Late Ceiling Splash

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hmmm not sure. But it is in the text. Maybe the size needs updating on the web page?

Yes maybe. I remember the first upfiring module (circa 2013) used four woofers and a waveguide/CD. In 2015 that design was revised to a single coaxial 10” driver.

Either way, the module as depicted in the image at the website looks more like the original iteration. Below is a picture of the revised module from 2015:
image.php
 
Some of you are right on the money.

Maybe I can fill in a few blanks and/or add to the confusion...

Having been a SoundLab (large fullrange electrostat) owner and dealer for many years, I felt that much of what they do well comes from their spectrally-correct backwave and its relatively late arrival, given sufficient distance between the speakers and the wall behind them. So ten years ago I launched my first bipolar loudspeaker, the "Dream Makers", deliberately copying as much as I could about the SoundLabs' behavior (90 degree pattern in the horizontal plane over most of the spectrum, both front and back). I was quite surprised when Robert Greene of The Absolute Sound gave the Dream Makers a Golden Ear award.

When vital parts became unavailable I had to discontinue the original Dream Makers. But I kept reading about reflections, because the more I could learn about what makes the difference between "good" and "bad" reflections, the better my chances of making good speakers. Floyd Toole investigated the arrival angle of reflections, and found that about 60 degrees off the centerline (30 degrees forward of "directly to the side") was ideal. So my next design was a "twisted bipole" speaker, which combined strong toe-in of the front-firing array with a side-firing (instead of rear-firing) array.

Jim Romeyn had heard the Dream Makers and apparently thought they worked pretty well, and so he set up a bipolar system using speakers that he already had on hand (multiple 6" two-way mini-monitors). He was using a trinaural system at the time, so he had a center channel. Each of the three channels consisted of a front-firing mini-monitor sitting on top of a rear-firing mini-monitor, which in turn was atop a stand.

I told Jim of my latest thinking, as embodied in my "twisted bipole" configuration, so he tried that with his left and right speakers (re-orienting the rear-firing arrays to become side-firing arrays), and found that it was an improvement. He left the center channel as a normal bipole at first, then one day for whatever reason (Jim is extremely creative) he put the rear-firing speaker of the center-channel stack on the floor and aim it up at the ceiling. He was very happy with the result and reconfigured all three channels to have an up-firing array. I suggested he make a cardboard "shield" to block the radiation of the up-firing unit, so that its highs and most of its mids wouldn't "leak" around the smallish speaker stand and arrive too early. This was a further improvement. I was quite impressed with the results Jim was getting (he lived about a 45 minutes from me). We decided to work together.

Jim and I showed our first collaboration at the 2013 Rocky Mountain Audio Fest. They caught the attention of Tyson and Pez from AudioCircle.

A bit of background: Tyson and Pez may well have been the most thorough and organized reviewing team to ever hit an audio show. They planned in advance which rooms they would visit and had five tracks picked out that they would play. They would take turns sitting in the sweet spot. Then they would type up their impressions and move on to the next room on their list to stay on schedule. They covered more rooms more thoroughly, and with a more apples-to-apples comparison, than anyone else ever has, to the best of my knowledge. At least one major magazine - one you have all heard of for decades - tried to hire them, and they declined.

Anyway here are their comments on our system from the 2013 RMAF:

RMAF 2013 Day 3 - Tyson and Pez Coverage LIVE (scroll about 2/5 of the way down the page)

Pez and Tyson's Best of RMAF- the ultimate guide to awesomeness 2013 (fairly far down the first post on this page)

Since then we've made a few changes to the Late Ceiling Splash section, which we call the "Space Generators", because with proper setup they can make the room sound bigger than it really is. Here's how that works:

The ear/brain system judges the size of a room by the time delay between the first-arrival sound and the "center of gravity" of the reflections. By introducing some relatively late-onset reflections we can shift that "center of gravity" to a later time, which in turn tricks the ear into thinking the room is bigger than it really is. So we get less "small room signature". It is highly counter-intuitive to think that ADDING reflections results in hearing LESS of the room, but we think that's what's happening.

One thing we've learned is, you don't want these additional reflections to be too loud, or else they start to degrade the system's clarity. We had the Space Generators turned up too loud at that 2013 show. At RMAF 2018, the Space Generators were about minus 13 dB relative to the main speakers, measured at the listening position. So they were only adding about .2 dB to the in-room SPL. Tyson and Pez got burned out and no longer do their marathon show reports, but they did stop by our room, and you can find Tyson's comments here:

Tyson's Best of RMAF 2018 (scroll down to the bottom of the first post)

The lighting was poor in our room so the photos didn't turn out well. We had large stand-mount speakers sitting atop some rather unusual stands, which incorporate a Space Generator section and a subwoofer. We call them the "SuperStands", because there simply is not enough hyperbole in high-end home audio these days.

So I think there are a couple of things that differentiate the Late Ceiling Splash from previous systems that use up-firing drivers. I'm aware of the big LINN DMS speakers, and Ohm system from back in the day, a Lowther USA system that had a second, up-firing driver, and of course the late great Richard Shahinian's polydirectional speakers.

When the additional drivers are on top of the cabinet, you get a fair amount of relatively early-onset energy from the side of the up-firing drivers. This can cause coloration. Our configuration places a fairly wide speaker box in between you and the upfiring drivers. Also if the upfiring drivers are up higher, the time delay for the bounce off the ceiling may not be long enough for that reflections to be as beneficial as it can be (we shoot for 10 milliseconds, and I got that figure from Earl Geddes).

I have used the technique of adding a rear-firing driver to fix the power response for many years, imo that's not quite the same thing. In fact I do that in both of the systems you've seen at the links above, to compensate for the narrowed vertical coverage of the SEOS waveguides that I use.

And as some of you have noted, there is a conceptual similarity between what we're doing and a dipole speaker pulled out from the wall a bit. We haven't fully taken advantage of the ability to back out system right up against the wall under show conditions because it does sound better out in the room a bit, but it still works well quite close to the wall.

There are some aspects of the Space Generators that are "trade secrets", the frequency response curve that we shoot for being one of them. Actually that has evolved a bit too over the past few years.

Anyway I apologize for how long-winded this turned out to be.
 
Last edited:
  • Like
Reactions: 1 user
According to Toole & Olive, the work in Harmon & the spino-rama measurements, it appears that a loudspeaker's early-reflections response is a major contributor to listeners' experience. Apart from expanding the perception of room size, I'm wondering if the late ceiling splash also has some role in perceptually overriding (improving) the main speaker's own off-axis response by perceptually substituting a speaker with its direct on-axis spectrum (with a tweaked FR curve) being used as the primary reflective power?
 
According to Toole & Olive, the work in Harmon & the spino-rama measurements, it appears that a loudspeaker's early-reflections response is a major contributor to listeners' experience.

My understanding is that the first sidewall reflections expand the soundstage size and increase the sense of immersion. They also can degrade imaging precision and clarity. Apparently listeners generally find the benefits of a good wide-pattern speaker (which will generate a lot of reverberant energy, much of it in early reflections) to outweigh any downsides. Jim and I think there is room for improvement in delaying (rather than eliminating altogether) the arrival of that reverberant energy.

Back in the mid-80's Ken Kantor designed an interesting loudspeaker for Acoustic Research. It was called the MGC-1, and its nickname was the "Magic". It used a highly directional main array + a delayed side-firing array. It was not a big success in the marketplace, but imo it got some important things right and just needed a little fine-tuning. Perhaps in a sense what we are doing is a passive variation on that theme.

Apart from expanding the perception of room size, I'm wondering if the late ceiling splash also has some role in perceptually overriding (improving) the main speaker's own off-axis response by perceptually substituting a speaker with its direct on-axis spectrum (with a tweaked FR curve) being used as the primary reflective power?

Imo there is definitely improvement in "fixing" the spectral balance of the reverberant field. My instinct is that the spectral balance of the reverberant field would become a "weighted average" of the off-axis response of the main speakers + the power response of the LCS section. But thus far my first-hand experience with using an LCS section with "conventional" main speakers is nil - I've only heard them with my own main speaker designs, which tend to have fairly well-behaved off-axis response. With my speakers, the LCS section changes the "feel", but doesn't change the tonal balance. I'm not sure an LCS section could reliably "fix" the reverberant field of "conventional" speakers because I think the LCS section would have to be turned up fairly loud to do so, and our experience has been that if the LCS section is too loud, clarity is degraded. That being said, it might still be a "net improvement" to turn the LCS section up louder than we normally recommend.
 
Last edited:
I find it very important to have reflections from behind, this opens up the soundfield a lot. It is a little like holding your cuped hand behind your ear to hear better. It is not adviceble however to put your listening chair close to the rear wall because of the low freq boost, it is better to sitt away from the wall and then use an acoustic reflector standing behind you. This if you have an absorbent back wall or an open space behind you. I think it all has to do with how sound levels decreases with distance in small rooms vs a big room like a concert hall, in a big room the refections has not been attenuated that much since you sitt more in the far field. Reflections from front wall are are also crucial, i m not so sure about side walls reflections
 
My understanding is that the first sidewall reflections expand the soundstage size and increase the sense of immersion. They also can degrade imaging precision and clarity. Apparently listeners generally find the benefits of a good wide-pattern speaker (which will generate a lot of reverberant energy, much of it in early reflections) to outweigh any downsides. Jim and I think there is room for improvement in delaying (rather than eliminating altogether) the arrival of that reverberant energy.

Back in the mid-80's Ken Kantor designed an interesting loudspeaker for Acoustic Research. It was called the MGC-1, and its nickname was the "Magic". It used a highly directional main array + a delayed side-firing array. It was not a big success in the marketplace, but imo it got some important things right and just needed a little fine-tuning. Perhaps in a sense what we are doing is a passive variation on that theme.



Imo there is definitely improvement in "fixing" the spectral balance of the reverberant field. My instinct is that the spectral balance of the reverberant field would become a "weighted average" of the off-axis response of the main speakers + the power response of the LCS section. But thus far my first-hand experience with using an LCS section with "conventional" main speakers is nil - I've only heard them with my own main speaker designs, which tend to have fairly well-behaved off-axis response. With my speakers, the LCS section changes the "feel", but doesn't change the tonal balance. I'm not sure an LCS section could reliably "fix" the reverberant field of "conventional" speakers because I think the LCS section would have to be turned up fairly loud to do so, and our experience has been that if the LCS section is too loud, clarity is degraded. That being said, it might still be a "net improvement" to turn the LCS section up louder than we normally recommend.

Would it work even better if the LCS was firing into a ceiling mounted diffuser?
 
I find it very important to have reflections from behind, this opens up the soundfield a lot.

Ime it depends on the room - if you have a fair amount of space behind you then diffusion works well on the back wall. If you are very close to it, then you may have to resort to absorption. If so, I suggest using the minimum amount of absorption necessary. Put a mirror on the wall to find out exactly where the first reflection from each speaker happens, and base your minimalist absorption on that.

Would it work even better if the LCS was firing into a ceiling mounted diffuser?

Yes! My understanding is that we want to avoid "specular" (strong, distinct) reflections while preserving the spectral content of the reflections. Specular reflections are more likely to shift the image. Decorrelation in the reverberant field is desirable - see the last Griesinger quote below.

David Griesinger has researched into reflections & immersion & intelligibility
Your ideas on speaker design seem to concur with his research.

I am more a disciple of Griesinger than of Geddes or Toole when it comes to the reverberant field. Griesinger focuses on large rooms - concert halls, auditoriums, lecture halls - but I think his general principles are still applicable to small rooms (home listening rooms).

The main differences of course are the much longer reflection paths (and therefore longer time delays) in the larger rooms Griesinger works with, and the shorter reverberation times in our small rooms. I've written to him a couple of times hoping to find out more about applying his ideas to smaller rooms, but he hasn't replied.

Anyway here are some imo relevant Griesinger quotes, lifted from various papers that he has on his website. I didn't record which papers because I was just collecting information for my own benefit:

Envelopment is perceived when the ear and brain can detect TWO separate streams:
A foreground stream of direct sound.
And a background stream of reverberation.
Both streams must be present if sound is perceived as enveloping.

In a small room there is almost never sufficient late reflected energy to contribute to the background perception.
[This is why I believe in adding a bit more late-onset reverberant energy, the late ceiling splash being one way of doing so passively.]

Presence depends in the ability of the ear and brain to detect the direct sound as separate from the reflections.

When presence is lacking the earliest reflections are the most responsible.

The earlier a reflection arrives the more it contributes to masking the direct sound.

The issue of decorrelation is very important, and perhaps at present unresolved. It has been my
[Griesinger's] experience in everything I do that decorrelated reverberation sounds more natural than correlated reverberation.

This is my distillation of Griesinger's ideas:

There must be THREE things present for envelopment: A foreground stream of direct sound; a background stream of reverberant sound; and a distinct time gap in between the two. The latter is implied by Griesinger, but I like to make it explicit to emphasize avoiding early reflections.

So what I'm shooting for is "presence" (it sounds like the musicians are present) along with "envelopment" (it sounds like you're in the acoustic space they were in, or at least like you're in a bigger acoustic space than your small room).
 
Last edited:
One apparent effect of the additional late-onset reverberant energy for which I have no satisfactory explanation is the way it seems to let you hear more of the acoustic space that's on the recording. It seems to facilitate the ear/brain system getting a good fix on the reverberation in the recording, as opposed to the reverberation signature of your actual listening room.

At Axpona 2016, Andrew Quint of The Absolute Sound came into the room. I gave my schpiel and he listened politely, then pulled out a thumb drive and said, "I'd like to challenge that." So we played the recording on his thumb drive and I asked him how we did. He said, "It works, it's not a gimmick. I could hear the acoustic signature of the hall the recording was made in."

Quoting from an online post Andrew subsequently made:

"I'll elaborate a bit on the recording in question and what I heard in my brief audition of Duke's new speaker.

"The recording was a FLAC rip of the CD layer of an RCO Live SACD: Shostakovich—Symphony No. 15; Concertgebouw Orchestra/Bernard Haitink conductor. It's a live recording from March of 2010 (Haitink made a much earlier recording of the same piece with the London Philharmonic; he was the first person to record all the Shostakovich symphonies.) For a couple of years, this has been my go-to symphonic recording when I have just a short time to get a sense of an unfamiliar system. It's an excellent performance, something I can listen to repeatedly without going nuts, which is important at a show. In terms of audiophilia, it's an extremely detailed yet atmospheric representation of an orchestra, with excellent dynamics and fully characterized instrumental colors (bells, solo turns by violin, flute, piccolo, string bass, trumpet, etc.) And—with the right audio gear—it successfully renders the essence of (IMO) one of the greatest 3 or 4 concert halls on earth, the Concertgebouw (thus the orchestra's name) in Amsterdam. I've heard music there, and there's truly a sense of sound being present in the air around you.

"The multichannel program on the RCO Live SACDs (there are dozens) get this last aspect right; so did the Bienville Suite, nearly to the same degree, despite the presence of only two channels. My concern when Duke told me about the rear-firing drivers was that this would impart some generic, Bose-like spaciousness to the recording, but that wasn't the case—what I heard was the unique acoustic signature of the Concertgebouw.

"So, a splendidly realized design—a pleasure to listen to."

I wish I knew EXACTLY why this effect was happening, so I could then perhaps optimize further.
 
Last edited:
Yes, Griesinger is the man when it comes to concert hall acoustics - he even points out in one of his presentations where he finds Toole is incorrect.

The decorrelated reverberation is an interesting phenomena
Gary S. Kendall has published research & book (1995) on the 5 perceptual effects of decorrelated reverberation:
- the timbral coloration and combing is avoided because the constructive & destructive interference of multiple delayed signals is perceptually eliminated
- the diffuse sound fields akin to the late field of reverberant concert halls
- the externalization in headphone reproduction
- the lack of image shift because :
- the precedence effect is defeated

Perceptual research has progressed a lot since 1995 particularly the area of Auditory Scene Analysis (Bregman 1990) which is concerned with how we analyze/categorize all the vibrations that belong to separate auditory objects & how we are able to similarly follow separate auditory streams.

The perceived naturalness of decorrelated reverberation has to do with how sounds in nature are reflected off many surfaces before reaching our ears - essentially decorrelating the reflections. This, as you say allows us to perceive the sound as foreground & background auditory streams allowing us to separately evaluate the sound making object & the environment in which the sound was made. Recent research (2016) suggests that we use statistical summaries of such background streams & evaluate them as natural if they conform to the statistical regularities found in nature
We conducted a large-scale statistical analysis of real-world acoustics, revealing strong regularities of reverberation in natural scenes. We found that human listeners can estimate the contributions of the source and the environment from reverberant sound, but that they depend critically on whether environmental acoustics conform to the observed statistical regularities. The results suggest a separation process constrained by knowledge of environmental acoustics that is internalized over development or evolution.

You might find the above line of research of interest in your quest to get to the bottom of this & optimize it further in your speakers?

It strikes me that BMR (Bimodal Mode Radiator) drivers produce decorrelated signals by default but I wonder do they satisfy the above criteria? Have you experimented with these BMR drives in the role of your back/ceiling firing speaker?
 
Last edited:
Ime it depends on the room - if you have a fair amount of space behind you then diffusion works well on the back wall. If you are very close to it, then you may have to resort to absorption.

Diffusors (at least the ones I've looked into) don't work well close up, need some distance from the ear and the source to really diffuse well.

Wondering whether you have tried running your LCS speaker from another amp, fed with a delayed version of the main signal? 10msec is about all the delay that can be had in most rooms by reflecting off the ceiling. In my (very limited!) playing with ambience tweeters fed through DSP and delay, it seems that it works better with more delay up to near 20msec or so (in addition to the added delay of most of that signal from walls and ceiling).
 
Yes, Griesinger is the man when it comes to concert hall acoustics - he even points out in one of his presentations where he finds Toole is incorrect.

The decorrelated reverberation is an interesting phenomena
Gary S. Kendall has published research & book (1995) on the 5 perceptual effects of decorrelated reverberation:
- the timbral coloration and combing is avoided because the constructive & destructive interference of multiple delayed signals is perceptually eliminated
- the diffuse sound fields akin to the late field of reverberant concert halls
- the externalization in headphone reproduction
- the lack of image shift because :
- the precedence effect is defeated

Perceptual research has progressed a lot since 1995 particularly the area of Auditory Scene Analysis (Bregman 1990) which is concerned with how we analyze/categorize all the vibrations that belong to separate auditory objects & how we are able to similarly follow separate auditory streams.

The perceived naturalness of decorrelated reverberation has to do with how sounds in nature are reflected off many surfaces before reaching our ears - essentially decorrelating the reflections. This, as you say allows us to perceive the sound as foreground & background auditory streams allowing us to separately evaluate the sound making object & the environment in which the sound was made. Recent research (2016) suggests that we use statistical summaries of such background streams & evaluate them as natural if they conform to the statistical regularities found in nature

You might find the above line of research of interest in your quest to get to the bottom of this & optimize it further in your speakers?

YESSSS! Thank you VERY MUCH for this information and the links!!

One of the "trade secrets" I alluded to earlier has to do with decorrelation.

It strikes me that BMR (Bimodal Mode Radiator) drivers produce decorrelated signals by default but I wonder do they satisfy the above criteria? Have you experimented with these BMR drives in the role of your back/ceiling firing speaker?

My understanding is the bending-wave radiators have very wide radiation patterns, and that would make it difficult at best to avoid relatively early-arrival contributions. Please correct me if I am wrong, as I really don't know much about them.
 
Wondering whether you have tried running your LCS speaker from another amp, fed with a delayed version of the main signal? 10msec is about all the delay that can be had in most rooms by reflecting off the ceiling. In my (very limited!) playing with ambience tweeters fed through DSP and delay, it seems that it works better with more delay up to near 20msec or so (in addition to the added delay of most of that signal from walls and ceiling).

I have not tried delaying the LCS signal, but that totally makes sense, and there's no reason why it shouldn't be done. Your 20 millisecond target figure sounds right to me too.

I've been focused on a passive solution with a high enough impedance to be safely run in parallel with the main speaker in most cases, but the "next level up" would probably be what you describe.

Did we meet at Lone Star Audio Fest many years ago?
 
Extra path length of about 20 ms means about 6 m. So, if the wall behind you is about 2-3 m away you should be fine. Direct reflection from back wall: 2 x 3 m. A reflection from backwall to sidewall to LP means the distance behind you can be decreased to under 3 m. If possible, it should be advantageuos if sound from left speaker could be steered / reflected from the backwall to sidewall and end up at you right ear. -To get that decorrelated sound mentioned in previous posts. A reflection from straight behind is of "less value" than those coming in at an angle to your ears, as side reflections are more easliy heard and can be of lower stregth.

If you cannot get rid of audible reflections from ahead / side of you before about 15-20 ms has passed I see little use to try to fix up something behind you. Serious designers for control rooms often strive for 20 ms (ideal) silence between direct sound and reflected sound, -an Initial Time Delay Gap (ITD or ITDG). Depending on room conditions one may have to accept 15 ms = OK, 10 ms not so good anymore, passable ?. Ideal strength for reflections during the "silent stage": -20 dB and weaker, -20 to -15 dB or so = OK.

If you can avoid early audible reflections but get fairly strong ones later on, say 15-30 ms, you fool your brain into beliving your room is larger and more spacious than it actually is. -The walls / ceiling seem to be further away as audible reflections are late in time corresponinding to further away.

If one succeds with that ITD, the soundstage will narrow in width but you should hear more of what is actually recorded, the room where recording took place, position of persons / instruments. Widening of the sound stage can be seen as a "false width". That doesn't mean it is unpleasant to get that "false width. Pick your poison.
 
Last edited:
To get 20ms delayed reflections, then how large listening rooms should we strive for?

20 milliseconds roughly corresponds to a 22 foot path length difference between the first-arrival sound and the reflections. So you'd either need VERY tall ceilings if using late ceiling splash, or a very large room if using a bipole or dipole system (with the added reflections now in the horizontal plane instead of going vertical), OR a delayed signal as bwaslo suggests.

If you can avoid early audible reflections but get fairly strong ones later on, say 15-30 ms, you fool your brain into beliving your room is larger and more spacious than it actually is. -The walls / ceiling seem to be further away as audible reflections are late in time corresponinding to further away.

The first time I experienced this was in a recording studio control room designed by Jeff Hedback. When I closed my eyes, it sounded like the room was about twice as big as its physical dimensions. It was a bit disorienting.

If one succeeds with that ITD, the soundstage will narrow in width but you should hear more of what is actually recorded, the room where recording took place, position of persons / instruments. Widening of the sound stage can be seen as a "false width". That doesn't mean it is unpleasant to get that "false width. Pick your poison.

Yesss!!! The early sidewall reflections "broaden" the image, so when we remove them the soundstage correspondingly narrows, but then we can re-position the speakers accordingly.

I'm an advocate of Geddes' approach, using speakers with a 90 degree pattern (-6 dB limits) in the horizontal plane, then toeing them in by 45 degrees, which results in negligible early same-sidewall reflections. In fact the first significant sidewall reflection of the left speaker will be the long across-the-room bounce off the right side wall, and vice versa. This also introduces de-correlation, as the first sidewall reflection of the left speaker's direct sound arrives at the right ear, and vice-versa.

A bit of creativity is called for if we also want to suppress the floor and ceiling bounces. Fortunately these are of less audible consequence than the first sidewall bounces, but those who have addressed them usually say it's worth doing.
 
diyAudio Moderator
Joined 2008
Paid Member
it seems to let you hear more of the acoustic space that's on the recording. It seems to facilitate the ear/brain system getting a good fix on the reverberation in the recording, as opposed to the reverberation signature of your actual listening room.
This is well achieved by reducing early reflections.
 
Last edited:
This is well achieved by reducing early reflections.

That's a vital part of it, but in my experience the additional late-onset reflections do make a significant improvement.

At RMAF 2018, Kal Rubinson of Stereophile came by our room. He listened to a bit and then asked us to turn the LCS section off and play the same thing again. Literally within the first two seconds he was saying something like, "oh, I see what you mean."

(Kal made it clear to us that he was not at the show to do a show report. He usually covers multichannel systems, but occasionally reviews particularly interesting two-channel, like the Bang & Olufsen Beolab 90.)
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.