martin Display is utilizing 1,800,000 per virtual measurement microphone for a full array of full size MLA (1,200,000 and 1,800,000 are both in my mind and I do not recall which is correct). So a 100 meter deep audience area would have 540,000,000 calculations for just the audience plane (if I am doing my math correctly), add to that the height of the walls, length of the ceiling, height of the backstage wall, etc. and you get an idea for how much math is being done here.
In Display virtual measurement mics are placed every 1/3 meter in the listening plane of the audience area and every 1 meter in the non-audience area, on-axis. Calculations are based on a combination of positional, flanked, and isolated cabinet data. This is because it was found that utilizing isolated cabinet data alone (which is what was used by all that I am aware of when MLA was in development) resulted in errors of as much as 8dB. The requirement for accuracy for modeling was found to be 1dB (+/- 0.5dB) which could not be obtained by utilizing isolated data. MLA utilizes balloon data generated with a combination of positional, flanked, and isolated data and confirmed to be accurate to +/- 0.5dB with a resolution of 1 degree through 360 degrees both vertically and horizontally.
This is utilized to create for an 8 box array for example, 5 unique balloons plus three mirror image balloons.
When all array angle calculations and elemental EQ are complete in Display the resulting balloon data is utilized to create a single complex directional point source balloon. This CDPS can be utilized in EASE for room modeling where EASE is the preferred method.
Here is a link to an indoor measured hang of 12 boxes MLA from 2009. http://forums.prosoundweb.com/index.php/topic,159272.msg1463372.html#msg1463372
Since it is now a patented approach it will be interesting to see what other developments happen to avoid infringement yet allow for the flexibility and resolution that MLA has introduced.
Lee
Hi Lee,
I just took a quick look at the plots posted in the link you provided and though coverage looked relatively even, it appears that (according to the scale provided), hf is significantly down at about 14khz, some 20db and nosedives above that, pretty much in line with what Ivan was talking about earlier. Whilst I have had a subjectively good experience hearing MLA, it still looks like there are many of the issues that all systems suffer from due to interaction from multiple drivers in multiple boxes, though I would concede these would be likely much worse without the processing power MLA offers.
Post by: Lee Buckalew on July 26, 2016, 08:29:21 AM
Hi Lee,
I just took a quick look at the plots posted in the link you provided and though coverage looked relatively even, it appears that (according to the scale provided), hf is significantly down at about 14khz, some 20db and nosedives above that, pretty much in line with what Ivan was talking about earlier. Whilst I have had a subjectively good experience hearing MLA, it still looks like there are many of the issues that all systems suffer from due to interaction from multiple drivers in multiple boxes, though I would concede these would be likely much worse without the processing power MLA offers.
With MLA it is all based on configuration provided by the processing and numerical optimization. Up to the limits of air absorption you can make it give you what you want. Often that means tapering HF to limit reflective bounce from floor to wall, etc.
It is all based on choices for the particular space, is there a need for a large area of Hard Avoid on the stage, does the back wall or the floor create unwanted reflections that limit intelligibility. All of this is configurable within Display and the results can be modeled very, very accurately although it does not account for room modes or other significant acoustic room interactions only the simple boundary reflections as far as I am aware.
The point of the plots that I linked to was to show the impulse response in a highly reverberant space and the overall consistency of frequency response front to rear. At 75 meters there is no issue with HF roll-off if you don't want it there.
This thread started as a discussion about what each of the various newer processed systems are doing and how they compare. I would choose to compare their technical capabilities since " how do they sound" has significant variables that are far beyond the ability to type a few paragraphs on an online forum.
Cellular Drive is the only technology that starts with the listeners ear and allows the designer/system tech the ability to create the result they want to achieve (at the listener) and translate that into requirements for each individual speaker element in order to achieve that goal. All others also attempt to achieve a particular result but the focus of the processing is to create that result at the speaker system and project that created field out into the listening area. Cellular Drive does not care what is happening at the speaker system (it is also not limited to the current form factor).
Below is a quote from Martin's product specialist, Jim Jorgensen from a recent email discussion that we were having about various systems for a project that we have upcoming. The discussion was based around what is being done with the various approaches of different manufacturers and how do they differ. The most basic is, what problem do they seek to solve?
From Jim,
“To compare different points of sound propagation takes a bit more than a paragraph, and is even more often a theory than an application. What happens when we simply place a full range point source on top of another full range point source? What happens when you put one next to the other? Why would there be any difference (and there is) if they are both truly point sources? The true difference with “cellular technology” is not about source propagation. It is giving each driver or set of drivers its own information with-in any type of array in order to meet a predetermined set of goals. What and how that information is derived is now protected by a patent and the current form factor is only available in the MLA family of products. “
Lee
Post by: Doug Fowler on July 26, 2016, 11:30:17 AM
something similar to what Clair is doing and what VUE is doing in terms of applying array processing
AFMG FIRmaker
http://firmaker.afmg.eu/
Post by: Lee Buckalew on July 26, 2016, 11:47:59 AM
AFMG FIRmaker
http://firmaker.afmg.eu/
Exactly
Lee
Post by: Mark Wilkinson on July 26, 2016, 12:18:53 PM
Hi Lee,
I just took a quick look at the plots posted in the link you provided and though coverage looked relatively even, it appears that (according to the scale provided), hf is significantly down at about 14khz, some 20db and nosedives above that, pretty much in line with what Ivan was talking about earlier. Whilst I have had a subjectively good experience hearing MLA, it still looks like there are many of the issues that all systems suffer from due to interaction from multiple drivers in multiple boxes, though I would concede these would be likely much worse without the processing power MLA offers.
I just can't see how DSP processing can do fine optimization for multiple drivers at 14Khz and higher....given current processing speeds.
At 14khz frequency, one cycle takes 1/14,000 second, or about 0.07ms. So 0.07ms for 360 degrees.
A 98khz dsp takes 1/98,000 second for one sample, or about 0.01ms. So 0.01ms is the smallest time adjustment the dsp can make.
Doesn't that mean that the smallest adjustment that a 98khz dsp can make at 14khz frequency,
is 0.01/0.07 * 360 or 51 degrees ?
That seems pretty coarse, and obviously gets worse as freq climbs.
48khz processing looks like it would almost be worthless for aligning multiple VHF drivers.
But I guess the error between any two adjacent VHF drivers, each individually driven, would be +/- one half the minimum adjustment... so maybe summation isn't too bad...at 14khz and 98k processing...
It's just hard to see how this slippage fits together for more than a couple drivers..
Post by: Lee Buckalew on July 26, 2016, 05:13:06 PM
I just can't see how DSP processing can do fine optimization for multiple drivers at 14Khz and higher....given current processing speeds.
At 14khz frequency, one cycle takes 1/14,000 second, or about 0.07ms. So 0.07ms for 360 degrees.
A 98khz dsp takes 1/98,000 second for one sample, or about 0.01ms. So 0.01ms is the smallest time adjustment the dsp can make.
Doesn't that mean that the smallest adjustment that a 98khz dsp can make at 14khz frequency,
is 0.01/0.07 * 360 or 51 degrees ?
That seems pretty coarse, and obviously gets worse as freq climbs.
48khz processing looks like it would almost be worthless for aligning multiple VHF drivers.
But I guess the error between any two adjacent VHF drivers, each individually driven, would be +/- one half the minimum adjustment... so maybe summation isn't too bad...at 14khz and 98k processing...
It's just hard to see how this slippage fits together for more than a couple drivers..
Phase correction is not about time delay as you are thinking of it, it is about phase correction via the proper filter selection and design. With FIR processing delay is created to allow for time to process the required coefficients.
Here is some info from the QSYS information page on creating custom FIR in QSYS. A FIR coefficient includes; magnitude (20Hz - 20kHz, -20dB to 20dB), magnitude/phase (same magnitude adjustment but adds phase from -180 degrees to 180 degrees) , impulse (amplitude vs. time from 0.00ms to 21.3ms)
The larger problem for FIR filters is LF control, not HF control. The lower the frequency being affected by an FIR filter the longer the processing takes so you get greater processing delay. FIR is not always the correct filter choice depending on what you are trying to accomplish.
Lee
Post by: Tom Danley on July 26, 2016, 05:57:15 PM
Phase correction is not about time delay as you are thinking of it, it is about phase correction via the proper filter selection and design. With FIR processing delay is created to allow for time to process the required coefficients.
Here is some info from the QSYS information page on creating custom FIR in QSYS. A FIR coefficient includes; magnitude (20Hz - 20kHz, -20dB to 20dB), magnitude/phase (same magnitude adjustment but adds phase from -180 degrees to 180 degrees) , impulse (amplitude vs. time from 0.00ms to 21.3ms)
The larger problem for FIR filters is LF control, not HF control. The lower the frequency being affected by an FIR filter the longer the processing takes so you get greater processing delay. FIR is not always the correct filter choice depending on what you are trying to accomplish.
Lee
"The larger problem for FIR filters is LF control, not HF control. The lower the frequency being affected by an FIR filter the longer the processing takes so you get greater processing delay."
This part is correct but the reason is that since there is no such thing as inverse delay (read ahead processing), to correct the phase at low frequencies, everything else must be delayed back to the lowest frequency / phase one is correcting. The processing latency is added to this FIR imposed correction delay for a total processing delay
Tom
Post by: Mark Wilkinson on July 26, 2016, 08:06:10 PM
Phase correction is not about time delay as you are thinking of it, it is about phase correction via the proper filter selection and design. With FIR processing delay is created to allow for time to process the required coefficients.
Here is some info from the QSYS information page on creating custom FIR in QSYS. A FIR coefficient includes; magnitude (20Hz - 20kHz, -20dB to 20dB), magnitude/phase (same magnitude adjustment but adds phase from -180 degrees to 180 degrees) , impulse (amplitude vs. time from 0.00ms to 21.3ms)
The larger problem for FIR filters is LF control, not HF control. The lower the frequency being affected by an FIR filter the longer the processing takes so you get greater processing delay. FIR is not always the correct filter choice depending on what you are trying to accomplish.
Lee
Thanks Lee,
I've been doing a lot of FIR experimentation, with a lot of taps available...6000+ per channel, for a 4 way system...
Yes, as you say.....I get the the issues down low, and think they definitely extend into lower mid as well...
What has surprised me, and I what i don't see mentioned much on the web, are the HF/VHF issues I've experienced.
Maybe the issues don't exist and i am just in the state of self-learning that always has the huge learning holes.....good chance...:)
but as i try to tune HF to VHF that crosses at about 6300, I find that this has been more problematic than expected.
I know that sampling freq determines the smallest time alignments possible.
Past time alignment I am left with trying to fine tune and compensate phase alignment via FIR guesswork. I say guess work, because I've found that predicted flat phase via FIR generation, and then measured flat phase, seems questionable at VHF.
I'm even wondering how good our dual FFt's are at measuring VHF given their 48k sampling...
Maybe, this is where the engineers are doing what I can't yet see how to do...adjust minute phase at VHF via the FIR files they generate (or IIR embedded in FIR). Either way the same.
Dunno ...
But I do know enough to get that FIR correction has difficulty down low (due processing time)...and I currently believe it has difficulty up high too (due to processing speed).
So it makes me wonder about claims i see...
Having said that, I totally think the MLA processing tech is awesome.
Hell, I want to buy a big "multi-driver horn-loaded point-source" and try my hand with active FIR... to see if it can be tuned even further ;D
Post by: Tom Danley on July 26, 2016, 09:24:21 PM
If you’re having problems at high frequencies you might be running into something else.
The magic of FIR filters is that it separates adjustment of the magnitude from the phase where with a circuitry based conventional equalizer any change in the magnitude also changes phase following the ”minimum phase” relationship.
While transducers are mostly minimum phase what happens as the sound radiates may not be.
The issue is that while you can correct many things about a loudspeaker, the tough part up high becomes basing that correction on the “right thing”.
By that I mean you can correct the output which is tied to the voice coil but not spatial issues forward of that.
The problem up high is often the device does not radiate or have a pattern like a simple expanding sphere and could be shaped more like a sea urchin and so, one must take a number of measurements across the front and average them to get to the correctable portion of the radiation and NOT the spatially variant part.
If you don’t do that, you might be chasing your tail correcting spatial issues related to the radiation shape and not the transducer.
6300Hz is high enough where you might be dealing with the interference patterns radiation shape with two drivers operating at the same frequency. They need to be within about a half inch apart edge to edge to radiate as a single new source and at an inch or more are independent sources.
As for the "multi-driver horn-loaded point-source", what I have done is used a physical offset instead of signal time delay and use adapted shape crossovers to do most of the crossover phase correcting but one can go farther with FIR filters if you have an active system.
A low budget but good way to play is using mini-dsp and Re-phase
https://www.minidsp.com/
https://www.minidsp.com/applications/advanced-tools/rephase-fir-tool
Best
Tom Danley
Post by: Lee Buckalew on July 26, 2016, 10:20:10 PM
Thanks Lee,
I've been doing a lot of FIR experimentation, with a lot of taps available...6000+ per channel, for a 4 way system...
Yes, as you say.....I get the the issues down low, and think they definitely extend into lower mid as well...
What has surprised me, and I what i don't see mentioned much on the web, are the HF/VHF issues I've experienced.
Maybe the issues don't exist and i am just in the state of self-learning that always has the huge learning holes.....good chance...:)
but as i try to tune HF to VHF that crosses at about 6300, I find that this has been more problematic than expected.
I know that sampling freq determines the smallest time alignments possible.
Past time alignment I am left with trying to fine tune and compensate phase alignment via FIR guesswork. I say guess work, because I've found that predicted flat phase via FIR generation, and then measured flat phase, seems questionable at VHF.
I'm even wondering how good our dual FFt's are at measuring VHF given their 48k sampling...
Maybe, this is where the engineers are doing what I can't yet see how to do...adjust minute phase at VHF via the FIR files they generate (or IIR embedded in FIR). Either way the same.
Dunno ...
But I do know enough to get that FIR correction has difficulty down low (due processing time)...and I currently believe it has difficulty up high too (due to processing speed).
So it makes me wonder about claims i see...
Having said that, I totally think the MLA processing tech is awesome.
Hell, I want to buy a big "multi-driver horn-loaded point-source" and try my hand with active FIR... to see if it can be tuned even further ;D
A couple of things here.
Read and reread Tom's comments.
Think about what you said about trying to create a flat phase speaker output. This is not what cellular drive does.
A flat phase output at the speaker cabinets face can never create a phase coherent coverage at the audience plane when it is combined with any other cabinets. Cellular Drive is creating a phase and frequency coherent coverage across the audience plane, not at the speaker.
Lee
Post by: Roland Clarke on July 27, 2016, 05:18:09 AM
Perhaps we need to start looking more at finding practical solutions? It's well known that our hearing adapts, go to a concert with too much top end on the system and after about 10 mins it sounds about right. Similarly, engineers tweaking eq only to find the eq is by-passed. Perhaps more work in exploiting the weaknesses in our perception to gain where we are most sensitive?
Post by: Merlijn van Veen on July 27, 2016, 06:43:16 AM
Taking into the equation the number of possible changes available to a system be it MLA, ANNYA, KV2, Danley, L'acoustics, Meyer, D&B, et al, there wouldn't be enough permutations to provide close to ideal.
"An exact and unique solution does not exist since in general there are more equations than unknowns" - Dr. Evert Start
AFAIK Dr. Evert Start of Duran-Audio aka Axys in the Netherlands, now owned by Harman / JBL Pro, was the first to come up with a solution to this challenge.
Duran was seriously pioneering with beam-steering going back to 1994. Their column loudspeakers with Digital Directivity Synthesis can be found all over the globe in airport and train terminals, churches and cathedrals and HOW's (close to 300 in Mekka alone).
Click here (http://www.jblpro.com/www/jbl-story/innovation/technology/directivity/intellivox-dds#.V5iMNpDtSrU) to read the full story.
Verzonden vanaf mijn iPad met Tapatalk
Post by: Lee Buckalew on July 27, 2016, 06:44:31 AM
Ultimately, isn't the problem trying to achieve reasonably flat phase and frequency response at every listener location in the hall or on the field?
The question is how to do this, the problem(s) would be those things that stop us from doing this. That is precisely why Cellular Drive development began.
But that is not the question that was generally being asked, the question that was typically being asked by speaker builders was, "how do we achieve flat phase and frequency response from each driver, speaker enclosure, combination of speaker enclosures?" Which is a great question where we have a single listener and/or in the case of a fully coherent multi-driver assembly, when we never need to use more than one assembly/cabinet to achieve the necessary coverage.
Taking into the equation the number of possible changes available to a system be it MLA, ANNYA, KV2, Danley, L'acoustics, Meyer, D&B, et al, there wouldn't be enough permutations to provide close to ideal.
But it is being done with Cellular Drive in the MLA form factor now through a combination of purpose designed, purpose built; drivers, speaker cabinets, measurement models, modeling software, etc. It certainly does require many, many very intensive mathematical computations. It took a military supercomputer to do the model development.
It is a completely different approach that no one else has done because they have taken into account how the speaker elements and speakers actually behave in each possible combination and then used this to create accurate models.
Perhaps we need to start looking more at finding practical solutions? It's well known that our hearing adapts, go to a concert with too much top end on the system and after about 10 mins it sounds about right. Similarly, engineers tweaking eq only to find the eq is by-passed. Perhaps more work in exploiting the weaknesses in our perception to gain where we are most sensitive?
Unfortunately we can not hear what has been cancelled. We may be able to get accustomed to it (that's the way this room, band, concert, movie, sounds) but that does not mean that we do not want to translate an artistic vision, be it the music of a band or the soundtrack and dialogue of a movie, to the listener as accurately and with as high a resolution as is possible when given the opportunity.
Exploiting the weaknesses in our auditory perception is a job for sound designers of film, TV, and movies as they use psychoacoustic principles to trick us into believing we are hearing what we are not or we are not hearing what we are. In most cases, with a few exceptions, the job of the sound system designer, system tech, mixing engineer is to accurately translate an input or group of inputs so that it can be heard as intended by an audience.
Lee
Post by: Steve M Smith on July 27, 2016, 08:04:56 AM
engineers tweaking eq only to find the eq is by-passed.
As if anyone would ever do that!!!
Steve.
Post by: thirtha chengappa on July 27, 2016, 09:50:36 AM
As far as the audiences ears go, we would not have this if they had astute hearing that we members have on this forum -
http://forums.prosoundweb.com/index.php/topic,160129.0.html
And they, the audience eventually pay us, directly or indirectly.
The point is this, ...... everything comes down to price - Point source VS line array.
More importantly, it, unfortunately comes down to quantity. For the price of a KV2 VHD 2.0 or the VHD 5, the numbers that comes in comparison to a conventional "line array" manufactured by a "reputed" manufacturer would make it more attractive for the vendor, the event organizer, and the artiest to see, not to forget the audience. The proof of that, is that a particular well marketed brand's logo can virtually add +3 DB to the rig at play in the audience's perception. Sad but true.
I have attended 3 palm show (India) line array demonstrations in the past 3 years. The "point source" KV2 VHD2 stood out in comparison to the other 12 odd line arrays out there, each of these 12 odd lines hangs with nothing less that 8 numbers, and in a 3 way format to boot.
The KV2 was way ahead in term of clarity and SPL. And yet, the only sales that made every year were made by the manufacturers of "conventional" line array formats. To the best of my knowledge, there is only one VHD rig in India right now. Sad but true.
Is this the same problem in the rest of the world? As far as India goes, this seems to be the case. It's not that there is a lack of money. The sales personal at Harman India proudly told me that one Indian state capital alone has 12 Vertec 4889 rigs with 8 + modules per side minimum (matching number of subs). The same state capital has 5 VTX rigs with 10 per side minimum (matching number of subs). India has 22 states. Harman is just another manufacturer in India. This, after we play close to 48 % in tax on the Invoice value.
Why is it, that the point source concept like the KV2, and not to forget Danley, not taken off In a big market like India? I Am baffled. Is it just the lack of marketing or the lack of understanding of a concept?
Post by: Tom Danley on July 27, 2016, 10:32:46 AM
I really appreciate the tech talk that has taken place on this thread. Taking the liberty to deviate a bit, it would be relevant to pay attention to real world market, the quality of the audiences ears, and unfortunately, their eyes.
As far as the audiences ears go, we would not have this if they had astute hearing that we members have on this forum -
http://forums.prosoundweb.com/index.php/topic,160129.0.html
And they, the audience eventually pay us, directly or indirectly.
The point is this, ...... everything comes down to price - Point source VS line array.
More importantly, it, unfortunately comes down to quantity. For the price of a KV2 VHD 2.0 or the VHD 5, the numbers that comes in comparison to a conventional "line array" manufactured by a "reputed" manufacturer would make it more attractive for the vendor, the event organizer, and the artiest to see, not to forget the audience. The proof of that, is that a particular well marketed brand's logo can virtually add +3 DB to the rig at play in the audience's perception. Sad but true.
I have attended 3 palm show (India) line array demonstrations in the past 3 years. The "point source" KV2 VHD2 stood out in comparison to the other 12 odd line arrays out there, each of these 12 odd lines hangs with nothing less that 8 numbers, and in a 3 way format to boot.
The KV2 was way ahead in term of clarity and SPL. And yet, the only sales that made every year were made by the manufacturers of "conventional" line array formats. To the best of my knowledge, there is only one VHD rig in India right now. Sad but true.
Is this the same problem in the rest of the world? As far as India goes, this seems to be the case. It's not that there is a lack of money. The sales personal at Harman India proudly told me that one Indian state capital alone has 12 Vertec 4889 rigs with 8 + modules per side minimum (matching number of subs). The same state capital has 5 VTX rigs with 10 per side minimum (matching number of subs). India has 22 states. Harman is just another manufacturer in India. This, after we play close to 48 % in tax on the Invoice value.
Why is it, that the point source concept like the KV2, and not to forget Danley, not taken off In a big market like India? I Am baffled. Is it just the lack of marketing or the lack of understanding of a concept?
Hi Thirtha
In any significant market sector be it from cars, to wood preservatives, to politics, to food products, the largest, best selling and most well known brands are normally not the ones that deliver the biggest “bang for the buck” or best performance. Like they say in marketing “a dollar spent promoting the image of science produces more sales than a dollar spent on science”
The top sales positions instead normally goes to the ones with the most marketing and promotional efforts.
Since you mentioned us by name, I have not heard the KV2 products but they may be up against what we recognized to be a nearly unwinnable battle against the cumulative effect of so many companies marketing line arrays, in fact while a number of us are live sound folks, we elected not to even try to compete in the live sound area where the arrays are so dominant and the lore so deeply embedded.
Instead we focused on large venues and stadiums where the owners aren’t happy with line arrays they have and don’t give a hoot about brand recognition or rider acceptance and with essentially no advertising, in the last 6 years, as a result of side by side demo’s like you speak of, have replaced those systems with ours in about half of the country’s 100,000+ seat stadiums so far with more on the way next season.
So far as your country, I would guess the same things drive popularity in your sound market as they do here but like here, there are probably some who still seek the best results. If i were KV2 in your country, i would try to reach them and do demos.
Best,
Tom Danley
Post by: thirtha chengappa on July 27, 2016, 10:59:32 AM
Hi Thirtha
Like they say in marketing “a dollar spent promoting the image of science produces more sales than a dollar spent on science”
The top sales positions instead normally goes to the ones with the most marketing and promotional efforts.
Hats off to you sir. You said it best. Why I took your name is that I inherently understand what you say and I take the liberty to quote you .....
" It is the technology which allows us to put multiple drivers of different bandwidths into a single horn, and make them behave as a single full range loudspeaker. "
I put you and KV2 on the same page because both are, .... to the best of my understanding, please correct me if I am wrong, trying to do the same thing on a very broad spectrum.
Having heard many a line array, especially the Chinese brands that do the most damage via selling substandard stuff for a cheap price, I can say that the technology that you propound makes sense on many levels. I must admit that I have not heard any of your products, but would love to do so. It's very nice of you to take the time out to reply to my post. Thank you.
Post by: Mike Hedden on July 27, 2016, 11:13:23 AM
"An exact and unique solution does not exist since in general there are more equations than unknowns" - Dr. Evert Start
AFAIK Dr. Evert Start of Duran-Audio aka Axys in the Netherlands, now owned by Harman / JBL Pro, was the first to come up with a solution to this challenge.
Verzonden vanaf mijn iPad met Tapatalk
Why Dr. Start isn't a household name in the world of the dsp controlled arrays is beyond me. He developed these approaches many years before most of the products that are now considered "groundbreaking" were conceived. Every steerable column product currently on the market was developed in reaction to Dr. Start's Intellivox products and his concert products, Axsys, were also years ahead of their time. At the Syn Aud Con Loudspeaker workshop in Louisville Ky over a decade ago, Dr. Start did a thorough presentation as to how they measured their Axsys product and to date, I've never seen another line array technical presentation this detailed and revealing. He also appeared to be to be a humble gentlemen.
Mike Hedden
Danley Sound Labs, Inc.
Post by: Mark Wilkinson on July 27, 2016, 11:20:13 AM
Thanks Tom, Thanks Lee,
Yes, miniDSP and rephase was the only way this poor boy could find, to jump into FIR world with enough taps to really experiment down low.
Tom, your reply and Lee's posts are opening my eyes to where I've been having difficulty with VHF, and understanding what MLA is doing. I don't want to mire this great thread with my specific VHF issue, but for me at least it turns into an example of how Cellar Drive differs from traditional tuning.
I've used rephase to flatten driver-by-driver magnitude and phase, and add linear-phase x-over filters, via miniDSP.
It's pretty easy to get individual driver traces that look good in smaart...(even on the subs when I can live with the processing time lol.) The HF and VHF come from a single BMS4594 coax....both traces show flat zero phase throughout their passbands and 48LR x-over regions around 6300hz.
So then, I'd put it all together, adjust levels, and set time offsets via the dsp in my power amps running at 48khz.
System trace looks great, except VHF inevitably slopes up or down about 60 degrees from about 14khz to 20khz, depending on one click of the smallest time offset I can make via the amp (0.02ms). This was the genesis of my earlier doubt.
I've realized the fallacy of tuning to a single location, and have made a 3 mic rig on a boom for averaging measurements, and still get the same "can't phase flatten the final tail" phenom.
I'm thinking the interference issues you point me to may be...or I'm just still screwing something up :)
Truly would love to be able to try an active FIR setup on one of your boxes. Given their extraordinary physical alignment, I have to believe it would be some of the easiest FIR work ever !!!
At any rate, I apologize if this has swerved excessively, and to return...
Lee,
in a later post you say "the question that was typically being asked by speaker builders was, "how do we achieve flat phase and frequency response from each driver, speaker enclosure, combination of speaker enclosures?" Which is a great question where we have a single listener and/or in the case of a fully coherent multi-driver assembly, when we never need to use more than one assembly/cabinet to achieve the necessary coverage."
Yes, this is exactly what I've been doing and have trying to apply that thinking to line arrays, ultimately believing it just boiled down to adjusting driver by driver....
Let's just say the Cellular Drive lights went on and I'm seeing the need for MLA's immense computational power!!!!!
Thx guys, Mark
Post by: Lee Buckalew on July 27, 2016, 12:16:59 PM
Lee,
in a later post you say "the question that was typically being asked by speaker builders was, "how do we achieve flat phase and frequency response from each driver, speaker enclosure, combination of speaker enclosures?" Which is a great question where we have a single listener and/or in the case of a fully coherent multi-driver assembly, when we never need to use more than one assembly/cabinet to achieve the necessary coverage."
Yes, this is exactly what I've been doing and have trying to apply that thinking to line arrays, ultimately believing it just boiled down to adjusting driver by driver....
Let's just say the Cellular Drive lights went on and I'm seeing the need for MLA's immense computational power!!!!!
Thx guys, Mark
Glad to help turn the lights on :-)
That's the frustrating part of frequently running into the uninformed opinion that MLA is "just using beam steering", or is "just a processed line array". It is neither, Cellular Drive is a unique approach and is now a patented process. It is not a form factor or a cabinet layout, it is a complete process that is different than anything that has been done before.
That still does not make it the right choice for everything or a perfect solution by any means but it is a very powerful tool to understand and have in your toolbar.
Lee
Post by: Tom Danley on July 27, 2016, 04:02:23 PM
Thanks Tom, Thanks Lee,
Yes, miniDSP and rephase was the only way this poor boy could find, to jump into FIR world with enough taps to really experiment down low.
Tom, your reply and Lee's posts are opening my eyes to where I've been having difficulty with VHF, and understanding what MLA is doing. I don't want to mire this great thread with my specific VHF issue, but for me at least it turns into an example of how Cellar Drive differs from traditional tuning.
I've used rephase to flatten driver-by-driver magnitude and phase, and add linear-phase x-over filters, via miniDSP.
It's pretty easy to get individual driver traces that look good in smaart...(even on the subs when I can live with the processing time lol.) The HF and VHF come from a single BMS4594 coax....both traces show flat zero phase throughout their passbands and 48LR x-over regions around 6300hz.
So then, I'd put it all together, adjust levels, and set time offsets via the dsp in my power amps running at 48khz.
System trace looks great, except VHF inevitably slopes up or down about 60 degrees from about 14khz to 20khz, depending on one click of the smallest time offset I can make via the amp (0.02ms). This was the genesis of my earlier doubt.
I've realized the fallacy of tuning to a single location, and have made a 3 mic rig on a boom for averaging measurements, and still get the same "can't phase flatten the final tail" phenom.
I'm thinking the interference issues you point me to may be...or I'm just still screwing something up :)
Hi Mark
Try getting it as close as possible first without DSP and on that driver if i recall the mid needs to be inverted relative to the hf section and for a 6300Hz L&R 4th order xover, try delaying the hf about .2 to .18 ms with the crossover as the two sources are not perfectly aligned in time at the origin. Fiddle with the frequency and delay a tiny bit to get as close to " one source" as you can, then apply the dsp as a final correction.
Re-phase is fun but at first limit your corrections to broad "things" and don't try to fix every sharp peak and dip as the higher you go, the more likely these are spatially variable things past the voice coil and radiator which are not fixable globally speaking,
Hope that helps
Tom
Post by: Mark Wilkinson on July 28, 2016, 03:53:45 PM
Hi Mark
Try getting it as close as possible first without DSP and on that driver if i recall the mid needs to be inverted relative to the hf section and for a 6300Hz L&R 4th order xover, try delaying the hf about .2 to .18 ms with the crossover as the two sources are not perfectly aligned in time at the origin. Fiddle with the frequency and delay a tiny bit to get as close to " one source" as you can, then apply the dsp as a final correction.
Re-phase is fun but at first limit your corrections to broad "things" and don't try to fix every sharp peak and dip as the higher you go, the more likely these are spatially variable things past the voice coil and radiator which are not fixable globally speaking,
Hope that helps
Tom
Thank you Tom,
Yes, that helps. It's pretty much how I've been going about it... it's nice to sense some verification that I'm at least in the hunt :)
Using either 4th or 8th order linear-phase LR, I get 0.08 ms delay needed on the VHF coax section, along with polarity inversion. I'm guessing the linear-phase crossover is what makes the difference between the time I get and your recollection. (Oh, I should have said I have the HE version... no idea if that effects timing.)
I hear your advice about broad strokes. It's easy to get tempted to do too much fine correction.... averages seem to be the cure, along with off-axis readings.
And I've kind of adopted the theory that when measurements don't change with corrections, I've run into stuff like the spatial variables you describe, stuff that needs a different cure.
Best, Mark
Post by: kevin kiefer on July 28, 2016, 05:35:14 PM
MLA is a cellular array, and shouldn't be listed next to line arrays on riders and specifications. If the project needs MLA, there are no equivalent products from other manufacturers. Not every project needs it, just like thousands of shows happened on stacked up conventional PA's after VDOSC came out. However, lumping MLA in with other line arrays just isn't accurate.
There are also many companies out there with millions of dollars invested in line arrays that have a very vested interest in convincing engineers and others that their product "does the same thing" as MLA. Or that "there's too much latency", "it doesn't rig as fast", or any of the other nonsense I hear on a daily basis.
MLA systems are not line arrays, just like KF850's are not line arrays. Therefore the comparisons have to be made in the proper context. MLA is not easy or simple. I've watched many seasoned professionals' eyes glaze over when you describe the technology. It sounds like science fiction and a lot simply reject it out of hand as "impossible". It's only "voodoo" when you don't understand it.
This is from a series of MLA videos Martin released.
https://www.youtube.com/watch?v=_cL96sSMzgg
Glad to help turn the lights on :-)
That's the frustrating part of frequently running into the uninformed opinion that MLA is "just using beam steering", or is "just a processed line array". It is neither, Cellular Drive is a unique approach and is now a patented process. It is not a form factor or a cabinet layout, it is a complete process that is different than anything that has been done before.
That still does not make it the right choice for everything or a perfect solution by any means but it is a very powerful tool to understand and have in your toolbar.
Lee
Post by: Yoel Farkas on July 28, 2016, 05:44:01 PM
There are also many companies out there with millions of dollars invested in line arrays that have a very vested interest in convincing engineers and others that their product "does the same thing" as MLA. Or that "there's too much latency", "it doesn't rig as fast", or any of the other nonsense I hear on a daily basis.What about the EAW Annya. is it another technology? i'm not a speaker designer, but as far i see it is even more sophisticated then the MLA.
https://www.youtube.com/watch?v=GxnE18mMMSk
https://www.youtube.com/watch?v=IBKHSsOW7mM
Post by: Cailen Waddell on July 28, 2016, 06:19:31 PM
Many of the uninformed detractors sound *a lot* like those people that derided the first line arrays when the whole world was still stacking up mountains of MT4's and KF850's. I got scoffed at many times when flying the PA needed to wait 15 minutes while I put venue dimensions in my newfangled laptop.
MLA is a cellular array, and shouldn't be listed next to line arrays on riders and specifications. If the project needs MLA, there are no equivalent products from other manufacturers. Not every project needs it, just like thousands of shows happened on stacked up conventional PA's after VDOSC came out. However, lumping MLA in with other line arrays just isn't accurate.
There are also many companies out there with millions of dollars invested in line arrays that have a very vested interest in convincing engineers and others that their product "does the same thing" as MLA. Or that "there's too much latency", "it doesn't rig as fast", or any of the other nonsense I hear on a daily basis.
MLA systems are not line arrays, just like KF850's are not line arrays. Therefore the comparisons have to be made in the proper context. MLA is not easy or simple. I've watched many seasoned professionals' eyes glaze over when you describe the technology. It sounds like science fiction and a lot simply reject it out of hand as "impossible". It's only "voodoo" when you don't understand it.
This is from a series of MLA videos Martin released.
https://www.youtube.com/watch?v=_cL96sSMzgg
Due respect - I've heard MLA, it sounds good, is a tool in the box. It doesn't sound better or worse than a properly deployed line array to me, but perhaps I am a cretin
. Regardless I have been unable to have anyone explain to me in sufficient detail, how it does what it does. It would appear at first blush to violate the laws of physics. Obviously that's impossible, and what I think it is doing is making a series of compromises, using some very smart computing to make the least shitty series of compromises it can. It has been suggested that in a previous thread that I really need to take a whole day class to properly understand it. That's not in the cards. But don't mistake my eyes glazing over for me not understanding. My eyes glaze over because I've watched the videos and read the material, and I simply don't have a good enough understanding of what is going on to understand how this array does not have the same problems a traditional line array has. So when there is fanboy'ing without discussion of the physics, I glaze over 
.Point source physics are a bit easier to understand and have been around a while so I can tolerate the fanboys when they are right because I understand what's going on under the hood.
Of course Martin is under no obligation to convince me and I'm not going to buy one so it doesn't really matter whether I get it. But I'm always going to think their claims are suspect until it's clear how/if they are able to do what the claim.
I think that's what the heart of this thread is, understanding the physics of what's going on. Unfortunately without the participation of some more company's engineers, we may not be able to have that level of discussion.
Sent from my iPhone using Tapatalk
Post by: Lee Buckalew on July 28, 2016, 10:05:39 PM
What about the EAW Annya. is it another technology? i'm not a speaker designer, but as far i see it is even more sophisticated then the MLA.
https://www.youtube.com/watch?v=GxnE18mMMSk
https://www.youtube.com/watch?v=IBKHSsOW7mM
It is absolutely another technology and Cellulat Drive is far more complex than Annya and Anna.
Lee
Post by: Scott Holtzman on July 29, 2016, 02:06:20 AM
Due respect - I've heard MLA, it sounds good, is a tool in the box. It doesn't sound better or worse than a properly deployed line array to me, but perhaps I am a cretin
Point source physics are a bit easier to understand and have been around a while so I can tolerate the fanboys when they are right because I understand what's going on under the hood.
Of course Martin is under no obligation to convince me and I'm not going to buy one so it doesn't really matter whether I get it. But I'm always going to think their claims are suspect until it's clear how/if they are able to do what the claim.
I think that's what the heart of this thread is, understanding the physics of what's going on. Unfortunately without the participation of some more company's engineers, we may not be able to have that level of discussion.
Sent from my iPhone using Tapatalk
Cailen, I don't claim to understand all of the physics but I have experience with flat panel radar arrays that shift a radar beam without having a moving driven element by altering the phase relationship between a bunch of emitters. Since I see more in common with the physics of RF and antenna design to acoustics than I do dissimilarities I have to think that it is a valid comparison of the underlying physics.
And the cool thing is if I am wrong I will be corrected and learn something in the process.
Post by: Keith Broughton on July 29, 2016, 08:29:48 AM
It is absolutely another technology and Cellular Drive is far more complex than Annya and Anna.But how much "different" can they really be?
Lee
I will admit to not knowing as much engineering as some members here but I did watch some of the product info on these systems.
Once the physical orientation of the drivers has been established, what adjustments can be made?
Basically, time/phase-EQ-level between sources are it, unless I am missing something. So the DSP and amplifier end of things are not all that revolutionary.
Of course, the apllication of the DSP modeling is quite important but it also looks like the pysical layout of components has become very critical to these types of systems.
The EAW info was more in depth and I grasp the concept better than with the limited info on Martin's Cellular Drive.
I do like the idea of not curving the array as EAW has done.
I understand that my comments only brush the surface of what is going on, and I am probably missing something here, but are these 2 systems really that radically different from each other? Drivers organized a specific way then application of time/phase-EQ and level.
Is it the DSP/ amplifier configuration or the physical layout of components that really set them apart?
Certainly different from the average "line array"
Standing by for more education :)
Post by: Merlijn van Veen on July 29, 2016, 09:16:00 AM
Verzonden vanaf mijn iPad met Tapatalk
Post by: Keith Broughton on July 29, 2016, 09:20:50 AM
Small side note on straight arrays. In general, any applied processing to a straight array radiates inherently symmetrical in the vertical plane. The directional properties of the individual enclosures and their transducers will determine the level of the frequencies that fire BOTH forward and rearward. Ultimately even really large speakers are rapidly becoming omnidirectional below 250 Hz which means that down tilting the frontal lobe by means of processing in general results in an equal amount of tilt and orientation in the rear lobe. Pointing that lobe and / or its rear wall reflected sound at the stage reducing GBF.A good point!
Verzonden vanaf mijn iPad met Tapatalk
I was discussing this with a systems tech that was using time offset to steer the low mid pattern and I asked about the unintended rearward lobes that may be created. He didn't seem to care or wasn't willing to measure it.
Could be a real issue in some cases!
Begs the question as to what is happening at the back of these "steered" arrays.
Post by: Merlijn van Veen on July 29, 2016, 11:53:52 AM
Could be a real issue in some cases!
Begs the question as to what is happening at the back of these "steered" arrays.
Yep, now you have a great sounding system which you might not be able to turn up because it starts to feed back so you start killing those frequencies killing part of the art in the process...
Verzonden vanaf mijn iPad met Tapatalk
Post by: Ivan Beaver on July 29, 2016, 12:05:47 PM
Small side note on straight arrays. In general, any applied processing to a straight array radiates inherently symmetrical in the vertical plane. The directional properties of the individual enclosures and their transducers will determine the level of the frequencies that fire BOTH forward and rearward. Ultimately even really large speakers are rapidly becoming omnidirectional below 250 Hz which means that down tilting the frontal lobe by means of processing in general results in an equal amount of tilt and orientation in the rear lobe. Pointing that lobe and / or its rear wall reflected sound at the stage reducing GBF.That is something that most don't understand
Verzonden vanaf mijn iPad met Tapatalk
If you take a normal line array or single source and tilt it down the rear lobe will go up in the air
When you take a steerable system an point the beam down the rear lobe will also go down at the angle that the front beams are steered
Post by: Stephen Kirby on July 29, 2016, 01:52:31 PM
Post by: Tom Danley on July 29, 2016, 02:20:54 PM
Having worked with some audio beam forming technologies I can understand the concept and how it would allow multiple drivers to correlate at some point in the audience. What I don't quite get, is how you get them to correlate at multiple points in the audience. I know MLA doesn't claim to use beam forming as in Anya and others. But how else do you deal with multiple arrivals at multiple points in space? Regardless of whether it's 10 feet in front of the speakers or 100 feet out in the audience.
Hi Steve
There are two separate ways to examine the issue you raised. The popular steady state model shows that the different distances add phase shift to each source according to the path length differences between one source and another. At 0 degrees difference and at every N times 360 degrees difference there is constructive addition while at every odd 180 degree phase difference there is cancelation or destructive interference. When plotted as a polar plot or spherical plot, this is the 3d view or manifestation of comb filtering where at a given frequency, one can move from a cancelation notch to an additive lobe. It is possible to use delay or phase manipulation so that there is the best combination of summation and cancelation at some point in front or combination of locations. This is how the sonar and radar arrays work and follow Huygens theorem.
The “other way” to look is not steady state but to look at a transient view.
Here there is no way to make a single transient arrive and sum coherently into one impulsive event when you’re talking about more than one set of path lengths because being at point A in front one has one set of path lengths to compensate while being at point B in front there are another set of different path lengths to compensate.
This is how one can have it sound good at the mix position but it’s very different everywhere else because there are different path lengths everywhere else.
You simply cannot fix this time of arrival or transient issue globally when you have spatially separated sources, only for one location at best. Much of music is more steady state but things like transients, voice intelligibility and music articulation requires preserving time more closely to replicate the input signal.
Hope that helps
Tom Danley
Post by: Cailen Waddell on July 29, 2016, 03:09:30 PM
Hi Steve
There are two separate ways to examine the issue you raised. The popular steady state model shows that the different distances add phase shift to each source according to the path length differences between one source and another. At 0 degrees difference and at every N times 360 degrees difference there is constructive addition while at every odd 180 degree phase difference there is cancelation or destructive interference. When plotted as a polar plot or spherical plot, this is the 3d view or manifestation of comb filtering where at a given frequency, one can move from a cancelation notch to an additive lobe. It is possible to use delay or phase manipulation so that there is the best combination of summation and cancelation at some point in front or combination of locations. This is how the sonar and radar arrays work and follow Huygens theorem.
The “other way” to look is not steady state but to look at a transient view.
Here there is no way to make a single transient arrive and sum coherently into one impulsive event when you’re talking about more than one set of path lengths because being at point A in front one has one set of path lengths to compensate while being at point B in front there are another set of different path lengths to compensate.
This is how one can have it sound good at the mix position but it’s very different everywhere else because there are different path lengths everywhere else.
You simply cannot fix this time of arrival or transient issue globally when you have spatially separated sources, only for one location at best. Much of music is more steady state but things like transients, voice intelligibility and music articulation requires preserving time more closely to replicate the input signal.
Hope that helps
Tom Danley
This is the understanding of physics I have.
The voodoo I don't understand is how a product like MLA can claim phase coherency with the multiple arrival points of multi cellular point source - or whatever it is called.
Tom - I'm not asking you to explain it, simply saying that I understand the physics the way you have explained them, but can not align the marketing copy of MLA with those physics. I assume that perhaps I don't understand enough yet :). Of course there is also the possibility that I understand the physics correctly...
Post by: Lee Buckalew on July 29, 2016, 03:26:44 PM
But how much "different" can they really be?
I will admit to not knowing as much engineering as some members here but I did watch some of the product info on these systems.
Once the physical orientation of the drivers has been established, what adjustments can be made?
Basically, time/phase-EQ-level between sources are it, unless I am missing something. So the DSP and amplifier end of things are not all that revolutionary.
The most significant difference in the Cellular Drive process, as I am understanding it currently, and others is the modeling, both the models used to describe each individual cabinet and the model of the coverage plane.
Cellular Drive starts with a virtual model of the room created in section,
you then set the coverage requirements and goals and the software calculates the required acoustic source that would be necessary to create the user defined coverage/goals.
The software then uses the individual cabinet balloon data based on the relative position of each cabinet in the array, tests every combination of inter-cabinet angle to determine what interactions come the closest to creating the previously defined target result. The software then utilizes elemental equalization to configure each cell to behave, at the listening plane, as required to create the previously defined acoustic model.
Martin is not modeling to create specific interactions at the cabinet/driver and assuming/expecting that result to create a specific coverage in the room (because it doesn't) they are creating a model that creates a known result along the listening plane by making the individual cells within the array function together as a CDPS. They don't care what is happening at the driver or cabinet other than for the initial information so that the model can be accurate. The drivers/cabinets must be designed from the beginning with the physical characteristics required to allow the software to create the required interactions.
Of course, the apllication of the DSP modeling is quite important but it also looks like the pysical layout of components has become very critical to these types of systems.
Physical layout of components is critical as is the actual performance of the drive components.
The EAW info was more in depth and I grasp the concept better than with the limited info on Martin's Cellular Drive.
I would doubt that you will see truly in depth engineering info on the Cellular Drive process since it is patented.
Others would have to answer that as I am not in the know. :D
I do like the idea of not curving the array as EAW has done.
A great deal of initial testing was done with O-Line during early development of the Cellular Drive concept. There is far too much to go into on a web forum but part of that testing showed the limitations of a flat hung array vs. curved in terms of coherence and in terms of the quantity of individual drive cells required. I do not know enough about what EAW is doing to discuss their processing implementation.
I understand that my comments only brush the surface of what is going on, and I am probably missing something here, but are these 2 systems really that radically different from each other? Drivers organized a specific way then application of time/phase-EQ and level.
Is it the DSP/ amplifier configuration or the physical layout of components that really set them apart?
Certainly different from the average "line array"
Standing by for more education :)
The approach of the two systems, the starting point itself, is completely different
One very important point is that no system can be utilized to do anything asked of it. There are some obvious limitations to cellular drive (and some not so obvious) among them the requirement to have enough cells to create the required interactions.
Lee
Post by: Keith Broughton on July 29, 2016, 03:51:31 PM
among them the requirement to have enough cells to create the required interactions.Thanks for that comprehensive response Lee. Appreciated :)
Lee
Your quoted test is a very common fault in most of the "conventional" line arrays I see deployed.
Not enough cabinets and not deployed in the correct physical space to get good results.
Post by: Lee Buckalew on July 29, 2016, 04:05:27 PM
Due respect - I've heard MLA, it sounds good, is a tool in the box. It doesn't sound better or worse than a properly deployed line array to me, but perhaps I am a cretin
To my ears and testing, every line array that I have ever worked with has had some significant comb-filtering issues. Some are far better than others but all have them. Much of what we perceive at the highest frequencies as horn throat distortion (that really nasty sound of the EV Manifold Technology boxes that people equated with them "sounding" so loud) is really comb-filtering interactions creating the distortion.
These interactions are not present with any properly laid out MLA system that I have used. That said, there are certainly still comb-filter interactions between adjacent arrays such as mains and out-fill.
That said there are a number of great sounding line array solutions out there.
It has been suggested that in a previous thread that I really need to take a whole day class to properly understand it. That's not in the cards.
Then it would seem that you don't care enough about it to want to learn more in depth. The suggestion was not for 1 day it was to attend an MLA training event which is 3 days. That is a starting point.
Suggesting that it is not worth the time to learn more about it in depth but then spending time discussing it is like making comments comparing SMAART and SYSTUNE but being unwilling to take the time to take a training class on those systems. Like learning SMAART or SYSTUNE the training on an MLA system only begins with the classes,
But don't mistake my eyes glazing over for me not understanding. My eyes glaze over because I've watched the videos and read the material, and I simply don't have a good enough understanding of what is going on to understand how this array does not have the same problems a traditional line array has.
And yet spending time in a class where you can learn about this directly form the factory people involved and ask direct questions is "not in the cards". ???
Point source physics are a bit easier to understand and have been around a while so I can tolerate the fanboys when they are right because I understand what's going on under the hood.
And of course point source physics don't actually work when you try to apply them to actual speaker cabinets. Danley is using physics to allow multiple drivers to behave as a single source from a given horn. This is not the same thing as the physics model of a point source.
Utilizing multiple point source models results in the same interactions that are discussed with Cellular Drive and MLA. With 2 modeled point sources and the smallest of spaces between them you begin to alter the balloon data that is generated. If you take it beyond theory and actually place that point source into a cabinet you create interactions of the source to the cabinet (reflection, diffraction, etc.).
Of course Martin is under no obligation to convince me and I'm not going to buy one so it doesn't really matter whether I get it. But I'm always going to think their claims are suspect until it's clear how/if they are able to do what the claim.
I think that's what the heart of this thread is, understanding the physics of what's going on. Unfortunately without the participation of some more company's engineers, we may not be able to have that level of discussion.
Sent from my iPhone using Tapatalk
I doubt that there will be any in depth physics discussions of the actual process as it is intellectual property protected by a patent.
I don't know if there are any factory people from Martin who would/could participate. I don't know what would be allowed by PSW and what would be allowed by Martin or Loud.
Post by: Russell Ault on July 29, 2016, 04:34:26 PM
I doubt that there will be any in depth physics discussions of the actual process as it is intellectual property protected by a patent.
Actually, that's the real joy of patents: in return for legal protection, you have to immediately tell the whole world "how you did it" (with enough detail that someone else can do it too when the patent expires).
For example: US 20140348355 - Speaker Configuration (http://www.google.com/patents/US20140348355).
-Russ
Post by: Stephen Kirby on July 29, 2016, 04:56:21 PM
Hi SteveThank you Tom. This is pretty much elaborating on what I was asking. Getting the impulse to be uniform over a large area. Maybe being a musician, this is important to me.
There are two separate ways to examine the issue you raised. The popular steady state model shows that the different distances add phase shift to each source according to the path length differences between one source and another. At 0 degrees difference and at every N times 360 degrees difference there is constructive addition while at every odd 180 degree phase difference there is cancelation or destructive interference. When plotted as a polar plot or spherical plot, this is the 3d view or manifestation of comb filtering where at a given frequency, one can move from a cancelation notch to an additive lobe. It is possible to use delay or phase manipulation so that there is the best combination of summation and cancelation at some point in front or combination of locations. This is how the sonar and radar arrays work and follow Huygens theorem.
The “other way” to look is not steady state but to look at a transient view.
Here there is no way to make a single transient arrive and sum coherently into one impulsive event when you’re talking about more than one set of path lengths because being at point A in front one has one set of path lengths to compensate while being at point B in front there are another set of different path lengths to compensate.
This is how one can have it sound good at the mix position but it’s very different everywhere else because there are different path lengths everywhere else.
You simply cannot fix this time of arrival or transient issue globally when you have spatially separated sources, only for one location at best. Much of music is more steady state but things like transients, voice intelligibility and music articulation requires preserving time more closely to replicate the input signal.
Hope that helps
Tom Danley
I know that audiophiles are derided here, but in that world temporal coherence is often referred to as "pace". Meaning that the leading transients of all the overtones are aligned in time. One doesn't hear the upper overtones of a bass drum separated from the fundamental. The more accurate this is, the more the groove or pace of the music is maintained. Probably why I tend to prefer panel speakers for home listening, although they are obviously not scalable for large area/high SPL SR.
Getting steady state sine waves aligned within the 120* that Merlijn refers to as coupling without significant cancellation is one thing. It gets a bit tougher when the waveforms have steeper rise times (as most instruments other than a flute do) thus involving higher bandwidth or alignment to higher frequencies. But getting the initial transient of everything coherent in time is tougher. How you get it coherent in time at multiple points in space with multiple path lengths is the part I can't get. Maybe this isn't happening, and getting phase coherence by controlling addition and cancellation over an area is a sufficient improvement that it markedly improves sound quality.
Post by: Roland Clarke on July 29, 2016, 05:36:52 PM
Unless I'm getting it completely wrong what it boils down to is that they are doing what most other manufacturers of line array are doing, but taking it down to a driver by driver basis rather than a cab by cab. To suggest that it's not a line array is, for me, to ignore the fact I doubt it can create a completely coherent wave front any more than any line array can genuinely create a genuine curved wavefront.
Post by: Stephen Kirby on July 29, 2016, 05:54:42 PM
I used to be involved in holographic optics. These are done by creating an interference pattern at a point in space where you have placed a recording media. It takes a lot of wavefront manipulation to get a particular interference pattern that does what you want. In that case it was a 3 dimensional sort of Fresnel lens based on the change in refraction index in the exposed regions compared to the unexposed regions. This is probably the closest analog to the MLA that I can think of.
Post by: Josh Millward on July 29, 2016, 06:29:12 PM
...they are creating a model that creates a known result along the listening plane by making the individual cells within the array function together as a CDPS. They don't care what is happening at the driver or cabinet other than for the initial information so that the model can be accurate.
So they are only modeling what is happening at the listening plane and manipulating the DSP to make it happen.
What about the rest of the room?
What is coming off the sides, top, bottom, and backside of the array?
I'm asking because I don't know. I've never heard it... though I am intrigued by it. I'd like to learn more about it.
Post by: Lee Buckalew on July 29, 2016, 06:41:03 PM
So they are only modeling what is happening at the listening plane and manipulating the DSP to make it happen.
What about the rest of the room?
What is coming off the sides, top, bottom, and backside of the array?
I'm asking because I don't know. I've never heard it... though I am intrigued by it. I'd like to learn more about it.
Cellular Drive in Display 2.1.10 is modeling the surface planes of the whole room as entered by the user. It is only doing that right now in a section but the models used are fully 3 dimensional and I am told that the software can be as well but that is somewhere down the line.
The model does show what is happening from the back, top, bottom, etc. but at this point it does not include a model of the sides.
Lee
Post by: Lee Buckalew on July 29, 2016, 06:42:32 PM
Actually, that's the real joy of patents: in return for legal protection, you have to immediately tell the whole world "how you did it" (with enough detail that someone else can do it too when the patent expires).
For example: US 20140348355 - Speaker Configuration (http://www.google.com/patents/US20140348355).
-Russ
Patents are outside of my normal realm so, there you go. I guess those interested should be able to look up the specifics of what is patented.
Lee
Post by: Lee Buckalew on July 29, 2016, 06:48:16 PM
The only thing that makes me at all nervous about the whole "multicellular" technology is the lack of candour that appears about it. we all know that even with the best designed horn technology there are limits to off axis response control, that clever manipulation of time, phase, level and frequency response can assist in controlling this still isn't the magic bullet.
Unless I'm getting it completely wrong what it boils down to is that they are doing what most other manufacturers of line array are doing, but taking it down to a driver by driver basis rather than a cab by cab. To suggest that it's not a line array is, for me, to ignore the fact I doubt it can create a completely coherent wave front any more than any line array can genuinely create a genuine curved wavefront.
It is not about a coherent wavefront. It is about altering the output of every cell such that when each individual cell combines with the other cells of the same band pass that cover the same area of the listening plane they are phase and frequency coherent and the edges of their coverage is phase and frequency coherent with the next covered area, and on and on.
Lee
Post by: Russell Ault on July 29, 2016, 06:51:41 PM
US 20140348355 - Speaker Configuration (http://www.google.com/patents/US20140348355).
Patents are outside of my normal realm so, there you go. I guess those interested should be able to look up the specifics of what is patented.
Sorry, I should have been clearer: despite its generic title, I believe what I linked to is the patent for the technology behind MLA for anyone interested in having a look. The description certainly contains a lot of math beyond my ability!
-Russ
Post by: Lee Buckalew on July 29, 2016, 10:27:29 PM
Sorry, I should have been clearer: despite its generic title, I believe what I linked to is the patent for the technology behind MLA for anyone interested in having a look. The description certainly contains a lot of math beyond my ability!
-Russ
Sorry about that, I did not check the link, I was in a hurry to get to an event with my wife and kids.
Lee
Post by: Scott Holtzman on July 30, 2016, 03:08:15 AM
Yep, now you have a great sounding system which you might not be able to turn up because it starts to feed back so you start killing those frequencies killing part of the art in the process...
Verzonden vanaf mijn iPad met Tapatalk
That exactly described by night. We about 20' off a brick wall that formed the back of an aquarium building underneath a dome. The array was short so I should not have expected much but exactly what you describe, the rear lobe built up and bounced off the wall. Could not use all the gain and hacking up the EQ was ugly.
Post by: Roland Clarke on July 30, 2016, 03:51:01 AM
It is not about a coherent wavefront. It is about altering the output of every cell such that when each individual cell combines with the other cells of the same band pass that cover the same area of the listening plane they are phase and frequency coherent and the edges of their coverage is phase and frequency coherent with the next covered area, and on and on.
Lee
I fully understand that the calculation isn't aimed at making the sound good at the source, more at the audience position, however, the point I was making is that all speakers (cells as they are being described here), exhibit unevenness in their off axis performance, both in frequency and of course phase. As this is the smallest level of adjustment possible, the processing must by definition be a "fudge". I'm not disputing that it might be the best possible result available in real terms, more the marketing blurb that suggests it is better than this.
Post by: Merlijn van Veen on July 30, 2016, 09:08:09 AM
Actually, that's the real joy of patents: in return for legal protection, you have to immediately tell the whole world "how you did it" (with enough detail that someone else can do it too when the patent expires).
For example: US 20140348355 - Speaker Configuration (http://www.google.com/patents/US20140348355).
-Russ
I read the patent which is quite informatieve and inspiring. However it only reveals one aspect of the "magic" behind MLA.
Basically it describes a generic optimization routine that iteratively minimizes variance in SPL, frequency response, leakage, rate of change in SPL over distance and rate of change in SPL over frequency with respect to predefined parameters.
With respect, but not limited, to MLA, FWAICT it's used in the first round of calculations in the Display software to determine the splay angles. You could think of it as Martin's proprietary "autosplay" function. But that's just one particular application of the routine.
I imagine that the same routine, again not limited to MLA, could be used to distill the MLA processing parameters, but that's not being mentioned in the patent.
Verzonden vanaf mijn iPad met Tapatalk
Post by: Ivan Beaver on July 30, 2016, 11:07:50 AM
But unless they are VERY LARGE-that simply cannot happen.
And if the patterns are narrow (as they would need to be in a many cellular system), then they have to be even larger than a wider coverage pattern horn.
Since they are not large enough to cover a specific area, and therefore "spray" over the coverage areas of other devices that are physically in different places, the phase MUST be different at different seats.
Once the parameters have been set-when you have 2 different devices at physically different places, the response at different seats has to be different. How much? Does it matter? It depends.
There are 2 ways to "control the pattern" of a sound wave.
One is by the use of large horns and the other is electronic manipulation/cancellation/interference of the signal.
There is no way to preserve the original signal when it has somehow been "changed" electronically. Once it has been changed it is NOT the original signal anymore.
Once you start to have signals interfere-the sound quality must go down.
But sometimes the pattern control provided is more important than the sound quality. Just as with any directional sub configuration.
The sound quality will go down on axis, but the overall result may be better. Sometimes this is by not having people behind the subs complain.
Sometimes it is to steer a sideways lobe on the stage so the bass is reduced on stage.
Post by: Lee Buckalew on August 02, 2016, 11:34:15 AM
I read the patent which is quite informatieve and inspiring. However it only reveals one aspect of the "magic" behind MLA.
Basically it describes a generic optimization routine that iteratively minimizes variance in SPL, frequency response, leakage, rate of change in SPL over distance and rate of change in SPL over frequency with respect to predefined parameters.
With respect, but not limited, to MLA, FWAICT it's used in the first round of calculations in the Display software to determine the splay angles. You could think of it as Martin's proprietary "autosplay" function. But that's just one particular application of the routine.
I imagine that the same routine, again not limited to MLA, could be used to distill the MLA processing parameters, but that's not being mentioned in the patent.
Verzonden vanaf mijn iPad met Tapatalk
I have just been having the time to begin looking at this in greater depth.
The descriptions actually do get into calculating the actual required position of each speaker unit or, conversely, utilizing a constraint of known speaker unit to speaker unit orientation, calculating required amplitude, delay, and phase corrections to electronically create a shift in speaker to speaker orientation always attempting to create an optimized frequency response and SPL contour based upon the user defined parametres. This is discussed relative to both a 2D and 3D plot and to speaker elements constrained to a "Line Array" form-factor as well as other form-factors.
Lee
Post by: Keith Broughton on August 02, 2016, 11:54:42 AM
utilizing a constraint of known speaker unit to speaker unit orientation, calculating required amplitude, delay, and phase corrections to electronically create a shift in speaker to speaker orientation always attempting to create an optimized frequency response and SPL contour based upon the user defined parametres.However, considering the mechanical pattern control of the drivers in the "cell" and the limited adjustment of the array configuration, "optimized" doesn't mean "correct".
I still don't see how this kind of system can offer even frequency response, coherent phase response and even SPL over the entire audience coverage area.
There must be compromises somewhere.
Not saying it doesn't sound good, just pushing back against the advertised claims.
Need more info.
Post by: Lee Buckalew on August 02, 2016, 12:47:15 PM
However, considering the mechanical pattern control of the drivers in the "cell" and the limited adjustment of the array configuration, "optimized" doesn't mean "correct".
I still don't see how this kind of system can offer even frequency response, coherent phase response and even SPL over the entire audience coverage area.
There must be compromises somewhere.
Not saying it doesn't sound good, just pushing back against the advertised claims.
Need more info.
As I am understanding it there is not only mechanical pattern control, there is also electronic pattern control since each HF and MF cell is made up of multiple drivers that interact vertically to create vertical pattern constraint. These are further electronically controlled to change the pattern within the individual speaker cabinet and then from cabinet to cabinet. Each cell and each group of cells is not built to create a flat wavefront but to create a wavefront that propagates in a manner that will allow the control capability of the individual cells to interact properly.
There is compromise in the weighting of SPL profile (Target), vs. the Non-Audience area profile (Leakage), vs. the Exclusion profile (Hard Avoid). These compromises are user defined and are also affected by the number of cells available to provide control. The results, based upon the capability of the number of cells present and the user defined coverage requirements, are created in Display and show the actual ability or lack of ability to meet the desired target(s).
I can't say exactly what is being done (as in, I don't fully know) but I can tell you that the measurements hold up to the claims. The claim being that the predicted outcome calculated in Display will be achieved within +/- 0.5dB SPL.
Lee
Post by: Stephen Kirby on August 02, 2016, 01:32:23 PM
Which brings up the question for folks like Lee who have experience with the system. How does it match up in SPL for similar sized and powered systems that don't invoke as much cancellation? Granted, even and controlled SPL is more desirable (IMHO) than ultimate SPL, but it would seem that energy is being expended in creating the coverage effect.
Post by: Lee Buckalew on August 02, 2016, 01:54:54 PM
Since there is no way to have multiple path lengths arrive at multiple audience positions coherently, there has to be some very sophisticated destructive interference (cancelation) going on to minimize the effect over as much of the audience area as possible along with the ability to provide higher levels of cancellation in "leakage" and "hard avoid" areas. So you have a "virtual multi-cell" system rather than an actual multi-cell system of full bandwidth narrow pattern cells that can be adjusted for desired coverage.
Which brings up the question for folks like Lee who have experience with the system. How does it match up in SPL for similar sized and powered systems that don't invoke as much cancellation? Granted, even and controlled SPL is more desirable (IMHO) than ultimate SPL, but it would seem that energy is being expended in creating the coverage effect.
Overall SPL is certainly one of the variables. Since creating the correct coverage across the defined audience plane does limit coverage in the leakage area and the hard avoid area the question remaining is, how much additional cancellation needs to be created and what resources are available, both physical and electronic. Just as with any system, correct configuration for the wanted result is important.
There is no short answer for "how does it compare" for SPL capability, etc. as it depends on what the user is asking the system to accomplish.
In actual use my experience is that the system can achieve very similar output to other similarly sized systems even when providing a significant amount of leakage and hard avoid control because the interactions created in the audience area are combining more effectively than in a line array configuration. This is what is allowing such long distance coverage capabilities without delays.
Lee
Post by: Franz Francis on August 02, 2016, 02:00:51 PM
Since Lee has the most experience with the system my question, how does the mid-hi frequency behave under windy conditions in a properly deployed MLA rig?.
It's an indication of how good or better the alignment is with a multi box line array. Some designs are better behaved in a windy environments than others.
Franz
Post by: Lee Buckalew on August 02, 2016, 02:08:21 PM
Interesting thread.
Since Lee has the most experience with the system my question, how does the mid-hi frequency behave under windy conditions in a properly deployed MLA rig?.
It's an indication of how good or better the alignment is with a multi box line array. Some designs are better behaved in a windy environments than others.
Franz
Perhaps I have the most experience of those contributing here but there are many others with much more experience than me.
In windy conditions there is little to no phase shift/comb-filtering experienced. Obviously this can be deployment dependent since outfills or front/lip fills may affect this.
Lee
Post by: Yoel Farkas on August 02, 2016, 02:33:55 PM
Perhaps I have the most experience of those contributing here but there are many others with much more experience than me.i (still) didn't figure out how it works. but as far i understand it has to take in account the temperature, as it change the speed of sound and the wavelengths.
In windy conditions there is little to no phase shift/comb-filtering experienced. Obviously this can be deployment dependent since outfills or front/lip fills may affect this.
Lee
Post by: Lee Buckalew on August 02, 2016, 02:44:03 PM
i (still) didn't figure out how it works. but as far i understand it has to take in account the temperature, as it change the speed of sound and the wavelengths.
No, this is not varying with the temperature or wind to compensate, it is coherent enough that even though the wind does shift the sound (physics) as it shifts you are not hearing a bunch of different phase shifted sources hit you at different times. All the wind does is rapidly move the coverage past you rather than have you rapidly move through the coverage. If you are being presented with multiple path lengths that are not phase coherent you will hear it quite distinctly.
Lee