Corner Floor-to-Ceiling Line Array Using Vifa TC9

ra7

Member
Joined 2009
Paid Member
Short story: Corner line arrays can save space, can eliminate ceiling and front wall reflections, and can be fairly efficient. Using 24 Vifa TC9s, a 3.5" full range driver, a corner floor to ceiling line array was created. The array measures and sounds pretty awesome and achieves the design goals.

attachment.php


Long story: If you read Toole and other research, it says that the reflections that color the sound the most are the ceiling and front wall (the wall behind speakers). Reflections from side walls can add a sense of spaciousness but reflections from the ceiling and floor are perceived as modifying the frequency response.

Horns can be used to prevent ceiling and front wall reflections, but to control reflections at mid frequencies (sub 500 Hz) requires large horns. There are some active threads with folks trying to solve this problem by using cardoid cabinets for woofers, which is a pretty cool way of doing things, but there are problems.

I played with horns for a while and I could get good sound using Altec VOTT mid horns and constant directivity SEOS horns, but the sound was never quite of the peerless variety. The balance of the HF with the rest of the spectrum was never quite right. Horn speakers also tend to be big and occupy a lot of space and this annoyed me. And thus were born some design requirements. The new speaker should:
1. occupy less space,
2. control ceiling and front wall reflections, especially in the lower frequencies (below 500 Hz),
3. have a smooth on-axis response and uniform off-axis response and,
4. be fairly efficient, have low distortion, and have the dynamic headroom to hit 110 db peaks.

Toole kind of gives it away in his book that an ideal floor-to-ceiling line array would solve some of the key issues in sound reproduction in the home (figure 18.3). He describes the Keele CBT array as a practical implementation of an ideal floor-to-ceiling array with perfect drivers.

From here, putting the arrays in the corner was a no-brainer. It would completely get rid of the front wall reflection, and the floor-to-ceiling nature of the array would avoid ceiling reflections also. And I had a room that allowed a corner design.

The next question was the choice of drivers. Keele’s CBT uses a cone and a dome. But his arrays are relatively short. The floor-to-ceiling array would need a lot more drivers, meaning more cost, more wiring, and more worrying about unit to unit consistency.

If a smallish full range driver was to be used, the Vifa TC9 was an obvious choice. It has one of the smoothest frequency responses amongst full range drivers, a low distortion motor, and with Vifa manufacturing it, you could be relatively assured that there would be good unit to unit consistency. There are many designs here on diyAudio that use this driver. The Manzanita OB uses this driver. And many others have used it in their design. Seeing it used in Wesayso’s heroic tower build and other line array builds sealed the deal. Plus the TC9s are relatively cheap and I love cheap ;).

The construction is relatively straight forward. It’s a three-sided cabinet. The front baffle is as narrow as possible so that the drivers can be as close to the corner as possible. The drivers are flush with the front baffle. I had this design in my mind for at least a year but never got around to building it. Finally, after months of frustration at never finding the time to build the cabinets, I asked my friend John (carpenter) to build the cabinets and wire up the drivers.

John completed the build and the arrays are playing music. They occupy very little space, can go plenty loud, are pretty efficient and very low distortion (each driver is barely moving even at loud levels). Using a single 3.5” driver ensures a uniform off-axis response up to a high frequency, above which the Vifa does start to beam, but it is relatively well controlled. The on-axis response is flattened using DRC software, i.e., automatic room correction. The final result is quite good. Measurements coming up shortly. There is still a long way to go. The EQ is doing a pretty good job, but I think it can be better.
 

Attachments

  • IMG_1794.JPG
    IMG_1794.JPG
    77.3 KB · Views: 8,664
  • IMG_1797.JPG
    IMG_1797.JPG
    93.8 KB · Views: 3,971
  • IMG_1798.JPG
    IMG_1798.JPG
    100 KB · Views: 3,690
  • IMG_1800.JPG
    IMG_1800.JPG
    112.6 KB · Views: 3,632

ra7

Member
Joined 2009
Paid Member
Measurements

I wanted the drivers to be as close to the wall corner as possible and so that limited how wide the cabinet could be. And it was always going to be 7'-0" tall, the height of the ceiling. The cabinet dimensions sort of limit the bass extension. I can throw power at it and get it flat, but I am not doing that because I use a separate sub and am building two more at the moment. Besides, the resonant frequency of the TC9 is 125 Hz and it may not be a good idea to push it much below that region.

The measurement is made at the couch, just below and behind the projector. It is more than 12 feet away from the speaker. The HF has a 6 db boost past 4 kHz in all the uncorrected measurements. The uncorrected measurements show a pretty impressive impulse response. It is almost the perfect Dirac pulse. But one would expect that given the even frequency and nearly flat phase.

The uncorrected measurements are fed to DRC and the corrected response is convolved through JRiver. The corrected impulse does show some ringing. I suspect it is related to the measurement setup, or it could be that the subwoofer kicks in much before the array. Needs some investigating.

Uncorrected Left FR and phase:
attachment.php


Uncorrected Right FR and phase:

attachment.php


Uncorrected impulse response:
attachment.php


Corrected impulse response:
attachment.php


Corrected FR with sub 1/48th Octave smoothing:
attachment.php


Corrected FR with sub ERB smoothing in REW:
attachment.php
 

Attachments

  • L_R_Corr_Response.jpg
    L_R_Corr_Response.jpg
    72.1 KB · Views: 4,432
  • Impulse_Response_Corr_UnSmoothed.jpg
    Impulse_Response_Corr_UnSmoothed.jpg
    95.9 KB · Views: 4,839
  • Impulse_Response_Corr.jpg
    Impulse_Response_Corr.jpg
    69 KB · Views: 4,461
  • Impulse_Response_Uncorr.jpg
    Impulse_Response_Uncorr.jpg
    68.6 KB · Views: 4,474
  • Right_SPL_Phase.jpg
    Right_SPL_Phase.jpg
    91.7 KB · Views: 4,529
  • Left_SPL_Phase.jpg
    Left_SPL_Phase.jpg
    89.7 KB · Views: 4,575
Looks pretty impressive. But is the corrected IR response graphic right? L and R look a lot different from each other, like the L is actually the uncorrected maybe?

What DRC are you using? FIR based or IIR? And is it auto, or are you doing it manually with something like RePhase?

I've always seen best result from manual, but no experience with more modern setups like Dirac, maybe they do better. If you aren't already, I'd recommend eqing from relatively near to the speakers (so you are acting on the speakers, not on position sensitive room effects). You do not want to try to eq out reflections coming from behind you, particularly at above 400Hz or so.
 

ra7

Member
Joined 2009
Paid Member
@thunk303, Thanks. I know it looks bad (sort of) but it sounds fantastic. The response at the listening position is so smooth and "liquidy" and they do the "you are there" thing very well. DRC does a fantastic job at EQ.

Apart from the audiophile qualities, the one thing this speaker does that no other has done before is that it relaxes me. The music just draws me in and makes me pay less attention to all the nonsense we like to worry about.
 

ra7

Member
Joined 2009
Paid Member
Looks pretty impressive. But is the corrected IR response graphic right? L and R look a lot different from each other, like the L is actually the uncorrected maybe?

What DRC are you using? FIR based or IIR? And is it auto, or are you doing it manually with something like RePhase?

I've always seen best result from manual, but no experience with more modern setups like Dirac, maybe they do better. If you aren't already, I'd recommend eqing from relatively near to the speakers (so you are acting on the speakers, not on position sensitive room effects). You do not want to try to eq out reflections coming from behind you, particularly at above 400Hz or so.

Yes, it's correct. I know it looks bad and I do need to investigate. It is either the loop through (i.e., my convoluted measurement setup) or it is the subwoofer that does not have the right delay. But the proof is in the pudding, and the pudding tastes quite good :) I'll measure the arrays only (no sub) tomorrow and see what I get. The uncorrected measurement does not have the sub in it.

I am using DRC (DRC: Digital Room Correction), which is a free, auto correction software. It is quite brilliant. It uses a variable window for calculating the response to be corrected, i.e., short for HF, long for LF and smoothly varying in between. It also allows to change the windowing. It is kind of arcane, relatively speaking, but you (Bill) shouldn't have a problem with it. There is also a user guide on diyaudio with custom scripts to get you started (http://www.diyaudio.com/forums/full...ectrical-loudspeaker-correction-networks.html). I use the original ERB setting, which the software designer (Dennis) says matches the ear's response the best. REW has also recently introduced variable filtering. It also has psychoacoustic and ERB filters.

I used to think about EQ'ing the direct sound only, but DRC recommends measuring at the listening position. Dennis uses something called the spectral envelope to do the correction (DRC: Digital Room Correction). All I can say is that it sounds pretty fantastic. I have used rePhase to do manual correction and DRC is leagues ahead.
 
Last edited:
Apart from the audiophile qualities, the one thing this speaker does that no other has done before is that it relaxes me. The music just draws me in and makes me pay less attention to all the nonsense we like to worry about.

Lol! Exactly mimicking my original response to my arrays. I listened and thought: Should I really worry about woofers, tweeters and crossovers? Forget about it, just listen to the music!

I'm sure you'll be able to evolve them like I did. I'm jealous about your symmetrical setup though. What is the volume for each separate driver?

I do push the tiny drivers trough the impedance peak, which is still passively corrected with a conjugate network, even though I could not find definitive answers if that helped or not. Listening to it seemed to win me over but that may very well be a mind game. No answer from me there yet. But the cones still only move a small amount. Unless you play something like "A Perfect Circle - Lullaby". But even then they cope and bring unbelievable pressure from such tiny cones. My amp runs out of steam before the arrays do.

Try "Rodrigo y Gabriella - Hanuman" or other songs from the album 11-11 and play it loud. It's all midrange and very convincing. You will be there, listening to them play.

Kind of worried here (lol) not listening to flat on axis? Me neither ;).
 
A couple more questions if I may. What does the phase of the corrected signal look like?

Maybe you're up for showing a 4 cycle frequency dependant window for me to compare to? And I'd like to see a spectrum plot. I bet your setup has advantages to mine. That corner placement has to have advantages.

When listening to "Jenifer Warnes - Bird on a Wire, what imaging do you get, something like this:
bird_on_a_wire_wide.png


Or closer to this one:
bird_narrow1.png


From: B&O Tech: What is “Beam Width Control”? earfluff and eyecandy

Personally I have to listen again, at one point you hear Jenifer herself next to the central placed voice but moved a little back. Too bad it doesn't display on the graph. Unless it's the "bk" which makes sense. My cow bell and triangle (on left side) were more forward in the room. Have to listen again for the choir.

Had a lot of fun comparing maps...
 
Last edited:
Congratulations, ra7!
You have the benefit of having a dedicated AV room with symmetrical cornered loudspeakers - many acoustic problems avoided. DRC is doing the job well and you can easily adjust response to your taste. Your descending response is "standard" and gives fuul-bodied sound without chestiness/boominess etc. Obviously also room modes are minimal at your measuring point.

About construction - does each driver have it's own airspace?
I am dreaming of having same construcion in my HT room some day. I would have to tear down a bit of bookshelves though and there is no room for contents... (my vintage LPs and photo slides, remember those?)

Obviously you have read this classic https://www.trueaudio.com/array/
 
I wanted the drivers to be as close to the wall
Uncorrected Left FR and phase:

Uncorrected impulse response:
attachment.php


Corrected impulse response:
attachment.php

Very nice, but are you sure you are measuring the right speakers? ;)

I thought I read somewhere that one constant criticism of line arrays made up of multiple transducers was the inability to have a clean impulse response. It seems you have solved it rather nicely by picking larger drivers.

Best,


Erik
 

ra7

Member
Joined 2009
Paid Member
Lol! Exactly mimicking my original response to my arrays. I listened and thought: Should I really worry about woofers, tweeters and crossovers? Forget about it, just listen to the music!

Yes, I tried to pay attention to why I was enjoying it so much (that's the analytical type in me) and I think it's because the individual notes are so perfectly in balance and they just come one after the other right on track. At least that's the feeling I got.

I'm sure you'll be able to evolve them like I did. I'm jealous about your symmetrical setup though. What is the volume for each separate driver?
Yeah, I hope to enroll your help in improving the sound :)

All the drivers share the cabinet volume.

I do push the tiny drivers trough the impedance peak, which is still passively corrected with a conjugate network, even though I could not find definitive answers if that helped or not. Listening to it seemed to win me over but that may very well be a mind game. No answer from me there yet. But the cones still only move a small amount. Unless you play something like "A Perfect Circle - Lullaby". But even then they cope and bring unbelievable pressure from such tiny cones. My amp runs out of steam before the arrays do.

Try "Rodrigo y Gabriella - Hanuman" or other songs from the album 11-11 and play it loud. It's all midrange and very convincing. You will be there, listening to them play.

I'll have to get these songs to try them out.

Kind of worried here (lol) not listening to flat on axis? Me neither ;).

Dennis says in his DRC writeup that the correcting the spectral envelope using variable windowing produces the most neutral perceived response at the listening position. So, while it does not measure flat at the listening position (not that it should; only the direct sound emnating from the speakers should be flat), I'm perceiving it flat. I've played with RePhase and measurements a lot and this is the most neutral, most pleasing sound I've heard.
 

ra7

Member
Joined 2009
Paid Member
A couple more questions if I may. What does the phase of the corrected signal look like?

Here's the corrected phase:
attachment.php


attachment.php


Maybe you're up for showing a 4 cycle frequency dependant window for me to compare to? And I'd like to see a spectrum plot. I bet your setup has advantages to mine. That corner placement has to have advantages.

Will process and show it, but not right now.

When listening to "Jenifer Warnes - Bird on a Wire, what imaging do you get, something like this:
bird_on_a_wire_wide.png


Or closer to this one:
bird_narrow1.png


From: B&O Tech: What is “Beam Width Control”? earfluff and eyecandy

Personally I have to listen again, at one point you hear Jenifer herself next to the central placed voice but moved a little back. Too bad it doesn't display on the graph. Unless it's the "bk" which makes sense. My cow bell and triangle (on left side) were more forward in the room. Have to listen again for the choir.

Had a lot of fun comparing maps...

I'd say it's between the two charts you've shown. I'll have to listen again. Yes, I've heard the bit where Jennifer is next to her central voice. I'll say that the separation on orchestral works is better. I'll pick out a few recordings to share notes.
 

Attachments

  • Left_Corr_Phase.jpg
    Left_Corr_Phase.jpg
    73.9 KB · Views: 2,992
  • Right_Corr_Phase.jpg
    Right_Corr_Phase.jpg
    75.9 KB · Views: 1,855

ra7

Member
Joined 2009
Paid Member
Congratulations, ra7!
You have the benefit of having a dedicated AV room with symmetrical cornered loudspeakers - many acoustic problems avoided. DRC is doing the job well and you can easily adjust response to your taste. Your descending response is "standard" and gives fuul-bodied sound without chestiness/boominess etc. Obviously also room modes are minimal at your measuring point.

About construction - does each driver have it's own airspace?
I am dreaming of having same construcion in my HT room some day. I would have to tear down a bit of bookshelves though and there is no room for contents... (my vintage LPs and photo slides, remember those?)

Obviously you have read this classic https://www.trueaudio.com/array/

Thanks Juhazi! All the drivers share the internal volume.

Yes DRC is awesome, especially in the lower octaves where it is so hard to EQ. One has to rely on the ear because the measurements are usually corrupted by reflections.

Very nice, but are you sure you are measuring the right speakers? ;)

I thought I read somewhere that one constant criticism of line arrays made up of multiple transducers was the inability to have a clean impulse response. It seems you have solved it rather nicely by picking larger drivers.

Thanks Erik! I think I am measuring the right speakers but my measurement setup with the convolution engine in the measurement chain is not very reliable. I'm investigating it and I'll report back when I find out what's going on.

I don't think using larger drivers results in a cleaner impulse response. It is the floor-to-ceiling nature and also the fact that it's a single driver covering the entire passband that results in a clean impulse. If you have flat phase (i.e., all frequencies arriving simultaneously) and a flat magnitude response, you'll have the perfect impulse, i.e., a Dirac pulse.
 
Thanks Juhazi! All the drivers share the internal volume.

Yes DRC is awesome, especially in the lower octaves where it is so hard to EQ. One has to rely on the ear because the measurements are usually corrupted by reflections.



Thanks Erik! I think I am measuring the right speakers but my measurement setup with the convolution engine in the measurement chain is not very reliable. I'm investigating it and I'll report back when I find out what's going on.

I don't think using larger drivers results in a cleaner impulse response. It is the floor-to-ceiling nature and also the fact that it's a single driver covering the entire passband that results in a clean impulse. If you have flat phase (i.e., all frequencies arriving simultaneously) and a flat magnitude response, you'll have the perfect impulse, i.e., a Dirac pulse.

The issue, as I understand it, is that with small drivers, you end up with each radiating hemispherically and they can't create a single planar wave like you would with an electrostatic, or long ribbon. What should happen then is that if you measure the impulse response, you'll get the arrival of each driver as a separate incident. Meaning the driver directly in front of the mic arrives first, followed by the driver above and below, and so on until the signal from the drivers at the end arrives. The impulse response looks more like an ideal step response as a result.

However, if you can keep your drivers radiating in a planar, non hemispherical manner I would think you can avoid that issue. That's why I was so surprised to see your impulse response be so clean. :)

In any event, congrats. :) I'm learning new things just by lurking here.

Best,

Erik
 

ra7

Member
Joined 2009
Paid Member
Ah ah! So you're the guy who bought up all the VIfas at Parts Express, eh? :D
Interesting system. I will study it. :up:

Don't forget the shrink wrap, he bought all that too. :D

Erik

Yeah, I bought all of them. And lots of shrink wrap too. Planning to build a line array center, two sides and two rears. Charge $5 admission for the Pacific Northwest Stonehenge in my basement:devilr:
 

ra7

Member
Joined 2009
Paid Member
The issue, as I understand it, is that with small drivers, you end up with each radiating hemispherically and they can't create a single planar wave like you would with an electrostatic, or long ribbon. What should happen then is that if you measure the impulse response, you'll get the arrival of each driver as a separate incident. Meaning the driver directly in front of the mic arrives first, followed by the driver above and below, and so on until the signal from the drivers at the end arrives. The impulse response looks more like an ideal step response as a result.

However, if you can keep your drivers radiating in a planar, non hemispherical manner I would think you can avoid that issue. That's why I was so surprised to see your impulse response be so clean. :)

My understanding is different. A line array produces a cylindrical wavefront for frequencies proportional to its line length. Longer the array, lower in frequency extends the cylindrical nature of the wavefront. What that means is that when measured at any point along the vertical dimension on the cylinder, the sound is the same. Therefore the frequency response at a given height should not be thought of as a summation in time of multiple units along the length of the array. This is pretty much borne out in the uncorrected response. Floor-to-ceiling arrays are theoretically even longer because the floor and ceiling reflections extend the length of the array.

I'm not quite sure what happens when the wavelength is smaller than the driver size, i.e., the frequency above which the driver starts beaming. I need to re-read Speaker Dave's seminal paper on line arrays:
http://www.diyaudio.com/forums/mult...th-transducers-line-arrays-6.html#post2257395
 
Last edited:
Yeah, I bought all of them. And lots of shrink wrap too. Planning to build a line array center, two sides and two rears. Charge $5 admission for the Pacific Northwest Stonehenge in my basement:devilr:

Are you being serious here? :D
Well at least partly? With the right processing something like this would be mind-blowing!

I'd say it's between the two charts you've shown. I'll have to listen again. Yes, I've heard the bit where Jennifer is next to her central voice. I'll say that the separation on orchestral works is better. I'll pick out a few recordings to share notes.

I'd have to agree... need to play Jennifer again but do not have the house to myself this week. I haven't had the depth suggested in the lower chart, but have way more focus than the upper one suggests.
I agree to have heard more separation on other tracks. But I find these charts fun to compare to. I used a couple of tracks as reference all trough my journey and that Jennifer Warnes track did not make it into that selection.