I don't get much out of wavelet view. Spectrogram plot with short window (2ms) shows coherent reflections as continuous horizontal lines. Path length difference to microphone is seen on time axis. Based on known microphone and speaker location and timing of reflections is relatively easy to figure out where reflections are coming from.
With tall array, floor and ceiling reflections most likely smeared into near oblivion. Conversely, horizontal reflections from walls will remain highly coherent.
I guess that your listening position is close to back wall, likely less than 4ft. Such reflections can wreak havoc with imaging depth and width.
Here are spectrogram examples with my Linkwitz Pluto based speaker. Speaker was >5ft from walls, tweeter is 43 inches from floor and 53 inches from ceiling. Microphone is also 43 inches from floor. Floor is wall to wall wool carpet on thin foam padding. Ceiling is thin acoustic tiles attached to sheet-rock. Ceiling has several beams that hang 9 inches below tiles that act as wall/ceiling bondaries at shorter wavelengths. Walls are wood, brick, and lots of big glass windows. Microphone at 9 inches:
Cursor location is about 7ms, the horizontal line is reflection from ceiling/beam combination. Amplitude is -39dB to direct sound. Diffuse blob in lower left of plot is artifact of 2ms windowing. At 9 inches direct/reflected ratio is very high. Listening in stereo with speakers <2ft apart, and sitting quite close, the imaging is pinpoint. Front of phantom image starts just about a foot behind speaker and extends about 8-10 feet back. Image height is short.
Microphone at 42 inches:
Note the scales of these plots; for 9 inch measurement scale is 0dB to -74dB, and for 42 inch measurement it is 0dB to -42dB. The direct/reflected ratio at 42 inches is much lower than for 9 inch measurement. For 42 inch measurement cursor is at about 5ms, this is ceiling reflection with less ceiling beam influence. This reflection is -12dB to direct sound. Horizontal line at 4ms is floor bounce. Combination of off axis response and carpet attenuate much of floor reflection. Reflection at 12ms is combination of side wall reflections. Listening with speakers placed 4ft-5ft apart at about same distance back the image starts about 4ft behind speakers and extends 20-40 ft deep. Image height is natural sounding; symphonic recordings sound fantastic. From this listening setup if speakers are turned around to point at front wall, placing me behind speakers, image starts at about front wall (12ft from LP) and extends to way outside the house.
Speaker directivity and placement have impact that can not be underestimated. I built radial array for true omni-directional behavior in horizontal plane specifically to explore this. With radial arrays spaced 5-6ft apart and away from walls imaging has fixed behavior dependent on listening distance. From directly between speakers sound is like giant headphones with lots of sound inside head. As position moves back an image forms directly in front of face. As distance increases, image moves proportionately behind speakers, with proportionally increasing image depth. Image properties become stable beyond listening distance of about 7ft.
Attached are RT60 and water fall plots comparing 42 inch measurement of Pluto speaker to X's TC9 from early subjective round. These are good to think about in relationship to spectrograms and speaker/room interactions.
With tall array, floor and ceiling reflections most likely smeared into near oblivion. Conversely, horizontal reflections from walls will remain highly coherent.
I guess that your listening position is close to back wall, likely less than 4ft. Such reflections can wreak havoc with imaging depth and width.
Here are spectrogram examples with my Linkwitz Pluto based speaker. Speaker was >5ft from walls, tweeter is 43 inches from floor and 53 inches from ceiling. Microphone is also 43 inches from floor. Floor is wall to wall wool carpet on thin foam padding. Ceiling is thin acoustic tiles attached to sheet-rock. Ceiling has several beams that hang 9 inches below tiles that act as wall/ceiling bondaries at shorter wavelengths. Walls are wood, brick, and lots of big glass windows. Microphone at 9 inches:
Cursor location is about 7ms, the horizontal line is reflection from ceiling/beam combination. Amplitude is -39dB to direct sound. Diffuse blob in lower left of plot is artifact of 2ms windowing. At 9 inches direct/reflected ratio is very high. Listening in stereo with speakers <2ft apart, and sitting quite close, the imaging is pinpoint. Front of phantom image starts just about a foot behind speaker and extends about 8-10 feet back. Image height is short.
Microphone at 42 inches:
Note the scales of these plots; for 9 inch measurement scale is 0dB to -74dB, and for 42 inch measurement it is 0dB to -42dB. The direct/reflected ratio at 42 inches is much lower than for 9 inch measurement. For 42 inch measurement cursor is at about 5ms, this is ceiling reflection with less ceiling beam influence. This reflection is -12dB to direct sound. Horizontal line at 4ms is floor bounce. Combination of off axis response and carpet attenuate much of floor reflection. Reflection at 12ms is combination of side wall reflections. Listening with speakers placed 4ft-5ft apart at about same distance back the image starts about 4ft behind speakers and extends 20-40 ft deep. Image height is natural sounding; symphonic recordings sound fantastic. From this listening setup if speakers are turned around to point at front wall, placing me behind speakers, image starts at about front wall (12ft from LP) and extends to way outside the house.
Speaker directivity and placement have impact that can not be underestimated. I built radial array for true omni-directional behavior in horizontal plane specifically to explore this. With radial arrays spaced 5-6ft apart and away from walls imaging has fixed behavior dependent on listening distance. From directly between speakers sound is like giant headphones with lots of sound inside head. As position moves back an image forms directly in front of face. As distance increases, image moves proportionately behind speakers, with proportionally increasing image depth. Image properties become stable beyond listening distance of about 7ft.
Attached are RT60 and water fall plots comparing 42 inch measurement of Pluto speaker to X's TC9 from early subjective round. These are good to think about in relationship to spectrograms and speaker/room interactions.
Attachments
Thanks Barleywater,
You are correct, the wall behind the microphone is relatively close. If it weren't a family room the speakers would be at a very different place too and along the short wall. But these are the constraints I have to live with.
The wall behind the speakers is where I did hang my damping panel/poster. I also drape cushions between the panel and the couch for measurements and critical listening.
By a bit more toe-in I have all reflections in the first 20 ms down by 20 dB or more judging the Filtered IR. The biggest one was showing up at ~6 ms
As said before, my ceiling is thin metal and rattles if I play my electric guitar. I also hear it during measurements and try to stay within an acceptable range.
I just looked at the results from Halair (again, as it's my only reference to another array), it shows similar "hair" as I have after the impulse. I'm not sure if this measurement was after hanging all his damping panels. I think it's from before installing the panels.
I do not have as clear horizontal lines in my spectrogram. I tried to put damping at the first reflection points where applicable.
Here you see the trouble I still have in the high frequencies. Some of it from comb filtering and some are resonances perhaps? Not sure yet. Maybe the baffle needs more damping between baffle and enclosure with butyl like I have between the aluminium.
My waterfall looks a bit cleaner than the one from X's first subjective round (different scale here in favour of mine) though I'm not pleased with the high frequency part yet (not showing it publicly just yet) see the yellow blob in the spectrogram at ~6 - 7 KHz above.
The midrange looks like this:
Most measurements I see published are taken at close range. All my measurements are at 3 meter distance at the listening position. Hard to compare them sometimes. My RT60 still shows the trouble in the 100-200 Hz range (right speaker) but is reasonable from 200 Hz and above. Much better than before I had the 3 big panels installed.
Again taken at the listening/measurement position:
This is for the stereo pair. I can't hang up more damping without getting into trouble though I still want to put something behind the right speaker. I'll have to beg to make that happen though.
At some point I do hope to replace the ceiling.
You are correct, the wall behind the microphone is relatively close. If it weren't a family room the speakers would be at a very different place too and along the short wall. But these are the constraints I have to live with.
The wall behind the speakers is where I did hang my damping panel/poster. I also drape cushions between the panel and the couch for measurements and critical listening.
By a bit more toe-in I have all reflections in the first 20 ms down by 20 dB or more judging the Filtered IR. The biggest one was showing up at ~6 ms
As said before, my ceiling is thin metal and rattles if I play my electric guitar. I also hear it during measurements and try to stay within an acceptable range.
I just looked at the results from Halair (again, as it's my only reference to another array), it shows similar "hair" as I have after the impulse. I'm not sure if this measurement was after hanging all his damping panels. I think it's from before installing the panels.
I do not have as clear horizontal lines in my spectrogram. I tried to put damping at the first reflection points where applicable.
Here you see the trouble I still have in the high frequencies. Some of it from comb filtering and some are resonances perhaps? Not sure yet. Maybe the baffle needs more damping between baffle and enclosure with butyl like I have between the aluminium.
My waterfall looks a bit cleaner than the one from X's first subjective round (different scale here in favour of mine) though I'm not pleased with the high frequency part yet (not showing it publicly just yet) see the yellow blob in the spectrogram at ~6 - 7 KHz above.
The midrange looks like this:
Most measurements I see published are taken at close range. All my measurements are at 3 meter distance at the listening position. Hard to compare them sometimes. My RT60 still shows the trouble in the 100-200 Hz range (right speaker) but is reasonable from 200 Hz and above. Much better than before I had the 3 big panels installed.
Again taken at the listening/measurement position:
This is for the stereo pair. I can't hang up more damping without getting into trouble though I still want to put something behind the right speaker. I'll have to beg to make that happen though.
At some point I do hope to replace the ceiling.
Last edited:
I've no doubt that you have fantastic sound. Indeed, 3m measurement that has microphone close to back wall yet produces such waterfall, spectrogram and tight IR on which all plots are based, are best indicators I have on this end to what words alone can not convey. Perhaps some day I will have resources for such a grand project. Surely would like to find similar line array locally that I could audition.
With 12 driver radial array each driver is 30 degrees apart around the circle. Detailed plots at 2.5 degree intervals obtained by rotating the array between measurements shows some crazy ripple that is completely inaudible listening to music. I guess that 25 driver line array has much less ripple, also completely inaudible.
Keep the music playing. I've found with great speakers, lots of recordings show faults, sometimes tragically. Many recordings are helped with a little custom EQ, settings that can be saved, or applied to copy of recording for play.
With 12 driver radial array each driver is 30 degrees apart around the circle. Detailed plots at 2.5 degree intervals obtained by rotating the array between measurements shows some crazy ripple that is completely inaudible listening to music. I guess that 25 driver line array has much less ripple, also completely inaudible.
Keep the music playing. I've found with great speakers, lots of recordings show faults, sometimes tragically. Many recordings are helped with a little custom EQ, settings that can be saved, or applied to copy of recording for play.
I agree with you on the faults in recordings becoming more evident the further you improve. Though thankfully a lot of material is quite enjoyable. Some songs I have sound awful on one CD and yet another pressing or release can be beautiful.
I especially like the older stuff where the recording room/environment is quite evident.
I always remembered my old Led Zeppelin LP's fondly though they are stored in the attic at this moment. It wasn't until the recent 2015 re-masters that I've been able to truly enjoy them in a digital release.
When I was working on my car stereo I noticed I fell in the trap as to only play certain songs that would have the best imaging/stage etc. But the key point was to enjoy the music I actually love. So I worked on tonality for a while and let the best imaging concept slide a bit and started enjoying playing music again.
The same mode I had in the car: over critical to every song I play I now have in my house. And in the car it doesn't worry me anymore. So I'll know I won't be able to really start enjoying the system when I quit working on it. But it's too much of a challenge not to do it! I do enjoy watching movies with my Son, whatever he wants to watch, I just enjoy the effects I hear in the mean time sitting beside him with a smile (lol).
Before I started with the Array idea I was focussed on Omni speakers and considered building a Pluto like clone. I won't show the horrid design, it was like B&W meets Linkwitz Pluto. It looked like a big kettle (sub in foot) with a strange flute on top... So glad I didn't pursue that one!
By the way, Spectogram updated from 5 ms window to 2 ms like your plot(s)... it does show the lines (reflections) a bit more clear that way.
I especially like the older stuff where the recording room/environment is quite evident.
I always remembered my old Led Zeppelin LP's fondly though they are stored in the attic at this moment. It wasn't until the recent 2015 re-masters that I've been able to truly enjoy them in a digital release.
When I was working on my car stereo I noticed I fell in the trap as to only play certain songs that would have the best imaging/stage etc. But the key point was to enjoy the music I actually love. So I worked on tonality for a while and let the best imaging concept slide a bit and started enjoying playing music again.
The same mode I had in the car: over critical to every song I play I now have in my house. And in the car it doesn't worry me anymore. So I'll know I won't be able to really start enjoying the system when I quit working on it. But it's too much of a challenge not to do it! I do enjoy watching movies with my Son, whatever he wants to watch, I just enjoy the effects I hear in the mean time sitting beside him with a smile (lol).
Before I started with the Array idea I was focussed on Omni speakers and considered building a Pluto like clone. I won't show the horrid design, it was like B&W meets Linkwitz Pluto. It looked like a big kettle (sub in foot) with a strange flute on top... So glad I didn't pursue that one!
By the way, Spectogram updated from 5 ms window to 2 ms like your plot(s)... it does show the lines (reflections) a bit more clear that way.
Last edited:
Wesayso,
Your are using FIR for equalization so, your arrays has to be time & phase coherent, right? But what about phase performance of your arrays without any FIR processing?
I am curious because my arrays has phase variation peaks within +- 90 degrees upto about 5KHZ and beyond that phase changes to around -700 degrees.
This is true for my arrays with IIR PEQ as well as without any EQ.
Aren't full range speakers supposed to be phase coherent? 😕
I am using REW, so maybe there are some mistakes in my measurement techniques.
Again, your RT60 plots are pretty good for the lower frequencies as well.
What type of damping are you using? any bass traps?
Your are using FIR for equalization so, your arrays has to be time & phase coherent, right? But what about phase performance of your arrays without any FIR processing?
I am curious because my arrays has phase variation peaks within +- 90 degrees upto about 5KHZ and beyond that phase changes to around -700 degrees.
This is true for my arrays with IIR PEQ as well as without any EQ.
Aren't full range speakers supposed to be phase coherent? 😕
I am using REW, so maybe there are some mistakes in my measurement techniques.
Again, your RT60 plots are pretty good for the lower frequencies as well.
What type of damping are you using? any bass traps?
You can easily think that, yes... FIR is going to straighten everything out.
But that isn't going to sound very pleasant. We are measuring in one spot in the room, if you're going to correct the response in that one spot with a long window you can correct everything in that one spot. But I'd say try it.
I've done just that and it sounds weird. Some parts are pleasing, other parts are horrifying. The sound in the rest of the room is going to be off.
The reason it takes me this long to get anywhere is to try and find the optimum window that's going to correct most faults while not messing up the rest of the room. I want to correct the smallest window I can with the biggest benefits.
I bet if you listen to the arrays they are pleasing to hear, very dynamic sounding, clean in the midrange, right? And yet they probably will be quite messy in a measurement still. Can't know for sure, I don't have all the answers either, but this is based on my results and the file I saw from Halair.
If you are looking at the phase, what are you looking at? What window in REW? If it's the standard 500 ms window it's going to be all messed up by the room. You aren't looking at the arrays alone, which is a big problem as is but all reflections and resonances in the room are going to add to that result.
A single full range driver has a shot at being reasonable in phase behaviour.
But we aren't using a single big driver but 25 separate sources. We want those sources to act as one big full range. But all of them separately are going to have some problem of their own, not to mention they are playing in a room.
My main concern is the first wave front hitting the listening spot. Many ways to look at that so that's what I do. What is the first wave front? Is it the first ~ 20 to 30 ms? Or is it the first couple of cycles that we perceive as the signal is being played. My guess is a little of both.
First thing to do is get rid of as many first reflections as possible. You can do this by placing the arrays far from walls. Or you'll have to absorb some of those early reflections or make them go anywhere but the listening spot.
Easier said than done. The front of my arrays is 50 cm from the back wall. The shape of my arrays is chosen to have the least amount of diffraction edges within reason to keep them as slim towers.
Measure the IR of a speaker and look at the spikes after the big one. Those are the first reflections. My goal was to have those spikes at least 20 dB down from the initial impulse in the first 20 ms. More is better here. You can check that by looking at the Filtered IR. To get there I build 3 damping panels.
One huge panel 2.30 m x 0.6 m directly behind a curtain next to my left array.
It's 7 cm thick and 2 cm away from the wall. All I could fit behind the curtain that was there already. The damping material is fibre glass insulation between two layers of 0,5 cm real wool felt. Covered with a fitted sheet made of terry cloth.
That still gave me a peak from behind the listening position. So that's where the poster comes in. Basically the same recipe but 5 cm thick with one layer of wool felt on the back side. The cover is printed Kona cotton as per recipe from another array owner, koldby.
It helped out the room by bringing the ~250 Hz and up within the 500 ms window.
My speaker placement is asymmetrical. On one side (the left) I have a wall. The right side is free, there's a chimney but the front of the array is sticking out past that wall. So looking around I figured I'd get reflections from the right array firing in the left corner next to the listening position and coming back to me. The left side doesn't have that problem. So that's where the last damping panel is. Measuring 2.30 m x 0.7 m and very similar to the first described panel. Except this time a cotton sheet as cover as it was no fun working with the very elastic terry cloth.
After that, all the problems that remain are up to the FIR correction to be solved. Ideally I would want absorbers behind my speakers and on that chimney wall beside the right array. No chance of getting permission to do that!
I may be typing a lot more than you asked for, but I just wanted to tell the story as complete as I can. This is what I could do to the room to improve my measurements. I have a sheet steel ceiling (very bad!) and a wooden floor, with a rug on it in the center. So it's never going to be perfect.
But that isn't going to sound very pleasant. We are measuring in one spot in the room, if you're going to correct the response in that one spot with a long window you can correct everything in that one spot. But I'd say try it.
I've done just that and it sounds weird. Some parts are pleasing, other parts are horrifying. The sound in the rest of the room is going to be off.
The reason it takes me this long to get anywhere is to try and find the optimum window that's going to correct most faults while not messing up the rest of the room. I want to correct the smallest window I can with the biggest benefits.
I bet if you listen to the arrays they are pleasing to hear, very dynamic sounding, clean in the midrange, right? And yet they probably will be quite messy in a measurement still. Can't know for sure, I don't have all the answers either, but this is based on my results and the file I saw from Halair.
If you are looking at the phase, what are you looking at? What window in REW? If it's the standard 500 ms window it's going to be all messed up by the room. You aren't looking at the arrays alone, which is a big problem as is but all reflections and resonances in the room are going to add to that result.
A single full range driver has a shot at being reasonable in phase behaviour.
But we aren't using a single big driver but 25 separate sources. We want those sources to act as one big full range. But all of them separately are going to have some problem of their own, not to mention they are playing in a room.
My main concern is the first wave front hitting the listening spot. Many ways to look at that so that's what I do. What is the first wave front? Is it the first ~ 20 to 30 ms? Or is it the first couple of cycles that we perceive as the signal is being played. My guess is a little of both.
First thing to do is get rid of as many first reflections as possible. You can do this by placing the arrays far from walls. Or you'll have to absorb some of those early reflections or make them go anywhere but the listening spot.
Easier said than done. The front of my arrays is 50 cm from the back wall. The shape of my arrays is chosen to have the least amount of diffraction edges within reason to keep them as slim towers.
Measure the IR of a speaker and look at the spikes after the big one. Those are the first reflections. My goal was to have those spikes at least 20 dB down from the initial impulse in the first 20 ms. More is better here. You can check that by looking at the Filtered IR. To get there I build 3 damping panels.
One huge panel 2.30 m x 0.6 m directly behind a curtain next to my left array.
It's 7 cm thick and 2 cm away from the wall. All I could fit behind the curtain that was there already. The damping material is fibre glass insulation between two layers of 0,5 cm real wool felt. Covered with a fitted sheet made of terry cloth.
That still gave me a peak from behind the listening position. So that's where the poster comes in. Basically the same recipe but 5 cm thick with one layer of wool felt on the back side. The cover is printed Kona cotton as per recipe from another array owner, koldby.
It helped out the room by bringing the ~250 Hz and up within the 500 ms window.
My speaker placement is asymmetrical. On one side (the left) I have a wall. The right side is free, there's a chimney but the front of the array is sticking out past that wall. So looking around I figured I'd get reflections from the right array firing in the left corner next to the listening position and coming back to me. The left side doesn't have that problem. So that's where the last damping panel is. Measuring 2.30 m x 0.7 m and very similar to the first described panel. Except this time a cotton sheet as cover as it was no fun working with the very elastic terry cloth.
After that, all the problems that remain are up to the FIR correction to be solved. Ideally I would want absorbers behind my speakers and on that chimney wall beside the right array. No chance of getting permission to do that!
I may be typing a lot more than you asked for, but I just wanted to tell the story as complete as I can. This is what I could do to the room to improve my measurements. I have a sheet steel ceiling (very bad!) and a wooden floor, with a rug on it in the center. So it's never going to be perfect.
Man, I need some of this wall treatment in my lively speaker lab room. 🙂
Framing, fiberglass, is there a substitute for wool felt? That stuff is pricey.
Framing, fiberglass, is there a substitute for wool felt? That stuff is pricey.
Okay, next part.
In the above story I've said what my goals were in relation to the room. Getting the first reflections down by 20 dB. I still have early reflections but they are no where near as bad as what I started with. Clever toe in aiming at the poster helped there to get the last bit down under the 20 dB limit. But toe in/toe out is a huge difference in sound stage. So play with it!
Here are my graphs, all with a 4 cycle window in REW in various stages:
You see viewing it like this ain't that bad. The smoothing is automatic in REW, I would have liked a little less smoothing to really see what's going on.
Next I apply my famous pré EQ:
Notice how soft handed the EQ is? All low q boosts and cuts to get a somewhat flat target. No big changes in the phase. I'd rather use no pré EQ at all but I'm doing this to get the response within a reasonable window for DRC.
After that my FIR recipe comes in:
Subject to change but I'm quite pleased with the result. But we're still looking at smoothed plots. The reality is going to be a bit more ragged.
Here's a plot of the first 20 ms, no cycle gating and smoothed 1/12:
This is the right side, no damping panels near so it is the most rough plot.
Still you see it's reasonable with a lot of phase wrap at higher frequencies. And I assume this is what you were getting at?
I've only seen measurements from one room that didn't have the wraps and I'll never be able to replicate that! But it had a single tweeter and not 25 sound sources. Maybe a horn speaker could stand more chance as it wouldn't illuminate the room like these arrays do.
Still I'm working to better the results, but DRC isn't going to help here. The window used in DRC is much to small to solve that. So it's back to looking at the room and the speakers for the best cure.
I don't know what kind of window is needed at high frequencies to represent what we hear. The first sound it probably going to dominate so I look at the output gated in cycles. But I still look at the first 20ms too to try and improve. The balance between direct sound and reflected sound is also interesting. So I compare
what I get with a 0 to 20 ms window and a 20 to 500 ms window. They are very much alike which I think is a good sign.
In the above story I've said what my goals were in relation to the room. Getting the first reflections down by 20 dB. I still have early reflections but they are no where near as bad as what I started with. Clever toe in aiming at the poster helped there to get the last bit down under the 20 dB limit. But toe in/toe out is a huge difference in sound stage. So play with it!
Here are my graphs, all with a 4 cycle window in REW in various stages:
You see viewing it like this ain't that bad. The smoothing is automatic in REW, I would have liked a little less smoothing to really see what's going on.
Next I apply my famous pré EQ:
Notice how soft handed the EQ is? All low q boosts and cuts to get a somewhat flat target. No big changes in the phase. I'd rather use no pré EQ at all but I'm doing this to get the response within a reasonable window for DRC.
After that my FIR recipe comes in:
Subject to change but I'm quite pleased with the result. But we're still looking at smoothed plots. The reality is going to be a bit more ragged.
Here's a plot of the first 20 ms, no cycle gating and smoothed 1/12:
This is the right side, no damping panels near so it is the most rough plot.
Still you see it's reasonable with a lot of phase wrap at higher frequencies. And I assume this is what you were getting at?
I've only seen measurements from one room that didn't have the wraps and I'll never be able to replicate that! But it had a single tweeter and not 25 sound sources. Maybe a horn speaker could stand more chance as it wouldn't illuminate the room like these arrays do.
Still I'm working to better the results, but DRC isn't going to help here. The window used in DRC is much to small to solve that. So it's back to looking at the room and the speakers for the best cure.
I don't know what kind of window is needed at high frequencies to represent what we hear. The first sound it probably going to dominate so I look at the output gated in cycles. But I still look at the first 20ms too to try and improve. The balance between direct sound and reflected sound is also interesting. So I compare
what I get with a 0 to 20 ms window and a 20 to 500 ms window. They are very much alike which I think is a good sign.
Last edited:
I stumbled over an add for real wool felt locally. It wasn't expensive and is the seal deal. I'd say the underlay carpet might help? But I'd use in on the back side only. The real wool felt made a huge difference inside my enclosure. I even have it as a shield between the fiberglass insulation and the speaker on the inside and along all internal walls. And it's only 5 mm to 7 mm in thickness.
I guess having different materials for the sound to go trough helps. You could leave more air space behind the panels, it really helps. I couldn't do that in my living room.
It will definitely improve your room! Preferably about 7 cm thick panels would be my goal.
F15 sound deadening 5/8in thick x 10 ft x 72in is $473 + $100 shopping. Ouch!
What grade did you use ? I am reading on it and SAE F13 to F15 seems to be recommended for sound absorption.
https://www.pricefalls.com/product/...Zq529cSo7GrJcCtcCf5D-CM241Lz8Q3mdgaAkFr8P8HAQ
What grade did you use ? I am reading on it and SAE F13 to F15 seems to be recommended for sound absorption.
https://www.pricefalls.com/product/...Zq529cSo7GrJcCtcCf5D-CM241Lz8Q3mdgaAkFr8P8HAQ
No grading available, this is a private person living in the Netherlands and collects this stuff from his home country, France.
Wolvilt 15 m2 op rol 5 mm dik - Te koop | Tweedehands.nl
It was pretty much a gamble but it played out well. It is real wool and even smells like sheep. I could not find it anywhere else at these prices. I had a friend collect it at the seller's home as he lived nearby.
He sells it as insulation material as an alternative for glass fibre insulation. It looks like the stuff you linked though, but that's not saying much of coarse. I got a roll of 1 m x 15 m x 5 mm at a fair price of about € 95,00. Much more than I needed so I had enough to experiment.
Wolvilt 15 m2 op rol 5 mm dik - Te koop | Tweedehands.nl
It was pretty much a gamble but it played out well. It is real wool and even smells like sheep. I could not find it anywhere else at these prices. I had a friend collect it at the seller's home as he lived nearby.
He sells it as insulation material as an alternative for glass fibre insulation. It looks like the stuff you linked though, but that's not saying much of coarse. I got a roll of 1 m x 15 m x 5 mm at a fair price of about € 95,00. Much more than I needed so I had enough to experiment.
I don't think the phase response you measure has any relation to reality, it is most likely an artifact from phase unwrapping produced by your software.
€6 per m^2 is an awesome price for 100% wool felt that is this thick. I have heard real wool makes a big difference in speaker boxes vs foam. The pro studios use BASF "Basotect" melamine reticulated cell foam. I wonder how that compares? I use small pieces inside speakers around driver chamber and it absorbs back reflections well. Available as Mr Clean eraser pads. Recently found at dollar store.
http://construction.basf.us/files/pdf/Basotect_brochure.pdf
I use REW, as many do, the question is what should we look for when measuring 25 or even 50 drivers all playing at the same time?
If it's not the real phase (it isn't when it's smoothed) I'm measuring than isn't it pointless to base a FIR filter upon that? Yet it seems to work. Can you elaborate?
I get it that I'm not measuring a big full range driver here, they are always going to be 25 separate sources. I do treat it as one big source. But in reality it's going to have more peaks and valleys close together where the drivers can't combine due to combing. But I'm not hearing that combing. I do see an advancement in my measurements but are you saying I'm looking for the wrong things? I'd like to learn here so teach me 🙂.
The FIR filter seems to straighten out the measurement, even though it will still be peaks and valleys, just a little more organised. The way I view it is with the smoothing I'm averaging the separate results into something useful that somewhat relates to how we hear. Am I wrong? I wouldn't know how much smoothing is needed exactly to relate to what I hear. The phase wrap origins seem to be traceable to reflections, resonances and other time disturbances. I don't look at the wraps themselves as being real, but try to view where they start and push that further out in time by either treating the room or speaker. In combination with the group delay I try to figure out the cause. Am I doing it wrong?
Last edited:
One thing you could do is also look at the phase with another program, Holmimpulse or ARTA come to mind.
I have lots early reflections; 700-800 ms+ decay for -10dB according to REW RT60 plot. Don't even ask about T30 or Topt 😡
This is where I am getting harsh & confusing high frequencies & low speech intelligibility performance from my arrays, I guess.
The phase plot in REW depends on a lot of factors, thats why mine might be so messed up. I am planning on adding some damping to my room, until then maybe I should not look at the phase plot as I have neglected waterfall data too.
Even if I may have some actual problems with the phase (which I doubt), there is little I can do as I use IIR PEQ. The group delay plot on the other hand, is basically a flat line through out the whole frequency range with a few peaks & dips ( 10ms max) in low & mid range.
I read somewhere in this forum, someone claiming that side wall damping treatments are somewhat more necessary for line arrays in comparison to point source speakers to deal with high amount of co-related side wall early reflections. Do you agree, based on your own experience?
I haven't had the opportunity for any room treatment so far. But, even with my horrible reverberant room, the line arrays were a lot fun as far as I remember 🙄
Are you done with your sub bass performance experiments?
Like having a fair idea or SPL vs cone excursion vs amplifier output.
For me its very important to test the limits of my system, even if for sake of specifications.
One problem, I am facing in the low frequencies is that the frequency response curve changes all over over the room. Maybe it varies less than a point source in comparison ( a room simulation in REW will give you an idea of point source Subwoofer SPL variation wrt listening position), but it still varies for different position across the room. Dips in one particular frequency at one listening position are peaks in another position. Now if I equalize it flat for my listening spot, peaks/dips may get far worse in another spot! What do I do? Use bass traps to minimize SPL variation? Equalize only for my listening spot & forget it? Or equalize for an average response for all possible listening spots across the whole room?
Low frequency equalization are so taxing on both excursion & amplifier power.
This is where I am getting harsh & confusing high frequencies & low speech intelligibility performance from my arrays, I guess.
The phase plot in REW depends on a lot of factors, thats why mine might be so messed up. I am planning on adding some damping to my room, until then maybe I should not look at the phase plot as I have neglected waterfall data too.
Even if I may have some actual problems with the phase (which I doubt), there is little I can do as I use IIR PEQ. The group delay plot on the other hand, is basically a flat line through out the whole frequency range with a few peaks & dips ( 10ms max) in low & mid range.
I read somewhere in this forum, someone claiming that side wall damping treatments are somewhat more necessary for line arrays in comparison to point source speakers to deal with high amount of co-related side wall early reflections. Do you agree, based on your own experience?
I haven't had the opportunity for any room treatment so far. But, even with my horrible reverberant room, the line arrays were a lot fun as far as I remember 🙄
Are you done with your sub bass performance experiments?
Like having a fair idea or SPL vs cone excursion vs amplifier output.
For me its very important to test the limits of my system, even if for sake of specifications.
One problem, I am facing in the low frequencies is that the frequency response curve changes all over over the room. Maybe it varies less than a point source in comparison ( a room simulation in REW will give you an idea of point source Subwoofer SPL variation wrt listening position), but it still varies for different position across the room. Dips in one particular frequency at one listening position are peaks in another position. Now if I equalize it flat for my listening spot, peaks/dips may get far worse in another spot! What do I do? Use bass traps to minimize SPL variation? Equalize only for my listening spot & forget it? Or equalize for an average response for all possible listening spots across the whole room?
Low frequency equalization are so taxing on both excursion & amplifier power.
I don't expect all those 36 degrees phase turns at the high end to be there in reality.
Now I get what you're talking about. And I agree. It isn't the phase that gets rotated in the true sense. But it does point to some form of time behaviour distortion. But it isn't as severe as REW makes it look. Browsing trough the IR and it's reflection peaks I can trace back most of these "rotations" in time.
I'm not viewing them as true phase wrap (though I called them that) but more as a point of interest, to see if I can improve upon it.
It reads like you have some work to do! Yes, I do believe that treating the side walls is very beneficial for these array type speakers. Try a mirror on the side wall and look from the listening area in that mirror to see what spot would be best to treat.
On the bass part, I guess I got lucky in my room as I have a very even bass response throughout the listening area. I described I equalise left and right differently to get to that point. Judging distortion and output I trade some bass between the two arrays. Left has a problem at ~70 Hz, so I EQ that down a bit and bring the right side up a bit. Outside of the listening area is a different story. Never measured it but I expect some bass is missing there compared to the mid frequencies. But that's way off axis in the other part of my room. I think what I see there is the line acting as a line source when you're close enough and gradually falling off in low frequencies (not enough length in the array) when I move to the sides. Does that make sense?
In your case I'd first treat the room with panels on first reflection points. After that if the bass is still off I'd go for some bass traps in corners if you can do that. Then I'd average my listening area with multiple measurements and EQ for that average in the lowest FR range.
Whatever you can do passively is always better than trying to do it all with EQ.
It will take some experimentation but I'm sure you'll get a huge improvement.
I was aiming for sub 500 ms times in RT60 and at least -20 dB in the first 20 ms.
Did you check out Halair's build? He has the arrays in a basement and before damping it was fun but messed up. He paled a lot of panels around the room and the improvement was huge!
http://www.diyaudio.com/forums/full-range/253367-halair-column-my-line-array-build-20-fr-drivers.html
and
http://www.diyaudio.com/forums/full-range/261157-room-treatment-full-range-line-array-rew-measurement.html
The threads cross each other a bit in the results. But inspiring non the less!