Full range line array for wall or corner placement

Thanks. I hope you are right.

Also hope that multiple subs will neutralize effect of that 60 hz null.

Now here is boundary contribution prediction by Vituix for my array location in front of the garage door (green trace in middle graph):

attachment.php


This showing a 71 hz null and another at 220. It does predict the hump in between that we see. 6 db effective absorption is a guess. There should be at least that much at 200 Hz, declining rapidly as frequency decreases.
 

Attachments

  • boundary interference prediction vituix.jpg
    boundary interference prediction vituix.jpg
    223 KB · Views: 269
Here is the boundary prediction for the array tight to the corner.

attachment.php


With 2" thick pad, I believe I can get to the spacings shown which push the first null out to 720 Hz and reduce the null to a slight depression of 2.5 db or so. At lower frequencies there is a lot more corner support than shown as absorption drops off rapidly.
 

Attachments

  • tight to corner boundary.jpg
    tight to corner boundary.jpg
    237.8 KB · Views: 264
Yeah, now if only I had a corner I could use out here. There is one inside though.

This morning it looks workable to rotate my listening axis so that sounds go across the garage . That gets one of the speakers away from the garage door where it restricts the opening. Any absorption I put on what are now un-insulated block side wall will benefit me thermally.
 
Ages ago it seems, I turned my room on its ear (well rotated it 90 degrees) because the 2nd array crowded the garage door too much and thinking it would be easier to treat the new rear wall.

attachment.php


Since then its been a recovery operation trying to get back to where I was. the real challenge was implementing a circa 100 Hz XO in the presence of boundary interference near XO. Normally one would do this with outdoor measurements and/or nearfield of the LF but one can't nearfield an array...
But I persisted through several long days of trying again and again until I got something that seemed worth trying.

attachment.php


attachment.php


That isn't very impressive but it took all the miniDSP resources. Clearly there is work to be done both on the room and in DSP. But I listened anyway and was encouraged.
 

Attachments

  • across the garage.jpg
    across the garage.jpg
    103.3 KB · Views: 341
  • Across garage with MiniDSP IIR EQ.jpg
    Across garage with MiniDSP IIR EQ.jpg
    68.7 KB · Views: 246
  • Left spectrogram with MiniD EQ.jpg
    Left spectrogram with MiniD EQ.jpg
    110.2 KB · Views: 235
This morning I got up with more energy and did a quick run through DRC Designer.

attachment.php


Left channel
attachment.php


Right channel
attachment.php


DRC did a much better job on the right than the left but right was closer to begin with. This underscores the point that the better it is before DSP, the better the correction DRC does. That means more treatment. My 4" thick rigid absorbers don't go low enough, doing only a few db at 200 Hz. When I get time I will augment them with 8" thick pink and fluffy which porous absorber calculator tells me will be highly effective down to 100 Hz.

There is also a lot more to be done with DSP. This was just a quick check. Curious that DRC didn't flatten the phase at the low end.

Unfortunately duty will take me away for much of the rest of the day
 

Attachments

  • Before and predicted after DSP FR.jpg
    Before and predicted after DSP FR.jpg
    98.1 KB · Views: 239
  • Predicted spectrum on left.jpg
    Predicted spectrum on left.jpg
    81 KB · Views: 347
  • Predicted spectrum on right.jpg
    Predicted spectrum on right.jpg
    79.8 KB · Views: 236
Normally one would do this with outdoor measurements and/or nearfield of the LF but one can't nearfield an array...

You can significantly reduce the amount of "room" in a listening position measurement by impulse averaging 4 or more measurement s from different positions in a line around the listening position.

By impulse averaging you still keep the room resonances and nulls below transition but get rid of a lot of the positional dependent stuff above. About as close to an outdoor measurement as you will get indoors.

I showed how I did it in my thread somewhere around here
Full Range TC9 Line Array CNC Cabinet

REW calls it vector averaging, take the measurements align the impulses to the same point then vector average.
 
Hi Fluid:
Thanks for that - this fresh point of view helps. I know the value of positional averaging but it never occurred to me to use it in crossover work. The closest I'm coming to that is using subs <100Hz both front and rear, which does help.

I'm actually surprised how little room shows up in the FR measurements compared to conventional speakers and smaller rooms, but then looking at the spectrograms, how much there is.

But most of that is front to back stuff and its the reflection from the rear wall that is making my life difficult. Side to side response variation is small enough for my purposes. Should I try averaging responses at different distances from the rear wall? Its worth a try...

So far all I've done is to try to find the optimal arrangement of rear wall absorption and the measurement distance with the least interference. If I build a couple more absorbers I might be there...
 
You can measure anywhere you like and average it out, when changing front to back distance you can run into levels differences which might skew the results unless you normalize them. I tried with 12 points and found that 4 to 6 was the sweet spot for effort vs results. More points reduces the room in the measurement but becomes tedious and the most important stuff is down enough with 4 to 6 measurements.
 
yes, but my question was more to should I want to measure at various distances and then average. Think about the simplified case of a reflection from a boundary. At each distance you will get a null at a different frequency. If you average the multiple measurements with nulls at different frequencies, you pull down the response over the entire region spanned by those nulls. That doesn't get you closer to the direct response.

Instead let's say we have Mclose, a measurement taken close to the wall, which has a null at a high frequency and a Mdist, a measurement taken far from the wall which has a null at a low frequency. Perhaps we should merge these two measurements, like we do with a close mic and normal measurement, so that we use Mdist for the high range and Mclose for the low range and thus have a response relatively unaffected by the wall.
 
The impulse averaging will not get rid of the persistent nulls below transition with any realistic number of measurements.

It is more representative of you what you have over a listening area so that you don't go over the top with EQ that is only valid at that point in space.

Try it with 4 to 6 in a straight line across the area of a couch where you would listen, see if it gives you something different.

Here is link to a paper on beamforming if you haven't seen it before
https://www.researchgate.net/profile/Daniel_Krol2/publication/272498144_Quasi-Anechoic_Measurement_of_Loudspeakers_Using_Beamforming_Method/links/54e71b2b0cf277664ff79394/Quasi-Anechoic-Measurement-of-Loudspeakers-Using-Beamforming-Method.pdf

What you suggested is more along the lines of a moving microphone technique or frequency average where phase is not considered.

Arrays aren't that hard to measure outside because they don't show the ground reflection in the same way a normal speaker would. Maybe open your garage on a quiet day and take a quick measurement before anyone gets annoyed by the loud sweep :)

I imagine you will get a 12dB fall off after the midbass hump from the array, I did.
 
"Impulse averaging" did not suggest the delay sum beam forming technique described in the paper you linked. That sounds powerful. I think I read about it in a different paper some years ago but had forgotten all about it. That other paper had a motorized track to move the microphone along a line.
 
those were with subs using DRC Designer's Normal template - which I can do with a few mouse clicks. No work on templates yet, keep getting diverted into sub configuration and room treatment. Thought I was done but found out I'm not:) XO is about 100 Hz but its not clean. I'm not able to equalize the response slopes to an LRx acoustic profile, too much room in the response.

When I had the arrays working up/down the garage, I got decent bass from my 3 sub config with the rear sub working on both channels. Crosswise that doesn't work so well so I spent time rewiring the rear dual opposed, dual voice driver sub for separate left and right inputs. Not quite finished yet but close.

This will give me another, needed degree of freedom. I will attempt to use the rear sub to fill in the dips in the responses of the stereo front subs.

I also spent time trying to EQ a mono sub configuration but that didn't go so well. I figured the additional spacial diversity would help but it didn't work out. the two front subs are in relatively symmetrical positions, each near a front corner. The rear sub is centered on the rear wall.
 
I believe it mission impossible to set xo that way. Place a sub just next to the array and mic on the floor 60cm from them. Now you will catch mostly direct sound even with long gating. Adjust slopes that match. Then adjust delay and phase to get summation. Check also step response.

Only after this start doing farfield and spot room eq to the signal before xo! Sounds easy, but believe me it's not.

APL style presentation is in REW too, use Spectrogram with wavelets.
 
He probably doesn't want the sub next to the array. With these arrays you want to get the best result you can get at the sweet spot, there are a number of ways to get the sub to integrate without relying on nearfield measurements.
I'd start by measuring the arrays separately, apply DSP correction and next do the same for subs. Only after that try and combine the two to get them in sync. You've got to "see" the strong points of the array and subs separately to make them work as "one".
Nobody forces us to use named crossovers to make these two speakers sound as one. You use what gets the job done. The speakers and room should all work together for best results. We're not trying to get a correct speaker independent from the room, We're trying to make it work with the room. At least I know I am. At these lower frequencies we can use the strong points of either the sub or array to make them work as one strong unit per side. If all goes well you don't have any floor or ceiling bounce, the side wall energy is properly absorbed and you get a crazy clear end result that takes you "there".

Very different philosophy from designing the ideal speaker and then puting it into a room.
But it does take a lot of work to see what the room is doing (and when) and working with that. So the more data that is collected (separately), the better the strategy can be to make it work (together).
 
Last edited:
more seriously

well, the bass response I had back in post 305 was actually pretty good. With DSP, I was able to beat it into decent shape. I could have EQed those remaining dips but held off to see what DRC would do and it didn't do much. The perfectionist in me would have preferred to do something more in line with Juhazi's suggestion but the pragmatist won out.

But there is also array vs point source sub. If I XO at 1m like I did with my Synergies, then out at 4m where I listen, the sub level could be off 6 db.

OTOH, I am looking for a better approach. I need at least one more absorber panel for that 200 hz dip. More on the back wall wouldn't hurt. It might take active absorption there to kill the boundary interference from it that are making the sub xo difficult.

My subs could play through my absorber panels next to the arrays. The panels would simply act as low pass filters. This is another sub position to try or a place for additional subs. A bass array playing through the panels is interesting but that is a potential future project'
 
I suggested near field of both just to get phase match. Levels must be adjusted after that. Wavelength of 100Hz is 3.3 meters and 1,65m for reverse phase. Good timing of sources gives punchy and tight bass, it is more important than table-flat response at spot. Just my opinion based on experience of playing with dsp.
 
I suggested near field of both just to get phase match. Levels must be adjusted after that. Wavelength of 100Hz is 3.3 meters and 1,65m for reverse phase. Good timing of sources gives punchy and tight bass, it is more important than table-flat response at spot. Just my opinion based on experience of playing with dsp.

I favor your approach. Ime, a sub should always be co-located with the main speaker to achieve the punchy tight bass you speak of.

I know many favor the multiple subs approach for evening out freq response. My guess is they are the same folks who like a more disperse sound, a greater sense of ambiance.

When I built a large listening room and was planning the dimensions to m inimize room modal buildup...you know, the golden ratios etc...I also studied the multiple sub idea.

My homespun conclusion is there are basically two approaches to handling low freq modes....one is to minimize their number and strength.
Single or at most dual subs, corner placement if possible or wherever minimizes virtual sub relections, room dimension/acoustic traps , EQ out buildups, leave nulls alone.

Second is to maximize the number of modes, saturate the room with a bunch of peaks..so response becomes a smooth dense cloud so to speak...

Both work, just depends what you like I guess..
I so favor tight reflection-free sound, as heard outdoors.
Inside, I just grin and bear mission impossible ;)

On reflection, rereading this before posting...I guess there is a third method as suggested by Ron...specific tuning to a specific spot. Probably used my most.
I wish I could get into that again, but after ditching electrostatic sweet spot listening 20 years ago...well hell, I haven't even listened while sitting down in years. This old man like to move :)