The above measured response is perhaps the best from any speaker in DIY or even in all of hifi. Well done Wesayso! As Howard Beale said "I have seen the face of God..."
Has anyone seen better? I don't think so.
Amazing accomplishment.
Nearly 6 years ago, I built nearly an identical alignment using the TC9 which i proposed in a thread here on diyaudio.net. My measured results were nearly identical to Wesayso's using 31 band eQ.......easy to dial in a very listenable and. Dynamic response.
Sadly, measurements ( including smooth FR) at face value don't tell nearly the whole story and you're left with subjective opinions......so here's mine......
Fullrange arrays are VERY genre dependent which I experienced first hand and attribute it heavily to destructive combing in which complex electric rock (Metallica, Maiden, Boston, Ozzy, etc) are basically unlistenable. It may be just my subjective note, but it flies with the theory behind he fundamentals of line arrays and combing. Classical, acoustic Jazz, Symphony.....all really nice, warm and smooth. .....
Mayhem,
Cool that you built this before - but if you say you used 31 band graphic equalizer - I don't believe it would be possible for you to achieve flat phase like Wesayso's arrays which utilize state of the art FIR techniques combined with room treatments. Not to take anything away from you - but I would say your subjective impressions of your arrays would be very different of your impressions of Wesayso's arrays. If you would like to share with us your frequency response, phase, impulse and step response - we will see if you achieved the same level of sonic perfection. I am a believer in measurements - if it measures well in all aspects, it will play all genres equally well. There is nothing that I can find at fault with the measurements shown here.
Cool that you built this before - but if you say you used 31 band graphic equalizer - I don't believe it would be possible for you to achieve flat phase like Wesayso's arrays which utilize state of the art FIR techniques combined with room treatments. Not to take anything away from you - but I would say your subjective impressions of your arrays would be very different of your impressions of Wesayso's arrays. If you would like to share with us your frequency response, phase, impulse and step response - we will see if you achieved the same level of sonic perfection. I am a believer in measurements - if it measures well in all aspects, it will play all genres equally well. There is nothing that I can find at fault with the measurements shown here.
I am sorry but you are wrong. Combing from arrays with reasonably small elements, such as the Vifa TC9, is simply not audible at the listening position. The random phase additions from multiple elements plus the floor and ceiling reflection of the array combine to completely negate whatever comb filter effects there are, especially if you are more than a few feet away from the arrays. Read Speaker Dave's paper on line arrays and then read Toole on comb filtering. Basically, Toole says that comb filtering from baffle edges is a non-audible effect that is negated by the auditory system.
I've built a similar speaker to Wesayso's and I'm a fan of Led Zepp and Nirvana, and I'm sorry, but these speakers have the most amazing reproduction of all the speakers I've heard ever.
Well I'm pretty much heavily into the Rock category and build these arrays to please. Here's a preliminary list of things I tried so far:
There are more not mentioned here but does this cover some of the category you call "complex electric Rock"?
I'd love to see those measurements, as I think I have an idea what went wrong. I did not get these measurements in one day. And I did make wrong decisions along the way which could result in bad impressions like the one you mention.
So for now I'd have to side up with xrk971 and ask to see those measurements.
What I think went wrong for you is the correction of things that shouldn't have been corrected. While this is a simple sentence it is way more complicated to achieve.
Measurements are hard to interpret sometimes. I could have used brute force to force my measurements into shape with FIR filters to make them look good.
And get real bad sound in the process. But I don't think those measurements would resemble what I've shown here. If you read my entire thread about the correction there is one key point to notice that I tried to achieve: Find the bare minimum correction with FIR leading to the best results possible.
It took me literary months to find it. Many different factors play a role here. But one of the most important things is to use a short window to correct the response. Especially in mid and low frequencies. The area where the room starts to dominate.
Do you still have the mentioned arrays? I bet we could work on the results and come out on top.
If there is one thing I've learned it's measurements are the best indication of the perceived sound. Way more reliable than my ears alone.
But you cannot use brute force and expect good results. So far these arrays play every genre I tried. But I did not get there overnight.
If you're willing to try, it really isn't that hard. Except for the fact that you do need to be able to have a convolver in the audio chain.
But even without FIR, I bet we could improve on your current results. Drop in this thread and start reading from post 206 onwards:
http://www.diyaudio.com/forums/full-range/275730-convolution-based-alternative-electrical-loudspeaker-correction-networks-21.html#post4548875
I bet this method Greg mentions could make an improvement your current EQ use.
It will not be as powerful as full FIR correction but I bet it will sound more right than EQ-ing a 500 msec window. This should be easy to try.
But even without FIR, I bet we could improve on your current results. Drop in this thread and start reading from post 206 onwards:
http://www.diyaudio.com/forums/full-range/275730-convolution-based-alternative-electrical-loudspeaker-correction-networks-21.html#post4548875
I bet this method Greg mentions could make an improvement your current EQ use.
It will not be as powerful as full FIR correction but I bet it will sound more right than EQ-ing a 500 msec window. This should be easy to try.
Not really. The simple fact that the FR of one drive is vastly different than 24 confirms the phase cancellations and offset from quadrature exist.
Now I'm not talking about edge diffraction so I'm not sure what you mean by combing at the baffle edges?
Yes.......in the vertical the floor and ceiling reflections are basically completely avoided which is quite evident in the modal region response.....very smooth where no other alignment can compare so agreed.
But there's a lot more to line array pros and cons to discuss here where isn't the point of my posting in Wesayso's thread.
Combing is a physical and audible reality. You can reduce it, ignore it, whatever......but you can't negate it sadly. We DON'T all hear the same RA7 and some of us are more or less sensitive to certain anomalies.
I'm glad you enjoy your arrays and I applaud your efforts. I've wanted to try a corner load system for a long long time as the benefits are simply self evident by design. Good work!
The difference in the response comes in the HF when the 3.5" drivers start beaming. They don't add in phase, but the random phase additions of the array and the floor and ceiling reflections of the array cause the combing to be inaudible.
Maybe you're right, mayhem. Maybe we do hear differently and there is something to be said about that. But line arrays weren't invented today. There are a lot of folks out here (diyaudio) and out there with arrays. Keele's CBT also exists. It's a proven concept.
And I've never argued against arrays.......I've been involved with line array theory for nearly 25 years. But just as you indicated, multiple drivers whose wavelengths exceed their diameters and are cylindrical are problematic when arrayed.........period. You can't lump Keele's CBT and yours together.......VASTLY different animals and so are the pro arrays I've worked with/on.
Like I mentioned earlier.........I'm genuinely happy and in awe of your efforts in your project and others like it. I'm not here to tear down your work or Wesayso's. But let's not ignore physics in the process either. Subjectively, the speaker systems meet your needs and are enjoyable to listen to.........not much else to say is there?
Like I mentioned earlier.........I'm genuinely happy and in awe of your efforts in your project and others like it. I'm not here to tear down your work or Wesayso's. But let's not ignore physics in the process either. Subjectively, the speaker systems meet your needs and are enjoyable to listen to.........not much else to say is there?
Well I'm pretty much heavily into the Rock category and build these arrays to please. Here's a preliminary list of things I tried so far:
There are more not mentioned here but does this cover some of the category you call "complex electric Rock"?
I'd love to see those measurements, as I think I have an idea what went wrong. I did not get these measurements in one day. And I did make wrong decisions along the way which could result in bad impressions like the one you mention.
So for now I'd have to side up with xrk971 and ask to see those measurements.
What I think went wrong for you is the correction of things that shouldn't have been corrected. While this is a simple sentence it is way more complicated to achieve.
Measurements are hard to interpret sometimes. I could have used brute force to force my measurements into shape with FIR filters to make them look good.
And get real bad sound in the process. But I don't think those measurements would resemble what I've shown here. If you read my entire thread about the correction there is one key point to notice that I tried to achieve: Find the bare minimum correction with FIR leading to the best results possible.
It took me literary months to find it. Many different factors play a role here. But one of the most important things is to use a short window to correct the response. Especially in mid and low frequencies. The area where the room starts to dominate.
Do you still have the mentioned arrays? I bet we could work on the results and come out on top.
If there is one thing I've learned it's measurements are the best indication of the perceived sound. Way more reliable than my ears alone.
But you cannot use brute force and expect good results. So far these arrays play every genre I tried. But I did not get there overnight.
Sadly since I deemed the project a fail, nothing was archived. Even the arrays themselves now serve up PA use in a V.F.W. Hall which they were donated to.....a much larger space where they serve up vocal intelligibility like no other!
But I should add some more details as to help you identify where my design failed as compared to yours. These were 2nd order high passed at 120hz actively and only a impedance flattening circuit was used........no DSP of any kind....only a pair of mono 31 band parametric eQs which along with room placement and some treatments for first reflections allowed me to dial in a very smooth FR. Again, all analog, no DA to AD to DA conversions ( another topic best left alone for now). Off axis 5,15,30,45......Polars very similiar to yours horizontally. Never bothered vertically which was the point of an array in the first place.
Again, we're subjectively speaking here at least from my point of view. If you don't agree that measurements are significantly lacking in representing complex music reproduction, our viewpoints might be too far off to find any consensus. And that's ok........it's what makes the world go round and fuels scientific discovery.
Yes, I do think we are further apart here than you thought. Because over the course of this project I've come to believe that the measurements do tell us all about the listening experience. If you are willing to look hard enough.
I've learned a lot doing this project and invested a huge amount of time into that study. And the more I learn, the better I could relate what I hear to what I measure.
But you were talking about Frequency Response. What about the other graphs I showed?
I try to show them all (well almost all, maybe it's time to let go of my pride an cough up the early waterfall plot as well) as they are measured.
An impulse is much more than most take it for. Dissect it and it contains a wealth of information. That's exactly what I did to learn how to really "read" it.
No AD DA steps here, only one DA by the way. All music is stored digitally.
Remember: the real magic is in the music! Our job is to let it out the box!
I've learned a lot doing this project and invested a huge amount of time into that study. And the more I learn, the better I could relate what I hear to what I measure.
But you were talking about Frequency Response. What about the other graphs I showed?
I try to show them all (well almost all, maybe it's time to let go of my pride an cough up the early waterfall plot as well) as they are measured.
An impulse is much more than most take it for. Dissect it and it contains a wealth of information. That's exactly what I did to learn how to really "read" it.
No AD DA steps here, only one DA by the way. All music is stored digitally.
Remember: the real magic is in the music! Our job is to let it out the box!
.......and I'm not faulting the measurements either. Excellent work from concept to completion to measuring. What a wonderful world it would be if that's all there was to it.
Again, your own work with fullrange drivers has led you to some conclusions based on measurements and experience. The little Vifas and similiar have been on my workbench for a long time and they offer excellent value for their purpose
.......but it's the subjective that can't be overlooked. I don't doubt that Wesayso's arrays after multiple iterations of DSP would sound different from what I built. There's also physical differences such as the aluminum baffle and enclosure differences to consider as well. I concede all that. I just won't concede that there's correction available for the destructive combining of multiple drive units across such a wide FR. If you disagree, then we'll just have to accept eachother's opinion and move on to more positive discussions
I've followed your DIY efforts and enjoyed you busting some audiophile myths.......but this is a subjective hobby so when we get to the core of the discussions.......we like what we like and that's all there is to it. The audible effects of DSP and ADA conversions are of particular interest to me these days and i seem to be one of the very few who aren't as excited with the capabilities as I've been able to audibly detect the effects of these processes. My first go round was a complete accident where a simple native 44.1 to 48 conversion was clearly audible to me with ringing on solo violins and piano with cymbals being the most objectionable. I've also found the PC to be a very noisy environment for DSP and ADA processes inserting measurable and audible jitter all along the signal chain. But another topic for a different forum......
Sorry, but I disagree with the word "problematic." What is so problematic about it? If you sit far enough from the array, it simply is not a problem. Consider this: when a woofer is crossed to a tweeter in a traditional setup, there are lobes. At certain angles, away from the design axis, the phase will get 180 degrees out of sync and there will be cancellation. By designing the lobes to point away from our ears, we avoid the problem. Now consider the line array. Two nearby units are radiating the same frequency (just like a woofer and tweeter in a crossover) and there will be cancellation at some angles and summation at others. Now add multiple units to the equation, and their reflections from the floor and ceiling. If you do sit ups, you will hear the swish-swish from cancellations and summations. But if you sit in one place, and are far enough away from the array and in the vertical dimension of the array, then the random additions and cancellations don't matter so much. What matters is what is result of the random additions/cancellations at that point. If you can EQ it flat at that point, that's good enough. You have avoided the problem. And all this is just for the region where the drivers start beaming.
Finally, Keele's CBT array. Shorter arrays need more work to come close to replicating the performance of floor-to-ceiling arrays. Shorter arrays need shading, or need progressive roll off of drivers father away from the center of the array to maintain the constant beamwidth. They may be vastly different animals but they are all the same species.
But you see.........without insult......you're not as open minded as your writing and work attempts to suggest. From the beginning of this discussion which came about by chance of me searching related topics of mechanical manipulation of wavefronts........you immediately dispelled such as resultant potential diffraction........something that SL has studied and JBL implementing.......based on what?........experience and measurements of your arrays?
Sorry.......let me apologize in advance. I've been in this hobby for maybe far too many years to get grouchy now! I'm always looking for new and better ways to accomplish something.
To some it up, I just can't throw the baby out with the bath water if you're familiar with the expression. It's probobly best that I just simply and empirically state that I've moved on from all iterations of the line array as a best practice for home audio within typical domestic spaces. These discussions were based on curiosity of mechanical alterations of wavefronts and not intended for the improvement of full range line arrays.
.......and I'm not faulting the measurements either. Excellent work from concept to completion to measuring. What a wonderful world it would be if that's all there was to it.
Again, your own work with fullrange drivers has led you to some conclusions based on measurements and experience. The little Vifas and similiar have been on my workbench for a long time and they offer excellent value for their purpose
.......but it's the subjective that can't be overlooked. I don't doubt that Wesayso's arrays after multiple iterations of DSP would sound different from what I built. There's also physical differences such as the aluminum baffle and enclosure differences to consider as well. I concede all that. I just won't concede that there's correction available for the destructive combining of multiple drive units across such a wide FR. If you disagree, then we'll just have to accept eachother's opinion and move on to more positive discussions
I've followed your DIY efforts and enjoyed you busting some audiophile myths.......but this is a subjective hobby so when we get to the core of the discussions.......we like what we like and that's all there is to it. The audible effects of DSP and ADA conversions are of particular interest to me these days and i seem to be one of the very few who aren't as excited with the capabilities as I've been able to audibly detect the effects of these processes. My first go round was a complete accident where a simple native 44.1 to 48 conversion was clearly audible to me with ringing on solo violins and piano with cymbals being the most objectionable. I've also found the PC to be a very noisy environment for DSP and ADA processes inserting measurable and audible jitter all along the signal chain. But another topic for a different forum......
Not sure if you are referring to my setups but I don't have a PC in the chain. MiniDSP handles all the ADC and DAC. PC is used to program and then disconnected. I think native bit rate inside miniDSP is 24bit and 48kHz. When I make samples of sound clips I have been using 48k and 320kbit/Sec for MP3 as a practical reality of getting a sound clip to fit in a 45second file size limit with higher quality.
.........but you can't design 48 lobes on a forward radiating alignment to point 'away' from the listening axis.......you can only reduce the number and amplitude of the lobes.......the more there are, the more destructive in nature.
Again, I'll agree with the advantages of the floor and ceiling but there's plenty of ways to skin that cat without the drawbacks of destructive combing. If moving to the transitional midfield for critical listening is an acceptable solution to you and your practicalities, then ok........but I would concede that I would do it differently but that's not the point of these discussions.
All speaker designs are flawed to some degree.....there is no perfect system or transducer or alignment or..............I've come to accept that. What you've accomplished is nothing short of fantastic applying a quality driver in multiples along with Murphy's work to build a system in your space that fits your subjective tastes. Do we have to make it more than what it is? Do you want me to say that full range arrays and complex DSP are equivalent to the discovery of the wheel? Of course not. The design is flawed just as every design. To what magnitude has been discussed in every book, white paper and forum in creation. To what more can we add?;