• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Murphy's Corner Line Array project

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,752
Likes
4,633
Location
Sweden - Слава Україні
What listening distance will you have? It will dip in higher frequencies, do you think of EQa?

With a straight line source of limited length, problems are obtained because the frequency response becomes distance dependent and also when the receiver position is above the top point of the source. However, these problems can be avoided if the length of the source is infinite, which is the case via infinitely many mirroring effects if one has two ideally reflective surfaces at the ends of the line. Now, of course, floors and ceilings are not completely reflective, but close enough.

When creating a line source with dynamic elements, however, these will together (since the individual contributions are not summed in phase due to the distance differences) give a frequency response that decreases by 3 dB / octave, so you have to equalize all elements with a pitch of 3 dB / octave from a couple of hundred Hz and up. You can thus connect the elements in series parallel in a simple way and then equalize "as usual".

However, one should be aware that the efficiency due to the properties of the scattered source do not become particularly high for high frequencies and that one should therefore be prepared for quite a lot of power being consumed here if one does not choose to use only a single element for the highest frequencies. Here, long band elements have great advantages because they, through their low mass, purely acoustically receive a frequency response, even though the sound pressure contributions from the various sub-elements are not summed in phase here either (unless you listen at a distance of 20-50 meters or so) .



Edit:
If you look at that thread, reference is made to this thread. The Hifi world is still quite small. Especially when it comes to line speakers for home Hifi use.:)
(no, not me who wrote that post)
 

Attachments

  • Linjekälla-100mm-element Power+DI-hor-ver.png
    Linjekälla-100mm-element Power+DI-hor-ver.png
    235.1 KB · Views: 101
  • shot_2022-01-20_05-44-47.png
    shot_2022-01-20_05-44-47.png
    255.3 KB · Views: 92
Last edited:

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
I know the question was not addressed to me but if I may share my experience:
I have simulated with BassBox Pro at least half a dozen of different drivers (using 24 pcs/cabinet), like multiple Daytons, Vifa, Visaton, Peerless, Tectonic Elements, etc. and the results were by far the best when using Dayton ND91-8
I was comparing not only the FR curves but cone displacement (very very important for line array projects), impedance, phase and group delay

Hence I have decided to go for that model - speaking of which I just got informed that shipment is further delayed, new ETA is like 4 weeks from now on

For what it is worth, I am looking at doing a line array at home as well. I have worked on them professionally but never for home. I was looking at that Dayton as well, but I will get a few speakers in and measure the distortion first. The Dayton based on Kippel report looks optimized for distortion at a medium excursion at the expense of lowest distortion at lower levels, but proof will be in getting it and measuring it. I plan on using a sub though, so not concerned about distortion at high excursions. I will also be using tweeters. The plan right now is to use aluminum. It will obviously cost more, but I can design it on CAD and then just have it laser cut to spec including mounting holes. As opposed to CNCing flush, I will probably just get another piece the same thickness as the speaker front mounting and have it laser cut too and then glue them together.
 

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
Looking sweet! Super jealous right now!

You could even (if the technology allows in the future) assign a delay to each driver so you would get a constant beam transducer instead.

index.php

JBL has that in pro speakers using a passive network. Remember though, if you are on the floor (or ceiling effectively), that the graphs on the left are not accurate.
 

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,752
Likes
4,633
Location
Sweden - Слава Україні
JBL has that in pro speakers using a passive network. Remember though, if you are on the floor (or ceiling effectively), that the graphs on the left are not accurate.
Sure for pro, PA but for home Hifi to have a couple of such speakers at home would look crazy, purely aesthetically. In any case, I think that even though the taste is divided.:D

ECED8502-4986-454F-A213-BE52FA1154E3.png

Although there are those who have created curved line speakers for home hi-fi use, see attached picture.:)
 

Attachments

  • DSC_0072[1].JPG
    DSC_0072[1].JPG
    12.4 KB · Views: 132
  • DSC_0117[1] (1).JPG
    DSC_0117[1] (1).JPG
    51.6 KB · Views: 139
Last edited:

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
Sure for pro, PA but for home Hfi to have a couple of such at home would look crazy, purely aesthetically. In any case, I think even though the taste is divided.:D

View attachment 180357
Although there are those who have created curved line speakers for home hi-fi use, see attached picture.:)

When I get some free time, I am going to put more thought into the exact implementation I will do with mine. I have no intention of building a curved unit, but I am thinking about the feasibility of how many distinct banks I could have per speaker. I am almost tempted to distribute a digital signal and design mini 1-2 channel, processor/DAC/amp board. A discussion for another thread though. I don't want to take away from the ops great build!
 
OP
P

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,220
Likes
1,360
Location
Budapest
Thanks for sharing, looking good! Can I ask how do you connect your terminals to the drivers and how are you going to put the baffle onto the back?

I am using crimp terminals - 3 types of them for this project
The ND91-8 have a 6.3mm terminal for the + and a 2.8mm terminal for the -
For these I am using standard insulated sliding crimp terminals
The third type is a ring crimp terminal that I use for the parallel connections
I never use soldering, I prefer crimping (personal taste I guess)
You might see these in the above pictures but I am happy to post more pictures, let me know

One of the design principles for this project was that the baffles must remain detachable - for servicing, etc. reasons
If you take a look at the pictures you will see insulation tape applied to the edges where the back and the front baffle will be connected
We will then use (many) screws to attach the front baffle to the back (the holes will be covered so you won't see the screwheads)
I understand this will not be perfect but that is best idea I could come up with
 
Last edited:
OP
P

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,220
Likes
1,360
Location
Budapest
What listening distance will you have? It will dip in higher frequencies, do you think of EQa?

With a straight line source of limited length, problems are obtained because the frequency response becomes distance dependent and also when the receiver position is above the top point of the source. However, these problems can be avoided if the length of the source is infinite, which is the case via infinitely many mirroring effects if one has two ideally reflective surfaces at the ends of the line. Now, of course, floors and ceilings are not completely reflective, but close enough.

When creating a line source with dynamic elements, however, these will together (since the individual contributions are not summed in phase due to the distance differences) give a frequency response that decreases by 3 dB / octave, so you have to equalize all elements with a pitch of 3 dB / octave from a couple of hundred Hz and up. You can thus connect the elements in series parallel in a simple way and then equalize "as usual".

However, one should be aware that the efficiency due to the properties of the scattered source do not become particularly high for high frequencies and that one should therefore be prepared for quite a lot of power being consumed here if one does not choose to use only a single element for the highest frequencies. Here, long band elements have great advantages because they, through their low mass, purely acoustically receive a frequency response, even though the sound pressure contributions from the various sub-elements are not summed in phase here either (unless you listen at a distance of 20-50 meters or so) .



Edit:
If you look at that thread, reference is made to this thread. The Hifi world is still quite small. Especially when it comes to line speakers for home Hifi use.:)
(no, not me who wrote that post)
Very happy to see that this project has generated so much interest kind of 'worldwide' :)

Yes, the highs (and the lows and in fact the whole spectrum) will definitely need to be equalized; that is something that is part of the project
I am using Dirac Live 3 to do its magic and it has always worked perfectly for me. I also heavily modify the target curve to match my taste or even to further linearize the response. (which some might argue to be overcompensation but it works extremely well for me at least)
I will post measurements before/after so we will be able to see how much compensation was needed at both end of the spectrum
 

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,752
Likes
4,633
Location
Sweden - Слава Україні
Very happy to see that this project has generated so much interest kind of 'worldwide' :)

Yes, the highs (and the lows and in fact the whole spectrum) will definitely need to be equalized; that is something that is part of the project
I am using Dirac Live 3 to do its magic and it has always worked perfectly for me. I also heavily modify the target curve to match my taste or even to further linearize the response. (which some might argue to be overcompensation but it works extremely well for me at least)
I will post measurements before/after so we will be able to see how much compensation was needed at both end of the spectrum
Sounds great! :D

Can you make a few different measurements at different listening distances? It would it be interesting to see those results. You do not need EQ for all distances just do a few swipes.

Edit:
Attached image from the thread I referred to earlier.
 

Attachments

  • shot_2022-01-20_08-13-32.png
    shot_2022-01-20_08-13-32.png
    387.6 KB · Views: 100
Last edited:
OP
P

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,220
Likes
1,360
Location
Budapest
Sounds great! :D

Can you make a few different measurements at different listening distances? It would it be interesting to see those results. You do not need EQ for all distances just do a few swipes.

Edit:
Attached image from the thread I referred to earlier.

This formula is great, thank you for sharing!
My listening distance will be ~3 meters so I will be well within the critical listening distance
Yes, I will do those sweeps at multiple distances (also to determine where I need to move my couch; I might not necessarily sit at the exact tip of the regular triangle but maybe a few inches closer and farther depending on the response I measure)
 

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,752
Likes
4,633
Location
Sweden - Слава Україні
This formula is great, thank you for sharing!
My listening distance will be ~3 meters so I will be well within the critical listening distance
Yes, I will do those sweeps at multiple distances (also to determine where I need to move my couch; I might not necessarily sit at the exact tip of the regular triangle but maybe a few inches closer and farther depending on the response I measure)
Adds a graph that Wesayso (who also wrote in your thread) commented on line speakers,# 1358

singlevsarray.gif


 

Attachments

  • updatepanel (2).jpg
    updatepanel (2).jpg
    159.5 KB · Views: 147
  • shot_2022-01-20_08-29-57.png
    shot_2022-01-20_08-29-57.png
    709.4 KB · Views: 142

bigjacko

Addicted to Fun and Learning
Joined
Sep 18, 2019
Messages
721
Likes
359
The ND91-8 have a 6.3mm terminal for the + and a 2.8mm terminal for the -
For these I am using standard insulated sliding crimp terminals
Are those sliding crimps good for connecting to drivers? For electronics people are already complaining everything should be soldered, for drivers they would not accept crimp terminals. I have heard that the drivers might vibrate and shake off the terminals. Whay is your view on this and how do you combat this issue?
 
OP
P

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,220
Likes
1,360
Location
Budapest
I have always used those sliding crimp terminals for all my previous projects (also for subs) and I have never ever had such an incident
You need to make sure that the crimps are firmly attached - there is a small protruding part on these terminals that actually prevents the terminal from sliding off
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Line arrays don't really work like that, even with ceiling bounce. You start losing the benefit of the line array well above 100Hz and with a sub-array whether you have 1/R or 1/R^2 really does not matter.
I always thought that the more sources of sound you have in different poles of a room mode the less seat to seat variance you have and at a certain number of sources even the effect of the room mode is smoothed out entirely.
 
OP
P

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,220
Likes
1,360
Location
Budapest
I always thought that the more sources of sound you have in different poles of a room mode the less seat to seat variance you have and at a certain number of sources even the effect of the room mode is smoothed out entirely.

Actually we will see that shortly based on my measurements (at least for my listening room)
My current system is now moved into the corners for exactly the same purpose: to be able to compare it with the corner line array in terms of response below the Schroeder frequency
I will post those comparison measurements too so we will see if/how the corner line array can 'tame' the room modes
 

Wesayso

Active Member
Joined
Dec 30, 2021
Messages
123
Likes
291
Location
The Netherlands
Adds a graph that Wesayso (who also wrote in your thread) commented on line speakers,# 1358

View attachment 180385

Be aware that the graph I posted there is from a frequency shaded line array. The higher one goes in frequency, the less number of drivers will still play at full volume. There is a distinct difference between the two.
shaded-unshaded.gif

Here's the difference in vertical pattern control of the frequency shaded array vs the unshaded array. The more bundled result is the 'frequency shaded' array. It won't have any combing till above 10 KHz, and even there it is no more than a slight wiggle despite using 3.5" drivers.

I don't quite agree with the statement by @audio2design that the advantage of the array is lost well above 100 Hz. As the floor to ceiling array uses the mirror images in floor and ceiling to make it seem taller than it is. Over on DIYaudio we have some measurements that show that the 3 dB drop continues to even below 100 Hz, but the drivers themselves obviously loose output. If one EQ's it back to flat the traits of the line array remain.

It meant I was able to get this result after DSP:
stereo.jpg

Letting my array play down to 20 Hz due to the room helping out. It could not play a note at full power at say 30 Hz, but not a lot of music has notes that strong. Nowadays I use subs playing together with the arrays to keep the advantage multiple sources have at lower frequencies. In a way it is a form of distributed bass.

The advantage of the floor and ceiling mirror has been conveniently neglected in Keele's CBT presentations. He simmed a short 1m straight array (without floor mirror) vs a CBT array that does use it's floor mirror. A floor to ceiling array using the floor and ceiling mirror vs the CBT array would have been quite a different result. I've simmed the CBT using the same driver I have in my straight array, it suffers from way more combing even after his form of volume shading and bending the array. Not an option I would choose for a driver that size. It would require much smaller high frequency drivers. I would never use the volume shading though, it is throwing away one of the true benefits of arrays.

shadedvsunshaded.gif

A full comparison of a frequency shaded array vs an unshaded array to show the difference in combing effects.
The frequency shading need not be complicated, but it does require some shifting of groups for best effects. With an even number of drivers, like @ppataki has, it can even be simpler (and more effective) than my 5x5 driver groups.
 
Last edited:

Ericglo

Senior Member
Joined
Apr 21, 2020
Messages
452
Likes
323
Be aware that the graph I posted there is from a frequency shaded line array. The higher one goes in frequency, the less number of drivers will still play at full volume. There is a distinct difference between the two.
View attachment 180398
Here's the difference in vertical pattern control of the frequency shaded array vs the unshaded array. The more bundled result is the 'frequency shaded' array. It won't have any combing till above 10 KHz, and even there it is no more than a slight wiggle despite using 3.5" drivers.

I don't quite agree with the statement by @audio2design that the advantage of the array is lost well above 100 Hz. As the floor to ceiling array uses the mirror images in floor and ceiling to make it seem taller than it is. Over on DIYaudio we have some measurements that show that the 3 dB drop continues to even below 100 Hz, but the drivers themselves obviously loose output. If one EQ's it back to flat the traits of the line array remain.

It meant I was able to get this result after DSP:
View attachment 180399
Letting my array play down to 20 Hz due to the room helping out. It could not play a note at full power at say 30 Hz, but not a lot of music has notes that strong. Nowadays I use subs playing together with the arrays to keep the advantage multiple sources have at lower frequencies. In a way it is a form of distributed bass.

The advantage of the floor and ceiling mirror has been conveniently neglected in Keele's CBT presentations. He simmed a short 1m straight array (without floor mirror) vs a CBT array that does use it's floor mirror. A floor to ceiling array using the floor and ceiling mirror vs the CBT array would have been quite a different result. I've simmed the CBT using the same driver I have in my straight array, it suffers from way more combing even after his form of volume shading and bending the array. Not an option I would choose for a driver that size. It would require much smaller high frequency drivers. I would never use the volume shading though, it is throwing away one of the true benefits of arrays.

View attachment 180400
A full comparison of a frequency shaded array vs an unshaded array to show the difference in combing effects.
The frequency shading need not be complicated, but it does require some shifting of groups for best effects. With an even number of drivers, like @ppataki has, it can even be simpler (and more effective) than my 5x5 driver groups.


I am sure you have seen this. Monte Kay wrote about his experience.

The main left/right speakers are "near" infinite line. The array is 8' tall mounted 1' off the floor thus leaving a 1' gap between array and boundary at each end. There are eighteen Dayton Reference 6” truncated frame mid-bass drivers and forty two 2” Aura mid-tweeters in the array. Originally no shading was used but with encouragement from Don Keele I later added Hann shading to the array. I no longer believe that shading is the best approach nor is CBT. (Sorry Don) I believe instead that in a situation where a true infinite line is possible, it is the better approach. Problem with any line array other than CBT is that they only work well if the listening position is in the center. With an infinite line, no matter where you are, you are in the center so problem solved. Where infinite line is not possible however, CBT is definitely the best approach. I view the CBT as the perfect solution for finite line arrays, in fact CBT is really the only mathematically correct way to build a finite line array.

 

Wesayso

Active Member
Joined
Dec 30, 2021
Messages
123
Likes
291
Location
The Netherlands
I can fully agree with that ;). In a room with parallel floor/ceiling, a straight floor to ceiling array has advantages. CBT is an option for a finite array.

One of the best theoretical threads on infinite line arrays is this one: https://www.diyaudio.com/community/threads/infinite-line-source-analysis.314917/
Written and explained way better than I ever could. Complete with the math etc. which isn't easy to plow trough.

The kind of shading I used isn't to mimic CBT behavior, just an experiment with more bundling of the vertical beam and combating comb filtering in the process. Sort of a variation of an expanding array. Out of pure curiosity I might add, as I wasn't at all disappointed with the unshaded version. I've done a lot of things out of pure curiosity and in the interest of learning. I see this as an ongoing and fun science project.
It has taught me a lot already and I have many more things I want to try. Meanwhile I do have a killer sound system to listen to.

One of the latest whims is replacing the drivers with these:
10f-stash.jpg


Do I expect differences? Yes. Do I expect it to be a huge difference? Not really, not after FIR processing.
I just have to try to know for sure. :D

(I've had some 10 year old TC9's fail on me, an old problem due to corrosion on the connection between lead wire and voice coil. Scan Speak was familiar with that problem and have addressed it using a different kind of glue to protect that connection. I could get no answer from Tymphany.
If I'm replacing drivers, I might as well satisfy my curiosity. YOLO)
 
Top Bottom