Just as i have the artocles fresh in mind:
*Each perfect reflecting surface should give a 6 db rise in SPL. So that should be 6 dB for floor + 6 dB for ceilling = 12 dB. (Assuming a perfect reflecting floor was not found outdoors)
*The 45 degree Field at wavelength same as element distance was interesting (16k in the paper, but much smaller element distance)
*Put a thick carpet on the floor between you and your multiway speaker
*Each perfect reflecting surface should give a 6 db rise in SPL. So that should be 6 dB for floor + 6 dB for ceilling = 12 dB. (Assuming a perfect reflecting floor was not found outdoors)
*The 45 degree Field at wavelength same as element distance was interesting (16k in the paper, but much smaller element distance)
*Put a thick carpet on the floor between you and your multiway speaker
Result looks great
Based yourself being a drummer does setup subjective perform well into this area.
Thanks Byrtt there has been a lot of work going in behind the scenes but any credit for the DRC processing goes to wesayso. The processing chain on my computer makes me look more and more a wesayso fanboy every day, I would like to be different just for the sake of it but every time I tend to prefer the same as wesayso....
My credit card number is 45... oh wait you said don't share it
I haven't listened to the DRC processing yet but I have listened to the PEQ used to get it closer to the target. I listened for quite a while yesterday to a range of music and I really liked it. That has encouraged me to just add the DRC processing in based on the slightly virtual EQ measurements and try it out.
I was listening to acoustic tracks in general which were sounding really good, I couldn't crank the volume to give a true impression of the drums. One of my favourite test tracks for drums is Fleetwood Mac's "The Chain" I listened to that and it sounded good. The overall sound is definitely heading in the right direction
Thanks, I am glad to see you have been reading up on arrays even if you still seem to be struggling to let go of your pre-conceptions.
This stuff is not new, here is a paper from the sixties with plenty of maths in it if you like that sort of thing. The author created a line array to use in a church specifically because of it's directivity.
http://p10hifi.net/planet10/TLS/downloads/line-source.pdf
The floor and ceiling do help but they are not necessary to reap some of the benefits. My outdoor measurement did have a reflective floor about 1m or so below but there was no ceiling
Jim Griffin's paper is another source of good information
http://www.audioroundtable.com/misc/nflawp.pdf
You will find that as long as there is 75% or more coverage between floor and ceiling the array acts in a similar way as if there was no gap. Floor to ceiling is the general way to describe an array that is 75% or more. You can make them go all the way but then they are a little tricky to install
The graph you linked that shows the dip around 500 to 600 is interesting and shows part of what I was saying about the averaging techniques. Impulse averaging weights the dips more heavily. Even though you can see that quite a few of the measurements have peaks at that same location. The dips outweigh the peaks with impulse averaging.
I don't know if DRC is smarter than you but the algorithm works well because it leaves that area fairly well alone. It has made it better but has not tried to fix the unfixable.
I think DRC correction is another area where you are assuming facts that are not in evidence here. This is predominantly speaker correction and any room correction is only really done over a very short time period. It was wesayso's more unconventional approach to using processing that attracted me to read more about arrays and I saw enough in that to build my own.
As wesayso said the measurement was based on an average and it won't be totally different 2cm to the left. It will be different but the amount will be less. That is no brute force correction to make everything flat no matter what. This is what is good about speakers that have well controlled directivity whether it be narrow or wide. The reflections are broadly similar to the direct sound so they are less distracting. They also respond better to correction because the power response and DI is flatter.
Maybe I have missed them but I don't recall seeing any measurements of your speakers.
I am all for discussion, questions and criticism of what I have posted but I would prefer it to be helpful and informed. Sometimes there is a fine line between discussion and argument, lets try and stay on the right side.
DRC: From the documentation
"Furthermore, with digital room correction it is worth to experiment with unusual speaker placements. Reflections from nearby walls are more difficult to correct when they are away from the main spike, so placing the speakers near to the walls, or may be even in the corners, might sometime give better results with DRC, provided that you place some absorbing material near the speakers to remove early reflections in the high frequency range, where DRC is able to correct only the direct sound."
So DRC was tested by the author. I believe he used conventional speakers. In any case I am pleased to hear that all the logic inside DRC also works with arrays
Array boundaries: Keele uses and models one boundary with reflecting index 1.
In real world that means granite or marble floors like in churches or large buildings.
This kind of reflecting index is not often found in small domnestic rooms.
If both floor an ceilling has index 1, e.g. both are flat polished granite, the room will be very hard to hold an conversation with more talking people inside. So very few rooms are made that way.
That floor and ceilling should also give you a 12 dB rise in SPL. Look like you have about 4 dB rise and it is not uniform at all frequencies. So I think it is safe to say that boundary condition of reflecting index = 1 is not met.
So then the array/room will generate reflexions from floor and ceilling that is different from the reflexion described by dbkeele
The solution still works it seems but I don't understand why reflecting image speaker should be brougth into the description of the solution
"Furthermore, with digital room correction it is worth to experiment with unusual speaker placements. Reflections from nearby walls are more difficult to correct when they are away from the main spike, so placing the speakers near to the walls, or may be even in the corners, might sometime give better results with DRC, provided that you place some absorbing material near the speakers to remove early reflections in the high frequency range, where DRC is able to correct only the direct sound."
So DRC was tested by the author. I believe he used conventional speakers. In any case I am pleased to hear that all the logic inside DRC also works with arrays
Array boundaries: Keele uses and models one boundary with reflecting index 1.
In real world that means granite or marble floors like in churches or large buildings.
This kind of reflecting index is not often found in small domnestic rooms.
If both floor an ceilling has index 1, e.g. both are flat polished granite, the room will be very hard to hold an conversation with more talking people inside. So very few rooms are made that way.
That floor and ceilling should also give you a 12 dB rise in SPL. Look like you have about 4 dB rise and it is not uniform at all frequencies. So I think it is safe to say that boundary condition of reflecting index = 1 is not met.
So then the array/room will generate reflexions from floor and ceilling that is different from the reflexion described by dbkeele
The solution still works it seems but I don't understand why reflecting image speaker should be brougth into the description of the solution
The processing that works better for arrays has different windows lengths and parameters than those for conventional box speakers. Other than that there isn't that much difference. The arrays benefit from having a reasonably uniform response through most of the frequency range. Much more so than a stand or floor speaker would have if placed in the same position. This helps the processing work better over a larger area and be more free of the issues that brute force room correction is known for. Whilst it may only be optimum for one point in space it can also be very good for quite a large area.
When it comes to the theory of arrays I prefer the mirror example where if you view the floor and ceiling as mirrors the array seems and acts longer.
The less reflective the floor and ceiling the more "dirty" the mirror and the less it helps to extend the imaginary length of the line.
Keels's CBT uses many of the same principles but there is a fundamental difference with the CBT being curved back. A backwards quarter curve looks a half curve with a mirror floor reflection. It would be weird with a ceiling added. The straight array works well in theory with either.
To me the mirror theory extending the length is valid but as you say in real situations the effect is reduced. I also don't consider it to be the primary mechanism under which an array like this is able to work well as a speaker in a real room. The actual physical length is the biggest part. I would like to do more outdoor testing to get more data like the Harman "spinorama" measurements but I can't see it happening anytime soon as it would take a huge amount of time and is very susceptible to outside interference.
I am satisfied that the theoretical descriptions work well enough in practice and produce a very usable speaker.
When it comes to the theory of arrays I prefer the mirror example where if you view the floor and ceiling as mirrors the array seems and acts longer.
The less reflective the floor and ceiling the more "dirty" the mirror and the less it helps to extend the imaginary length of the line.
Keels's CBT uses many of the same principles but there is a fundamental difference with the CBT being curved back. A backwards quarter curve looks a half curve with a mirror floor reflection. It would be weird with a ceiling added. The straight array works well in theory with either.
To me the mirror theory extending the length is valid but as you say in real situations the effect is reduced. I also don't consider it to be the primary mechanism under which an array like this is able to work well as a speaker in a real room. The actual physical length is the biggest part. I would like to do more outdoor testing to get more data like the Harman "spinorama" measurements but I can't see it happening anytime soon as it would take a huge amount of time and is very susceptible to outside interference.
I am satisfied that the theoretical descriptions work well enough in practice and produce a very usable speaker.
Torgeirs, did you come to terms the floor and ceiling have way less of an influence with these arrays? That was the primary goal of last night excersize.
You can let go of the real reflection index within our rooms.
The real benefit for me is having speakers placed well out of the way for day to day life, (that fit on an A4 size piece of paper and work well with a room.
Despite their unusual placement they have a lot of potential and do many things right.
They can work with the room. And their output stays largely the same wether you move left or right, up or down. Not many other concepts can do this.
You can let go of the real reflection index within our rooms.
The real benefit for me is having speakers placed well out of the way for day to day life, (that fit on an A4 size piece of paper and work well with a room.
Despite their unusual placement they have a lot of potential and do many things right.
They can work with the room. And their output stays largely the same wether you move left or right, up or down. Not many other concepts can do this.
fluid: Agree mostly, but i from the measurements I would consider the mirror so dirty it's no longer a mirror. But ist polemics. As long as it works with DRC for its purpose. (My system is used over 10 m^2 and for real time sound so I cant use speakers that has to be time corrected)
wesayso:
Can't say so. According to dbkeele and scaled to 2 meters and longer element distance, the speaker will beam vertically like most multiway speakers do (But have another beaming signature, se picture taken from the article) So I got back to start, reading the papers.
But guess the DRC saves the design and is an integral part, as the straight vanilla version don't perform well according to dbkeele. (Or maybe it was totally different conditions)
"The performance data for all the arrays was subjectively
ranked for each performance type on a scale from 1 to
10 and then totaled (scale 8 to 80) for each array, to
yield the final array ranking:
Keele Performance Ranking of Line Arrays
AES 129th Convention, San Francisco, CA, USA, 2010 November 4–7
Page 21 of 31
1. CBT circular-arc array: 80,
2. CBT delay-curved straight-line array: 77,
3. Un-shaded circular-arc array: 54,
4. Spiral-line array: 46,
5. “J”-line array: 26,
6. Straight-line array (Hann shaded): 23, and
7. Straight-line array (not shaded): 10 "
wesayso:
Can't say so. According to dbkeele and scaled to 2 meters and longer element distance, the speaker will beam vertically like most multiway speakers do (But have another beaming signature, se picture taken from the article) So I got back to start, reading the papers.
But guess the DRC saves the design and is an integral part, as the straight vanilla version don't perform well according to dbkeele. (Or maybe it was totally different conditions)
"The performance data for all the arrays was subjectively
ranked for each performance type on a scale from 1 to
10 and then totaled (scale 8 to 80) for each array, to
yield the final array ranking:
Keele Performance Ranking of Line Arrays
AES 129th Convention, San Francisco, CA, USA, 2010 November 4–7
Page 21 of 31
1. CBT circular-arc array: 80,
2. CBT delay-curved straight-line array: 77,
3. Un-shaded circular-arc array: 54,
4. Spiral-line array: 46,
5. “J”-line array: 26,
6. Straight-line array (Hann shaded): 23, and
7. Straight-line array (not shaded): 10 "
Attachments
So you don't buy any of our data and measurements but do fall for the commercially driven advertising by Keele. My graphs are actually prettier than his published measurements.
Go figure....
Yes, DSP is an integral part for me and my lines. But we have several users that are satisfied with simpler means.
To come back to Keele's comparison... his line array on the left side, you know, the mediocre one, is actually one (1) meter tall with floor reflection.
So why all these posts from you on this thread?
Personally, I wouldn't trade my arrays for Keele's best effort. Even though I don't think its a bad concept, I couldn't fit it into day to day life and have it work with my room.
By the way, none of those arrays mentioned resemble our best effort. Something that works with the room.
Go figure....
Yes, DSP is an integral part for me and my lines. But we have several users that are satisfied with simpler means.
To come back to Keele's comparison... his line array on the left side, you know, the mediocre one, is actually one (1) meter tall with floor reflection.
So why all these posts from you on this thread?
Personally, I wouldn't trade my arrays for Keele's best effort. Even though I don't think its a bad concept, I couldn't fit it into day to day life and have it work with my room.
By the way, none of those arrays mentioned resemble our best effort. Something that works with the room.
Last edited:
Well I suppose some agreement from you is better than where we started
So far I have limited listening experience with DRC so I am not ready to make any comments on that front.
I do not consider the phase correction to be an "integral" part of the design, there is a definite need for EQ but that could be achieved with little to no latency if needed. Wesayso has shown an extremely refined DRC correction that a number of independent listeners have confirmed as sounding excellent so I have no doubt that it can be done and for the icing on the cake the time correction.
The concept of the speakers wesayso, I and others have built is not flawed without DSP time correction. The IDS-25 of Roger Russell works pretty well with an analogue equaliser.
Don Keele is a fantastic scientist but in this case he has a commercial interest and he wouldn't be the first to stack the deck in his conclusions.
The CBT is a similar concept it is not the "same" and his descriptions and conclusions on straight arrays to me don't apply to my speakers. I have never heard one but there have been some measurements posted of the full range CBT in another thread and I think they are awful. A massive hole in the top end which is EQ'd back to not even flat. Even using a driver with a rising response. I know the fullrange CBT is not his best effort but even so I think you need to take a large spoonful of salt when reading his conclusions.
Read the data for what it is. That was point of directing you to it, it's background to understand the concepts involved as you seemed to be finding it hard to get your head around. Look at the CBT measurements and look at mine or wesayso's and decide for yourself where to rank them.
I would like to see the measurements of your system as it might help to understand where you are coming from. If you can't or don't want to show I can understand.
Hi Torgeirs,
I made some measurements to determine the changes in response with height for my floor-to-ceiling arrays. See here (starting post 127):
http://www.diyaudio.com/forums/mult...line-array-using-vifa-tc9-13.html#post4584857
In post 129, there is a measurement very close to the ground and you can see there is considerable drop in response there, which means the array is directional. If you look at the impulse response (not posted by me in my thread), there is no reflection where you would expect one in the array measurement. It is definitely there when measuring conventional cone/dome speakers in the same position.
It might help this discussion if the original line array papers by David Smith are read:
http://www.diyaudio.com/forums/multi-way/243861-floor-ceiling-array-vs-cbt-3.html#post3666436
And there was a whole thread dedicated to comparing floor-to-ceiling and CBT arrays:
Floor-to-ceiling array vs CBT
I would pay particular attention to speaker dave's posts.
I made some measurements to determine the changes in response with height for my floor-to-ceiling arrays. See here (starting post 127):
http://www.diyaudio.com/forums/mult...line-array-using-vifa-tc9-13.html#post4584857
In post 129, there is a measurement very close to the ground and you can see there is considerable drop in response there, which means the array is directional. If you look at the impulse response (not posted by me in my thread), there is no reflection where you would expect one in the array measurement. It is definitely there when measuring conventional cone/dome speakers in the same position.
It might help this discussion if the original line array papers by David Smith are read:
http://www.diyaudio.com/forums/multi-way/243861-floor-ceiling-array-vs-cbt-3.html#post3666436
And there was a whole thread dedicated to comparing floor-to-ceiling and CBT arrays:
Floor-to-ceiling array vs CBT
I would pay particular attention to speaker dave's posts.
No need for harch words on dbkeele. He has given us a lot of very good open info.
I just like to understand how things work and get as much unbiased info as possible.
The initial unDSPed room measurements are quite good.
The close wall reflection is taken care of by DRC as DRC recomends
The deep bass is controlled inside what achivable by placement of array.
I am impressed by fluids efforts as written many times.
A speaker that don't sound good without heavy DSP is actually a trend nowadays. (And I am not ironoc)
I just like to understand how things work and get as much unbiased info as possible.
The initial unDSPed room measurements are quite good.
The close wall reflection is taken care of by DRC as DRC recomends
The deep bass is controlled inside what achivable by placement of array.
I am impressed by fluids efforts as written many times.
A speaker that don't sound good without heavy DSP is actually a trend nowadays. (And I am not ironoc)
I don't think my words were harsh (they were certainly not meant to be) and I have huge respect for Don Keele. I was trying to be unbiased by pointing out that he has a commercial interest in his speakers and the information he presented in that paper on other types of arrays was at best incomplete. I think the CBT concept is great and the first version looked good. The Parts Express full range version does not look good to me this is the graph I was referring to
I think that is the best place to leave this discussion and I will say thank you for those comments.
That might be the case, I don't think this is one of them. They do need a decent amount of EQ which is most easily done through DSP but it doesn't have to be.
For the last few years I have been running a Celeron NUC as my music computer, Windows really bogs it down but for the processing I want to run now Windows is the best choice so...
With a 20% off ebay offer I decided to go for an upgrade to an i3 Kaby Lake NUC with 8Gb of RAM, 128Gb Nvme PCIe boot drive and 500GB SSD to hold the music on. To make sure it is silent I have splashed out the extra for the Akasa Plato x7 chassis as the NUC fan can get a bit noisy sometimes, plus it is thinner and will fit under my TV
The latest NUC chassis have a bit a disco light show going on with the LED ring on the front panel another reason not to use it
This is the fastest booting computer I have ever built, it takes Jriver longer to load than Windows does. I hit one snag when building as the wifi card uses an antenna built into the chassis and the Plato doesn't come with a cable to hook anything up to the wifi
With a 20% off ebay offer I decided to go for an upgrade to an i3 Kaby Lake NUC with 8Gb of RAM, 128Gb Nvme PCIe boot drive and 500GB SSD to hold the music on. To make sure it is silent I have splashed out the extra for the Akasa Plato x7 chassis as the NUC fan can get a bit noisy sometimes, plus it is thinner and will fit under my TV
The latest NUC chassis have a bit a disco light show going on with the LED ring on the front panel another reason not to use it
This is the fastest booting computer I have ever built, it takes Jriver longer to load than Windows does. I hit one snag when building as the wifi card uses an antenna built into the chassis and the Plato doesn't come with a cable to hook anything up to the wifi
Attachments
I got the wifi working but in a different way. I ordered some u.fl antenna adapters from ebay but who knew that u.fl sockets came in different sizes? Not me. So they didn't fit the wifi card. I tried to desolder them but it was impossible without destroying them due to the way they were manufactured.
So I took out the aerial assembly that is inside the NUC chassis and fitted those to the base of the case with the original tape that was holding them down.
And now it works
So I took out the aerial assembly that is inside the NUC chassis and fitted those to the base of the case with the original tape that was holding them down.
And now it works
...And now it works
Nice continue on that road, curios if on new Kaby Lake platform you have tried run FocusRite interface on latests Win10 build with the inbox USB audio class 2 drivers, ask because did a test other day with a ASUS U7 and wow sound stream is HQ improvement using MS driver instead of what ASUS offer for C-media chipset, only drawback is then ASIO dll is missing but JRiver fortunately support WASAPI, will try later with XMOS chipset in Behringer UMC204HD if it also will rock better with MS inbox driver.
I haven't tried it yet. I am using the Mix Control Software to adjust the main volume remotely at the moment and I don't know if that is tied in directly with the ASIO drivers but I can test and see.
I have the creators update of windows10 installed for the native USB2 driver but I haven't tried it yet, do you have to install it for a specific device or does it just show up like with Mac? I am leaning towards using my AK4490 DAC with an analogue volume board inside as with all the EQ cut and boost and volume levelling I think I am getting close to the edge of acceptable resolution in the convertors and digital volume control will tip it over.
That is assuming that the volume on the Focusrite is analogue and not just a digital pot
What do you hear that is better? I am not doubting your impression but I find it hard to reconcile that OS tweaks etc. can have a noticeable impact on the sound unless the driver was really badly written in the first place. Interested to know what you think.
I have the creators update of windows10 installed for the native USB2 driver but I haven't tried it yet, do you have to install it for a specific device or does it just show up like with Mac? I am leaning towards using my AK4490 DAC with an analogue volume board inside as with all the EQ cut and boost and volume levelling I think I am getting close to the edge of acceptable resolution in the convertors and digital volume control will tip it over.
That is assuming that the volume on the Focusrite is analogue and not just a digital pot
What do you hear that is better? I am not doubting your impression but I find it hard to reconcile that OS tweaks etc. can have a noticeable impact on the sound unless the driver was really badly written in the first place. Interested to know what you think.
Before MS had support for UAC2 we got a unknown device and needed manufactures drivers support, nothing need to be installed now as with Mac and looking into driver tab one can see for class 2 it loads USBAUDIO2.SYS where the old class 1 devices always used USBAUDIO.SYS, and in MS sound control panel we get right choices to select 24bit and much higher sample rates.
Sounds like a nice DAC you have there in that AK4490 : )
Think know what you mean about OS tweaks and etc plus drivers, it shouldn't make a difference but it sadly often does in my world, one example is for U7 they support four driver versions for Win7 and they actual had cleaned up a bit in there was 6 or 7 in the past. Remember many years ago experienced same phenomena when a old Creative SB device got MS inbox driver in a newer OS it also started sound better as ever only lacking ASIO-dll and manufactures control panel.
U7 use CS4398 on front set out and not a bad chip so to explain what i hear is better is really a perfect coherent clean non distorting non digital sound stream and is not difficult to sense that this computers audio setup rocks and pretty sure you would agree if we was situated in same room and changing OS HDD between as before and with new driver. To get my preferred PCI and PCIe soundcards to sound close to as good as the U7 now performs normally cost some tweaking with manufactures control panel mostly about buffer setting and same tweaking is often needed for buffer settings inside JRiver, but that U7 setup was much simpler and run defaults inside JRiver using WASAPI.
Sounds like a nice DAC you have there in that AK4490 : )
Think know what you mean about OS tweaks and etc plus drivers, it shouldn't make a difference but it sadly often does in my world, one example is for U7 they support four driver versions for Win7 and they actual had cleaned up a bit in there was 6 or 7 in the past. Remember many years ago experienced same phenomena when a old Creative SB device got MS inbox driver in a newer OS it also started sound better as ever only lacking ASIO-dll and manufactures control panel.
U7 use CS4398 on front set out and not a bad chip so to explain what i hear is better is really a perfect coherent clean non distorting non digital sound stream and is not difficult to sense that this computers audio setup rocks and pretty sure you would agree if we was situated in same room and changing OS HDD between as before and with new driver. To get my preferred PCI and PCIe soundcards to sound close to as good as the U7 now performs normally cost some tweaking with manufactures control panel mostly about buffer setting and same tweaking is often needed for buffer settings inside JRiver, but that U7 setup was much simpler and run defaults inside JRiver using WASAPI.
Last edited: