Modeling my pet peeve

[Jay Barracato]

You may have noticed that I really hate the idea of trying to increase output by doubling MI type boxes. So I did some modeling to get used to the G.P.A program. In my mind, the comb filtering that results is just not acceptable, not due to the narrow spikes at high frequency, but due to the wider big hit some frequencies in the middle of the vocal range take.



Here are my results for both 2 60 degree boxes per side, and for 2 90 degree boxes per side, set up spaced the way many bands would set them up, based on information from their GC salesman.



60 degree

90 degree

 

[John Roberts]

Re: Modeling my pet peeve

Another potential difference between MI and more professional boxes is that the allowed production variance between drivers is larger. In a single box, X amount of driver variance is not that audible, but they you array them together, errors between the sundry boxes become more problematic.



I recall the difficulty in trying to explain to PV dealers why some more professional PV boxes cost more, they were sure we were overcharging them (we weren't).



JR

 

[Dick Rees]

Re: Modeling my pet peeve

Jay....



Does the modeling program allow you to splay the boxes or is it just a side by side config???



DR

 

[Jay Barracato]

Re: Modeling my pet peeve

Yes it allows you to splay the boxes through a full circle. What seems to be missing from it is walls and ceilings although the developer is working on three-d.



This is the freeware program that Bennett is using.



http://gpa.hms2k.cl/



I set up this model looking for a worse case scenerio.

 

[Jay Barracato]

Re: Modeling my pet peeve

Here are the same 90- degree boxes with 10 degrees, 30 degrees and 60 degrees of splay.



It is neat to work through the angles and watch the combing change location/frequency.





10 degrees

30 degrees

60 degrees

 

[Justice C. Bigler]

Re: Modeling my pet peeve

Maybe I missed it somewhere, but did you mention what speakers these were? Wouldn't any speakers perform the same when set up similarly? And if not, how to you input which speakers you are dealing with?



What part of the manufacturing process is it that controls the amount of of comb filtering when using multiple speakers? And how is that different for MI speakers versus pro audio speakers?

 

[Jay Barracato]

Re: Modeling my pet peeve

These are just a generic 90 degree box. My point is that since 90 degree boxes are not designed for arraying, just about any way you arrange 2 of them per side, the combing is going to seriously affect the sound.



Once you get down to 60 degree boxes, a trap array starts to become a little more reasonable, but here pattern control really sets apart pro boxes from MI. An arrayable Pro trap shows its pattern control more consistantly over a wider range of frequencies. And that pattern control comes at a cost.

 

[Greg Cameron]

Re: Modeling my pet peeve

Once you get down to 60 degree boxes, a trap array starts to become a little more reasonable, but here pattern control really sets apart pro boxes from MI. An arrayable Pro trap shows its pattern control more consistantly over a wider range of frequencies. And that pattern control comes at a cost.

This is one major area line arrays excel at which has been mentioned in various articles - lack horizontal combing in the array since there aren't boxes sitting side by side. Of course, there will be some combing the vertical plane where are ears are least sensitive to it. But even that is minimal due to the tight vertical patterns on the wave guides. And box to box coupling of the low and mid drivers is decent due to the close proximity of the drivers.



Greg

 

[Ivan Beaver]

Re: Modeling my pet peeve-Arrayability concepts

What part of the manufacturing process is it that controls the amount of of comb filtering when using multiple speakers? And how is that different for MI speakers versus pro audio speakers?

There are several aspects that determine how well particular loudspeakers will array. I will not go into detail-as that is pretty much a paper and a bit long for a forum post.



A couple of ''bullet points''.



Horn size. If the horns used are not large enough to provide pattern control down to a low enough freq so the drivers are close enougn together so they will sum without interference.



Closeness of drivers in array configuration. When the cabinets are arrayed, the drivers need to be physically close together. In the perfect world the drivers would occupy the same physical space. But we can't do that-so getting them as close together as possible is the idea. The HF driver should be just short of touching the sides of the cabinet.



The sides of the cabinets should be cut at 1/2 of the rated coverage. So a 60° cabinet should have the side cut at 30° each. If it is cut at less than than-then the HF driver will be far from the sides of the cabinet-and far away (in terms of wavelength) to the other cabinet.



And of course the better the drivers inside a particular cabinet are aligned in time (this is not always physical and the phase response of the crossover has to be taken into account when considering alignment) helps to keep everything ''working together''.



Here is a link that helps to describe some of those properties.



The discussion about arraying is the 2nd part of the paper.



http://www.danleysoundlabs.com/pdf/danley_tapped.pdf



If the sides are not cut at that angle, then there i

 

[Jeffrey Knorr]

Re: Modeling my pet peeve

Once you get down to 60 degree boxes, a trap array starts to become a little more reasonable, but here pattern control really sets apart pro boxes from MI. An arrayable Pro trap shows its pattern control more consistantly over a wider range of frequencies. And that pattern control comes at a cost.

This is one major area line arrays excel at which has been mentioned in various articles - lack horizontal combing in the array since there aren't boxes sitting side by side. Of course, there will be some combing the vertical plane where are ears are least sensitive to it. But even that is minimal due to the tight vertical patterns on the wave guides. And box to box coupling of the low and mid drivers is decent due to the close proximity of the drivers.



Greg

Hi Greg,



This certainly isn't the best line array out there, but I found this file particularly interesting and eye opening. I would wager that most ''line arrays'' being sold don't come anywhere close to the theoretical ideal (especially at higher frequencies).



Jeff

 

[Jay Barracato]

Re: Modeling my pet peeve

I was really just playing with the modeling software. I don't think I was demonstrating anything anyone here is not familiar with. But I do think the pictures will be helpful next time I get caught in the 4 page discussion of ''I need more output so I am going to buy more...'' on the other boards.



I still think there are many rooms that can be well served by a well thought out trap rig. Actually, the pictures that Greg posted of his hang is probably a good example because you can clearly see each box is providing coverage for different areas.



I have absolutely no real scientific evidence but my gut feeling is that some of the higher frequency combing is actually part of what we experience as stereo (different time arrivals at each ear). Once again, what I was really trying to show was the whopping 30+ db of cancellation right in the middle of the guts of the sound (1000-2000 hz). Many of the cancellation areas are a couple of hundred of hz wide. Moving the mic in the model is a neat experiment that really shows you can move the cancellation around, in both location and frequency, but you can't get ride of it.

 

[Ivan Beaver]

Re: Modeling my pet peeve

Once you get down to 60 degree boxes, a trap array starts to become a little more reasonable, but here pattern control really sets apart pro boxes from MI. An arrayable Pro trap shows its pattern control more consistantly over a wider range of frequencies. And that pattern control comes at a cost.

This is one major area line arrays excel at which has been mentioned in various articles - lack horizontal combing in the array since there aren't boxes sitting side by side. <font color="red">Of course, there will be some combing the vertical plane where are ears are least sensitive to it</font>. But even that is minimal due to the tight vertical patterns on the wave guides. And box to box coupling of the low and mid drivers is decent due to the close proximity of the drivers.



Greg

That is one area that is often stated-but is basically flawed.



True- people do not usually go up and down in the coverage pattern (wihtout a ladder or being able to levitate

)-HOWEVER- they do go front to back

Which puts them in different parts of the coverage pattern.



And more people tend to walk from front to back and vice versa, than from side to side.



If you look at the ''rows'' of coverage ripples with a line array-you will hear those as you walk from back to front.



Just food for thought

 

[Bennett Prescott]

Re: Modeling my pet peeve

Jay,



Just be careful, that looks really ugly (as do most predictions) but it won't sound anywhere near as bad as it looks. Our brains tend to ignore standing frequency response errors, especially if one ear hears them and the other is not. Cover one ear and you will be able to hear it. All HF devices look pretty nasty in the upper part of their range, but it doesn't bother you because your brain filters it out.

 

[Jay Barracato]

Re: Modeling my pet peeve

Bennett,



I am not focusing on the upper part of the HF, that is the same as what I was saying our brain perceives as part of the stereo image (others say we filter it out). What I am focused on is the bottommost notch where the combing gets wider and farther apart. Usually in practice I see it somewhere between 1000 and 2000 hz, and I find it to be very noticable as you move around the room. For an acoustic string band, there is a lot of meat to the music in that range, and we are not covering up with either volume or other sources like amps on stage.





edit- I should say I dislike the practice not due to the model but due to experience. Too many of my shows have been in smaller venues with a bunch of GC boxes thrown together. A good sounding mandolin is great for making this stand out because as you move slowly across the room the ''wood'' will be there, and then it will be gone again

 

[Greg Cameron]

Re: Modeling my pet peeve

Indeed. It's not a perfect solution. But they do work pretty good. Until someone designs a loudspeaker that has infinite output that's about the size of a golf ball, it's all a compromise

Greg

 

[Greg Cameron]

Re: Modeling my pet peeve

This certainly isn't the best line array out there, but I found this file particularly interesting and eye opening. I would wager that most ''line arrays'' being sold don't come anywhere close to the theoretical ideal (especially at higher frequencies).

Yeah, the SLS is certainly not on par with the upper echelon line sources. That said, it's absolutely true the the HF waveguides from box to box aren't going to couple all that well. Hence the goal of tightening up coverage for minimal overlap whereas the lower frequency drivers will couple so long as they're within the fraction wavelength distance for summation. L'Acoustics came out of the gate with a very good design for their HF waveguide that a lot of the other guys have been trying to replicate the behavior of without infringing on patents. But there are some really good designs out their to minimize interference these days. And at the top of the array in a properly deployed system is where most of that interference is rather than at the bottom for the 'near-field' listeners where the boxes have more vertical splay. I've been very impressed with the latest systems, though I'm still impressed by well designed & deployed trap systems.



Greg

 

[RYAN LOUDMUSIC JENKINS]

Re: Modeling my pet peeve

I don't think a ''Generic'' box in the prediction with no walls or ceiling really does a lot for us other than making the neat pattern on the computer screen. Keep in mind that you can have a pile of different boxes that are supposedly 90 degree boxes but they do not have the same coverage pattern. You really need to look at the polar plots for the boxes to get an idea of what frequencies are truely close to the specifications.



The was a very cool post that Ivan did a couple years ago for Bob Leonard on the PSW forums that showed more of a balloon prediction based upon measured responses of actual JBL boxes. I would hurry and find the post just in case it gets lost in the archives. It was a great thread and very eye opening!

 

[Ivan Beaver]

Re: Modeling my pet peeve

I don't think a ''Generic'' box in the prediction with no walls or ceiling really does a lot for us other than making the neat pattern on the computer screen. Keep in mind that you can have a pile of different boxes that are supposedly 90 degree boxes but they do not have the same coverage pattern. You really need to look at the polar plots for the boxes to get an idea of what frequencies are truely close to the specifications.



The was a very cool post that Ivan did a couple years ago for Bob Leonard on the PSW forums that showed more of a balloon prediction based upon measured responses of actual JBL boxes. I would hurry and find the post just in case it gets lost in the archives. It was a great thread and very eye opening!

Here is part of a slide that I presented at one of the Boston get to gethers on pattern control.



This is an actual product (who's manufacturer shall remain nameless) and their measured/published data. The solid line is the rated pattern. Notice that it is only that pattern over a small freq range. Higher or lower and it is a good bit different.



So as usual you cannot describe a complicated question (such as pattern control) with a simple answer. You have to ask ''at what freq''? Because the answer will be different.



And when you ''array'' the boxes, you will end up with overlap and underlap-depending on the freq you are interested in.

 

[RYAN LOUDMUSIC JENKINS]

Re: Modeling my pet peeve

This is an actual product (who's manufacturer shall remain nameless) and their measured/published data. The solid line is the rated pattern. Notice that it is only that pattern over a small freq range. Higher or lower and it is a good bit different.



So as usual you cannot describe a complicated question (such as pattern control) with a simple answer. You have to ask ''at what freq''? Because the answer will be different.



And when you ''array'' the boxes, you will end up with overlap and underlap-depending on the freq you are interested in.

It is charts like this that show a lot more good information than a spec sheet. My mains are rated at 60 degrees. I have experimented and splayed them at many different angles and the angle that sound best it not even close to 60 degrees. I haven't put a protractor on them while they are setup but I think I get a lot closer to 120 degrees out of three per side then I do 180 degrees.



On the other hand, I have a bunch of powered speakers that are rated at 90 degrees and I need to splay them as such to get them to sound right if I need two per side. I have to do this frequently to cover wide areas and they work quite well at it but if I try to narrow the splay, they do not sound right.