For years I've been pursuing a way of getting better stereo sound like you get when you use a divider wall between stereo speakers to eliminate crosstalk and interference patterns. One way to do this is to add more channels using a matrix. A simple and well known matrix is to turn L and R signals into L-R, L+R, and R-L. What no one has ever told me is that this simple array will do amazing things if you space the speakers apart properly and listen at an appropriate distance. Center panned sounds only play through the center speaker, so there are no interference patterns so that solves the phantom center problem. But what happens when a sound is panned hard to the right or left? For a right panned signal you end up with the array playing a -R in the left channel, a R in the center and another R in the right channel. This of course creates interference patterns, but good ones! They work to create the proper stereo separation and never cause a null to cross our ears! This is because there is a null in the center of the sound field instead of a lobe. So each ear gets an appropriate stereo lobe! Each ear gets a lobe - so punny!
Here's a picture of my current array using basic little Sony SSCS-5 speakers
You can see I'm actually using 5 speakers here. The center channel plays full range down to 100Hz. The inside left and right play down to 600Hz. The outsides play 600Hz and down. This just helps to boost the stereo effect as the wavelengths start to get long compared to the width of the array. 3 speakers works great. 5 just gives it a little something extra and I happen to have lots of speakers and amp channels available.
Watch this video to see how the interference patterns produce excellent stereo throughout a huge range of frequencies. I used Falstadt's great ripple tank simulator to demonstrate how the effect works over a huge range of frequencies. http://www.falstad.com/ripple/Ripple.html. The three dots at the top of the screen represent an appropriately spaced array of 3 speakers playing a hard right panned signal. The circle down below represents the listener's head with strong signal reaching the right ear and weaker signal that is phase delayed reaching the left ear - as it should be. As I play a wide variety of frequencies you can see that strong stereo separation is maintained and no null from interference patterns ever crosses the right ear. You do get some weakening of the signal as the frequency goes down due to the center null widening so you can boost up the side channels in the lower frequencies or add a wider spaced set of speakers for the lower frequencies. I actually do both.
So what happens with slightly panned signals? The same thing except the lobes are not as different, so no matter the panning there are no unwanted nulls from the interference patterns at the listener's ears.
Also notice what's happening off to the sides - not a lot of energy to light up early side wall reflections.
So how does it sound? I'm elated! It's great! It's not phase-y or weird. The sound stage is natural and oh so clear and smooth with beautiful separation and clarity. It's a revelation. On recordings with wild stereo effects they get really wild like they should. I don't see myself ever going back to a two speaker setup for high quality stereo playback. I tried having the outside speakers playing regular stereo from 600Hz down and it's just a downgrade. This is MUCH better IMHO even in lower frequencies where it looks like it's falling apart and just turning into mono. It's blowing my mind.
As a general starting guideline I recommend spacing the speakers at about 1 foot apart center to center for the 3 center speakers and put the listening position about 8 feet back. You'll hear some good stuff with just 3 speakers but you may need to put a low frequency boost on the side speakers to correct the tone for side panned sounds. You can do this by playing correlated pink noise and panning it left to right. Adjust the lower frequencies on the side speakers until the pink noise sounds similar in tone and volume across left, center, and right of the sound field. Experiment with spacing and listening distance. It's surprisingly flexible.
One difficulty is how to setup the matrix array. I use my Mac mini running Audio Hijack to do it in the digital domain. This allows matrix mixing and access to all 8 channels of my Denon receiver over HDMI. You can do a basic 3 speaker array using a 2 channel playback system by creating a L+R signal for the center speaker and a L-R signal for the side speakers. I run the L+R signal into the right channel, and the L-R into the left channel of the amp. The left channel runs both side speakers in series or parallel. Just wire up the right speaker opposite of the left speaker and it works!
I don't recommend mixing line level signals in the analog domain unless you have appropriate signal mixing hardware that that properly buffers everything. YMMV but my results were lackluster just using Y-connectors. I also don't recommend hooking multiple amp channels up to the same speaker unless you really know what you are doing. If you are using low pass and high pass crossovers on the outboard speakers like I am, watch out for phase reversals. Suddenly your soundstage is backwards!
12/30/2022 Update. After trying a lot of things I'm back to just 3 speakers. With proper spacing and the voids filled in between the speakers to minimize edge diffraction this really produces the purest sound and best imaging. I think I was tricked by the increased overall volume of adding more speakers. So there's something to that. If you have a lot of the same kind of speakers lying around and they can't play very loud, 7 of them playing at once makes a noticeably higher playback level possible than just 2 or 3 of them, and you can get some really great imaging. Turning up the volume a little to make up for the fewer speakers and I'm actually hearing better sound overall. This is great news because it's easier to set up as there are no time delays required and so a simple 2 channel amp and 2 channel dac can be used to power all 3 speakers. You just have to get the stereo signal matrixed into L+R and L-R signals. I heard one person say that the electric summing and differencing causes irretrievable loss of information. It does if you just take the L+R signal by itself, or the L-R signal by itself. But if you mix all the signals, including the inverted L-R, which gives R-L, everything that should reach the ears does. And, much less of what shouldn't.
Here's a picture of my current array using basic little Sony SSCS-5 speakers
You can see I'm actually using 5 speakers here. The center channel plays full range down to 100Hz. The inside left and right play down to 600Hz. The outsides play 600Hz and down. This just helps to boost the stereo effect as the wavelengths start to get long compared to the width of the array. 3 speakers works great. 5 just gives it a little something extra and I happen to have lots of speakers and amp channels available.
Watch this video to see how the interference patterns produce excellent stereo throughout a huge range of frequencies. I used Falstadt's great ripple tank simulator to demonstrate how the effect works over a huge range of frequencies. http://www.falstad.com/ripple/Ripple.html. The three dots at the top of the screen represent an appropriately spaced array of 3 speakers playing a hard right panned signal. The circle down below represents the listener's head with strong signal reaching the right ear and weaker signal that is phase delayed reaching the left ear - as it should be. As I play a wide variety of frequencies you can see that strong stereo separation is maintained and no null from interference patterns ever crosses the right ear. You do get some weakening of the signal as the frequency goes down due to the center null widening so you can boost up the side channels in the lower frequencies or add a wider spaced set of speakers for the lower frequencies. I actually do both.
So what happens with slightly panned signals? The same thing except the lobes are not as different, so no matter the panning there are no unwanted nulls from the interference patterns at the listener's ears.
Also notice what's happening off to the sides - not a lot of energy to light up early side wall reflections.
So how does it sound? I'm elated! It's great! It's not phase-y or weird. The sound stage is natural and oh so clear and smooth with beautiful separation and clarity. It's a revelation. On recordings with wild stereo effects they get really wild like they should. I don't see myself ever going back to a two speaker setup for high quality stereo playback. I tried having the outside speakers playing regular stereo from 600Hz down and it's just a downgrade. This is MUCH better IMHO even in lower frequencies where it looks like it's falling apart and just turning into mono. It's blowing my mind.
As a general starting guideline I recommend spacing the speakers at about 1 foot apart center to center for the 3 center speakers and put the listening position about 8 feet back. You'll hear some good stuff with just 3 speakers but you may need to put a low frequency boost on the side speakers to correct the tone for side panned sounds. You can do this by playing correlated pink noise and panning it left to right. Adjust the lower frequencies on the side speakers until the pink noise sounds similar in tone and volume across left, center, and right of the sound field. Experiment with spacing and listening distance. It's surprisingly flexible.
One difficulty is how to setup the matrix array. I use my Mac mini running Audio Hijack to do it in the digital domain. This allows matrix mixing and access to all 8 channels of my Denon receiver over HDMI. You can do a basic 3 speaker array using a 2 channel playback system by creating a L+R signal for the center speaker and a L-R signal for the side speakers. I run the L+R signal into the right channel, and the L-R into the left channel of the amp. The left channel runs both side speakers in series or parallel. Just wire up the right speaker opposite of the left speaker and it works!
I don't recommend mixing line level signals in the analog domain unless you have appropriate signal mixing hardware that that properly buffers everything. YMMV but my results were lackluster just using Y-connectors. I also don't recommend hooking multiple amp channels up to the same speaker unless you really know what you are doing. If you are using low pass and high pass crossovers on the outboard speakers like I am, watch out for phase reversals. Suddenly your soundstage is backwards!
12/30/2022 Update. After trying a lot of things I'm back to just 3 speakers. With proper spacing and the voids filled in between the speakers to minimize edge diffraction this really produces the purest sound and best imaging. I think I was tricked by the increased overall volume of adding more speakers. So there's something to that. If you have a lot of the same kind of speakers lying around and they can't play very loud, 7 of them playing at once makes a noticeably higher playback level possible than just 2 or 3 of them, and you can get some really great imaging. Turning up the volume a little to make up for the fewer speakers and I'm actually hearing better sound overall. This is great news because it's easier to set up as there are no time delays required and so a simple 2 channel amp and 2 channel dac can be used to power all 3 speakers. You just have to get the stereo signal matrixed into L+R and L-R signals. I heard one person say that the electric summing and differencing causes irretrievable loss of information. It does if you just take the L+R signal by itself, or the L-R signal by itself. But if you mix all the signals, including the inverted L-R, which gives R-L, everything that should reach the ears does. And, much less of what shouldn't.
Last edited:
Being a diehard 2 channel analog guy, I'm usually turned off by attempts to create center channels for stereo music. Your experiments however have me curious being the attempts are in the analog domain.
Just in case you haven't thought about it already, a good quality digital mixer with eq, time and phase delay or a PC audio interface with sufficient mix outputs (aux sends) along with mixing software would be a better way to process your channels just to simulate the results in a more flexible fashion. I would be curious how this all would work with dedicated amplification and analog mixing.
Just in case you haven't thought about it already, a good quality digital mixer with eq, time and phase delay or a PC audio interface with sufficient mix outputs (aux sends) along with mixing software would be a better way to process your channels just to simulate the results in a more flexible fashion. I would be curious how this all would work with dedicated amplification and analog mixing.
The old matrixed multichannel systems & variations thereof can be quite effective. As well as the usual Hafler circuit with the rear / rears playing the stereo difference signal, uou can derive a centre channel from a stereo signal providing you have an amplifier that doesn't mind you shorting the negative terminals. Ditto for 4 channels (rear left + right) albeit separation is lousy, hence the active gain-riding circuits, & [subsequently] actively shifting coefficients of the higher priced matrix quadraphonic systems of the '70s (SQ, QS etc., with Tate Logic & Variomatrix being, at the last gasp of quad., the most effective).
Another variation that can be quite effective is a variation on width stereo. Two separate speakers for the left & right channels, preferably with their own amplifiers, the outer channels down a few dB & preferably BW limited above a desired point to reduce destructive interference. Can be surprisingly good, especially on material with strong stereo separation.
Another variation that can be quite effective is a variation on width stereo. Two separate speakers for the left & right channels, preferably with their own amplifiers, the outer channels down a few dB & preferably BW limited above a desired point to reduce destructive interference. Can be surprisingly good, especially on material with strong stereo separation.
Audio Hijack on the Mac mini is giving me a digital mixing and EQ/Crossover solution for now that allows me to access all 8 channels on my receiver over HDMI. This is very flexible because the speakers are hooked up to the receiver in the normal fashion so if I turn off Audio Hijack I can still send a standard Atmos or other multi-channel mix to the speakers, or standard stereo, which allows direct comparisons while listening to music or watching movies. I just have to move the speakers to more appropriate locations.
Doing comparisons I've found Atmos multi-channel music to be generally inferior to 2 channel music up-mixed with this matrix. For movies it's a close call but even then discreet multi-channel soundtracks are not adding much over this matrix. They're better though if more people are watching. But even with the 3 channel matrix the vocals stay in the center anywhere in the room so it's pretty darn good!
I too am interested in an all analog solution and it seems all it would take is a simple powered mixing box to do this with a very high level of fidelity. For the 5 speaker array the crossovers could be all done passively so the whole thing could still be run with a standard 2 channel analog amp and pre-amp, with the mixing box going between the amp and pre-amp. The array really acts as one big sound bar, so somebody could even build a speaker that looks like a huge sound bar which would have the left and right speaker terminals on the back as usual. Just hook it up like normal but be sure the matrix mixer is installed in the playback chain!
If you haven't seen it already, I do think this thread: https://www.diyaudio.com/community/...equations-for-3-speaker-stereo-matrix.222881/ might give you lots of inspiration.
I took a look at that long thread. They were basically talking about the same thing. The only different thing I'm bringing to the table is the tighter speaker spacing to make the panned interference patterns work out perfectly for stereo sound across the head at the sweet spot, which fortunately isn't so tight that you need to keep your head in a vice. You just have to be in the middle and not too close to the speaker array. I know people can be very particular about exactly how they want the sound stage to come across in terms of width and depth on some recordings so I'd be curious to find out what people think of the sound they get with the strong stereo separation at the ears created with my setup. There was a lot of talk also about recordings potentially using EQ on center panned vocalists and instruments to make up for comb filtering effects. I suspect they don't do that because then it will sound wonked if played back in mono or listened to off axis. I think the comb filtering is just accepted as a limitation with 2 speaker playback that generally goes unnoticed and still can sound really great when set up well. Recording the vocalist slightly off center as some have noted on that thread can also help minimize comb filtering. It used to bother me when I heard off center vocals because I feared something was off with my system. LOL! There's an album that's weighted pretty heavily to the right side - Colour Therapy by Bruce Davidson. I wonder if that was a mastering mistake or if it was intentional. I can't help but feel anxious about the state of my equipment when I listen to this album that is supposed to be relaxing. If you think of this as back ground music not intended to be listened to in the sweet spot it makes a lot of sense to center around one speaker for most of the sounds and allow ambience to stream off into the opposite speaker.
I can't understand why engineers would mix a main lead vocal off center. It makes no sense to me, having worked in the recording industry for this long. Back in the 60s when stereo was considered a new thing, people would hard pan vocals alot, mainly in pop music, which was seen as more of a novelty than anything else.
All forms of surround processing to date don't consider serious 2 channel playback in such a way that supports a wider defined stereo mix using the current technology available to the fullest possible extent. Everyone is now just trying to widen the stereo image into something more artificially vague rather than accurately defined. Also, much of the content now is steered by the taste (or lack of it) of newer engineers that don't understand accurate stereo recording and imaging.
So generally speaking, most movie soundtracks are mixed to be exaggerated and fake, catering to a younger crowd which doesn't know how real, accurate sound effects and live music actually are supposed to sound. This conditions people over time, skewing their sense of reality and perception. This corruption of the senses makes things worse for us serious audiophiles, sort of like Napster, MP3s and Ipod earbuds did back when they took over as a new method of music consumption.
Back when Apocalypse Now was done in 4 track surround, Walther Murch made a big attempt to be accurate about the type and placement of sounds, providing a believable soundtrack and supporting the movie's plot. Nowadays it feels like the added flexibility of more channels in mixing movie sound enables a free for all situation, where people feel like they can be more sloppy about image localiation based on just the audio itself. What that spells for me as an audiophile is cringing at the way many live concerts are mixed, usually killing the audio experience right off the bat with fake, overly hyped surround sound and bloated overhanging bass. Add to that overly compressed audio, ruining the experience completely. There are some exceptions to this, but they're rare.
If someone actually tried to mix a concert accurately and tie it to the inagry in such a way that sounds like I'm in the audience, I'd be all over it. For 2 channel stereo it hasn't been done so far to my knowledge and I doubt anyone is seriously interested in catering to more audio savvy consumers. I'm sure the current surround formats are capable of delivering an accurate 3.1 mix with a dedicated center channel, but the current crop of younger engineers don't care about it. Someone please correct me if I'm wrong about this.
All forms of surround processing to date don't consider serious 2 channel playback in such a way that supports a wider defined stereo mix using the current technology available to the fullest possible extent. Everyone is now just trying to widen the stereo image into something more artificially vague rather than accurately defined. Also, much of the content now is steered by the taste (or lack of it) of newer engineers that don't understand accurate stereo recording and imaging.
So generally speaking, most movie soundtracks are mixed to be exaggerated and fake, catering to a younger crowd which doesn't know how real, accurate sound effects and live music actually are supposed to sound. This conditions people over time, skewing their sense of reality and perception. This corruption of the senses makes things worse for us serious audiophiles, sort of like Napster, MP3s and Ipod earbuds did back when they took over as a new method of music consumption.
Back when Apocalypse Now was done in 4 track surround, Walther Murch made a big attempt to be accurate about the type and placement of sounds, providing a believable soundtrack and supporting the movie's plot. Nowadays it feels like the added flexibility of more channels in mixing movie sound enables a free for all situation, where people feel like they can be more sloppy about image localiation based on just the audio itself. What that spells for me as an audiophile is cringing at the way many live concerts are mixed, usually killing the audio experience right off the bat with fake, overly hyped surround sound and bloated overhanging bass. Add to that overly compressed audio, ruining the experience completely. There are some exceptions to this, but they're rare.
If someone actually tried to mix a concert accurately and tie it to the inagry in such a way that sounds like I'm in the audience, I'd be all over it. For 2 channel stereo it hasn't been done so far to my knowledge and I doubt anyone is seriously interested in catering to more audio savvy consumers. I'm sure the current surround formats are capable of delivering an accurate 3.1 mix with a dedicated center channel, but the current crop of younger engineers don't care about it. Someone please correct me if I'm wrong about this.
The off center panning is interesting. As for what I want the sound stage to be like I tend to say accurate and realistic, but being at a live music event recently and closing my eyes I noticed some very strange effects. The choir was singing and I was sitting off to the right. With my eyes open everything made sense and I could locate individual singer's voices. With my eyes shut the whole choir seemed to move further to my left and some people's voices seemed to move around as they sang different notes, this all being the acoustic effects of human voices singing in a room creating localization problems for my brain. My eyes could assist and straighten everything out. So when listening to the stereo with no visual aids I'm not sure what to expect. On some classical and opera recordings I think the stereo mics are a ways back form the singer and you can hear them moving around as they sing, and sometimes they even seem to suddenly shift position as they hit a different note. I always assumed this was a shortcoming of my system but hearing it in real life suggests it's just something that can happen unless you close mic. the singer in mono and pin their position down in the stereo mix. Other things I've heard too like stridency in voices on certain notes that I was trying to fix in my speaker designs occur in live, unamplified singing due to interference patterns caused by reflections off the walls. So, in some ways I actually think I like a groomed recording that removes some of those effects so they don't pile on again when my listening room adds more reflections of it's own. Sound recording and mixing is definitely an art and I prefer whatever happens to sound enjoyable, which usually means there has to be something realistic and natural about the way it comes across to me.
What I've heard on most of these multi channel mixes is that they are not fully using the center channel, but maintaining a phantom center with the two side channels, sometimes adding a little center channel in to stabilize it. This seems like a wasted opportunity to me, but there are a lot of complaints about using the center channel fully for music. People say it mucks up the sound stage, with the center panned vocal or instrument seeming to move forward too much, loose depth. I don't hear it that way on my set up but others might. Perhaps it's how a lot of people have learned to interpret the phantom center anomalies and have come to enjoy it.
What I've heard on most of these multi channel mixes is that they are not fully using the center channel, but maintaining a phantom center with the two side channels, sometimes adding a little center channel in to stabilize it. This seems like a wasted opportunity to me, but there are a lot of complaints about using the center channel fully for music. People say it mucks up the sound stage, with the center panned vocal or instrument seeming to move forward too much, loose depth. I don't hear it that way on my set up but others might. Perhaps it's how a lot of people have learned to interpret the phantom center anomalies and have come to enjoy it.
The notes and sounds from a singer doesn't just come from the mouth, but also from various vibrating cavities of the upper body, which will cause the effects you talk about if in a group setting with multiple singers. The plosives are usually the main localiation cues, as they mainly come from the singers mouth. The vowels being the main components of melody modulate with the mouth, nose and sinus cavities along with the fundamental notes emanating from the chest area, throat and face.
Our hearing processes the melody and plosives in different parts of the brain, so there are multiple things going on which distract from the image our brain tries to recreate while hearing all this at once. The resonances are not easily localized due to the wavelengths making them up being too long for our hearing to process. They also tend to travel at different speeds through other people and surrounding things compared to the plosives, reaching our ears at a different time (speed of sound is higher in solids and liquids than air). This effect smears the image our brain tries to reassemble.
If you listen to pipe organ recordings, these behave acousticallty the same way as the choir you noted.
Our hearing processes the melody and plosives in different parts of the brain, so there are multiple things going on which distract from the image our brain tries to recreate while hearing all this at once. The resonances are not easily localized due to the wavelengths making them up being too long for our hearing to process. They also tend to travel at different speeds through other people and surrounding things compared to the plosives, reaching our ears at a different time (speed of sound is higher in solids and liquids than air). This effect smears the image our brain tries to reassemble.
If you listen to pipe organ recordings, these behave acousticallty the same way as the choir you noted.
Last edited:
During the Christmas break I made some changes and improvements to this speaker array. I got my room rearranged so the speakers are out in front of and below the TV now, at closer to ear level. I've also switched from using electronic crossovers on some of the speakers to just playing them all full range down to 120Hz where they cross over to the subs. I've added two more speakers which just play the center L+R signal, one pointing backwards and upwards and delayed 15ms, and one pointing forward on the floor and delayed 20ms. This was necessary because without them the center speaker was playing center panned sounds all by itself so center panned singers were a little weak compared to side panned sounds, which always had all 5 speakers playing at once. The extra center signals add presence and air to both the center and side panned sounds and provides some good sense of depth. Another big improvement was to match-pair the speakers for the inner sides and outer sides. The Sony SSCS-5s have audible tonal discrepancies between them, some sounding a bit brighter than others on pink noise. Matching them side to side more closely makes a very noticeable difference in the stereo effect, and using EQ to correct the general tone of the speakers also adds to the spaciousness and realism. At this point it's really doing some incredible imaging work, with sounds easily going out to 180 degrees, and on a Tomita album which seems to use some HRTF effects I even perceived a sound panning all the way around behind me! It's startling how clean and solid these ultra wide side panned sound can be while the front sound stage is also so solid and clean.
I did another experiment that demonstrated to me that there is a way to do this whole thing without the need for any matrix mixer! A pure analog system with no extra components - just left and right signals played by an array of drivers in a properly built speaker setup with a 2 channel amp no signal mixing or processing of any kind required to do the trick of alleviate the phantom center image comb filtering while providing an excellent stereophonic effect and avoiding the weakness in the center panned sounds without the need for extra delayed ambience channels! Delayed ambience channels may still be desirable. I didn't keep it setup because it really isn't very practical to use Sony Bookshelf speakers for this method but they proved to me that it works. This method should be a specially built speaker system. It has a drawback in that you need to be positioned correctly for height and centered in the middle of the array. You should be at the correct height for most speakers anyway but this makes it fairly critical. I'm going to build a prototype as soon as possible.
I did another experiment that demonstrated to me that there is a way to do this whole thing without the need for any matrix mixer! A pure analog system with no extra components - just left and right signals played by an array of drivers in a properly built speaker setup with a 2 channel amp no signal mixing or processing of any kind required to do the trick of alleviate the phantom center image comb filtering while providing an excellent stereophonic effect and avoiding the weakness in the center panned sounds without the need for extra delayed ambience channels! Delayed ambience channels may still be desirable. I didn't keep it setup because it really isn't very practical to use Sony Bookshelf speakers for this method but they proved to me that it works. This method should be a specially built speaker system. It has a drawback in that you need to be positioned correctly for height and centered in the middle of the array. You should be at the correct height for most speakers anyway but this makes it fairly critical. I'm going to build a prototype as soon as possible.
I've been going to Comdex and CES for 20+ years. I've heard stereos at CES that cost a million dollars, once you factor in the amplification and speakers and cables, etc.
Yet the best imaging I've ever heard at CES was with $200 worth of speakers!
In a nutshell, Sound Blaster (the audio card company) had set up a video game demo. They had a GIANT booth, this was decades ago, when PC upgrades were a big business and Sound Blaster had money to burn.
They'd erected a ring that was suspended over their booth. I'd estimate that the booth was about 500-800 square feet.
Because the speakers were arranged in a ring, if you stood in the middle of the ring, the soundstage was just uncanny.
If you moved ten feet out of the center of the ring, the entire soundstage collapsed and it was nothing special. But when all of the pathlengths aligned, it was absolutely crazy.
The speakers were incredibly pedestrian, you can find them on Craigslist all day long for $40:
I've measured them myself, and they're astonishingly flat from 200Hz to 20khz. Just a single full range paper driver in a tiny sealed enclosure.
Those little speakers could probably do the trick for my little array just fine. I might just search those out to use for a prototype. I think something like that could be great to take outside, free from room reflections. That setup at the show sounds amazing. Having a lot of space to work with and getting setup like they did it doesn't surprise me that they could get something amazing to happen. I should say it doesn't surprise me now in the light of my recent experiences. What is surprising to me how much money is thrown in to simple 2 speaker 60 degree triangle arrangements that are so fundamentally limited in terms of imaging. People rage about how good some of those systems image and I'll agree that some of them are surprisingly good but there's a definite upper limit to what can be done with that.I've been going to Comdex and CES for 20+ years. I've heard stereos at CES that cost a million dollars, once you factor in the amplification and speakers and cables, etc.
Yet the best imaging I've ever heard at CES was with $200 worth of speakers!
In a nutshell, Sound Blaster (the audio card company) had set up a video game demo. They had a GIANT booth, this was decades ago, when PC upgrades were a big business and Sound Blaster had money to burn.
They'd erected a ring that was suspended over their booth. I'd estimate that the booth was about 500-800 square feet.
Because the speakers were arranged in a ring, if you stood in the middle of the ring, the soundstage was just uncanny.
If you moved ten feet out of the center of the ring, the entire soundstage collapsed and it was nothing special. But when all of the pathlengths aligned, it was absolutely crazy.
The speakers were incredibly pedestrian, you can find them on Craigslist all day long for $40:
I've measured them myself, and they're astonishingly flat from 200Hz to 20khz. Just a single full range paper driver in a tiny sealed enclosure.
The purpose for my experimenting was originally to derive a center channel to get rid of the phantom center. L+R is a simple start for the center. I knew that if you place a speaker playing -R on the left and -L on the right side at about ear distance from the center speaker it will make stereo again because it will cancel the appropriate signals at each ear. That's basically Polk SDA but with a combined center. But then it occurred to me that a L-R and R-L signal to the side speakers would enhance the correct channel in each ear and cancel the wrong channel as well, and it would remove any center panned sounds from the side speakers. I tried it and it worked really well - much better than just the cancelation. But then I realized if I added another set of L-R and R-L outboard of the first two at the same spacing it should catch the next wave of crosstalk as well. So I tried it and it worked even better! It might be better still with another pair outboard on each side for a total of 7 speakers. I don't have enough matching speakers to try that. In any case, not only do I have a center channel but also outstanding stereo imaging with the 5 speaker array. The added 2 center channels with delay were added to help the center speaker. By delaying and attenuating them and placing them appropriately they don't interfere with the imaging or color the sound. They're not always needed but I generally think they add something nice. And, it uses up all my speakers from my 7.1 setup and uses all The amp channels on my receiver.
The quality of the stereo imaging really took me by surprise. There's still crosstalk even with the 5 speaker array, and the spacing doesn't seem to be very critical which totally confused me at first. Some wave simulation revealed to me that as you get back from the array a little ways you can get very good beaming of each channel at the correct ear so it really does reduce crosstalk considerably. The further back you get from the array the wider the spacing should be. I think what helps it even more is that the crosstalk that does reach the opposite ears always has both channels in it at once, and they are both out of phase across the head, which means they have no meaningful directional information in them so the crosstalk doesn't have contradictory information to the direct signal to each ear in terms of imaging. In a regular 2 speaker setup the crosstalk that reaches the ears is a mirror image of what first reached the ears so it tries to flip the sound field backwards. The brain chooses the first one because it was first and a little louder but the crosstalk's own very coherent and contradictory imaging information still works negatively against the perceptual experience.
I think this array does add a lot of room ambience to side panned information, but actually reduces it for center panned sound because side panned signals play through all 5 speakers but center panned signals are canceled on the sides and only play in the center channel. So the extra time delayed centers help with that a bit.
Last edited:
Here's my setup now that I've moved into my new listening room. You can see the three Sony SS-CS5 speakers in the middle, tucked into an old entertainment center cabinet that I bought for my mother decades ago, when CRT TVs were still mostly what there was. There are black towels stuffed between the speakers to eliminate the hollow sound of the cabinet. The center channel speaker below is also just filler for the same purpose. It's not actually hooked up. It needed somewhere to go anyway. Off to the sides are a pair of corner horn speaker systems that are currently serving duty as subwoofers crossed over at 100Hz. They are just being powered by the Denon receiver's surround channels and have to be turned down 15dB with EQ at the crossover to match the Sonys. By 20Hz they're only turned down 3dB. Each stack has four 18" woofers each loaded into its own 6 foot long folded horn. Overkill? I don't know but now that I've got it level matched it sounds great! Way, way better than the pair of 10" subwoofers I was using before. These bass cabinets can play cleanly up to 200Hz but interestingly the imaging from the 3 channel array is diminished unless the crossover is turned down to 100Hz. I'm not sure why that is. They're 24 dB per octave crossovers so even at 200Hz I can't understand a word being said if I just listen to the woofers.
Last edited:
I've built a unified open baffle version of this system using Dayton waveguides and Faital Pro 10 FE200 10" mid-woofers. To start the experiment I just bought a 2' x 4' piece of plywood and cut some holes in it. It's just sitting on those speaker stands, leaning against the TV cabinet. I can't believe how great this sounds. I expected to cross the 10s to the horns at around 1500Hz but experimenting and listening has pushed that up to 3500Hz! The imaging and tone is just amazing. I'm becoming more and more convinced that keeping the crossover out of the range between about 300 and 3000Hz is a really good idea. The last time I employed a midrange driver in this way I used a 4" and I thought that was pushing things at 3500Hz. But it sounded great and so does this. Shouldn't it be beaming up top? Mabye it is, but I can't get those horns to sound any better covering that range. They sound more in your face if I cross them at 2000. Still good, but in a different way. These 10s push the soundstage back and sound sweeter.
I've got them crossed at 300 and 3500 24dB/octave on both ends.
The system works better than ever, producing a beatiful wall to wall soundstage with excellent tone and a rock solid center image. As excited as I am by this unusual arrangement's success, I think I'm just as excited about these awesome 10" drivers from Faital Pro! They do need some EQ to get them really flat, but they go nicely up to about 4K and then drop off fairly gracefully. No really nasty peaks up higher. They're efficient and can handle some power. And they're not expensive.
I've got them crossed at 300 and 3500 24dB/octave on both ends.
The system works better than ever, producing a beatiful wall to wall soundstage with excellent tone and a rock solid center image. As excited as I am by this unusual arrangement's success, I think I'm just as excited about these awesome 10" drivers from Faital Pro! They do need some EQ to get them really flat, but they go nicely up to about 4K and then drop off fairly gracefully. No really nasty peaks up higher. They're efficient and can handle some power. And they're not expensive.