Wanted: Proof of multiple subs and sub EQ

[AJ Soundfield]

Fazenda also researched the so called "source and sink" subwoofer method. This is using subs on opposite walls in opposite electrical polarity or phase. This method most certainly reduces rate of decay without any EQ. This fazenda experiment shows that listeners preferred it over other sub setups.

http://www.aes.org/e-lib/browse.cfm?elib=16324

http://www.academia.edu/5362839/Perception_Modal_Control

Mono bass below 120Hz, no compensation for group delay between subwoofer and full frequency range speaker....
Using sub(s) as a sink like Andy (H.G.) suggested earlier in thread.

cheers,

AJ

 

[dallasjustice]

Thanks for finding the paper.

There is delay applied to the rear sub so that the "sink" happens behind the MLP. I never said this was anything new. Fazenda said in his paper that he is expanding on the so called CABS subwoofer setup that Celestinos described in a previous sub study and he cited to it.

http://www.academia.edu/5362839/Perception_Modal_Control

Mono bass below 120Hz, no compensation for group delay between subwoofer and full frequency range speaker....
Using sub(s) as a sink like Andy (H.G.) suggested earlier in thread.

cheers,

AJ

 

[dallasjustice]

There's also a highly mathematical paper DIRAC Research published demonstrating how multiple speakers can be used to cancel modes and fill in Allison effect. I admit I don't understand the math but the basic concept has been proven to work. I believe Amir said he heard their method at CES demo a couple of times. I'd love to hear more about what that sounds like.

I would not be surprised if their method is a much more advanced version of the source and sink or CABS methods I mentioned above.

 

[Ethan Winer]

I guess I’ll go ahead and give it a go. So strap yourself in.

Excellent, thanks very much for all you posted in such a detailed manner. The only thing I'll add is:

According to Nyal Mellor at Acoustic Frontiers (not Sonic Frontiers), the spectrogram view in REW does indeed show the rate of decay independent of level. I haven't used that display enough to know it's features and limitations because, like you, I find it less intuitive and less precise. But for now I'll accept on the surface what Nyal says until shown otherwise.

The real issue for me is not whether ringing can be reduced at a tiny one-cubic-inch location, but what happens a foot away and five feet away. Amir has posted data showing a real reduction in ringing at one location. So does Nyal. So do a number of "scholarly" articles published in the AES Journal. But not one of them shows what I have been asking about for many years now! As I (think I) mentioned earlier in this thread, I've tested two different EQ systems and neither improved ringing. So it's not like I'm unwilling to do the work myself. Proponents can't claim I screwed up either, because both systems (Audyssey and Dirac) adjust their EQ automatically.

So while I'm willing to be shown evidence that EQ and / or multiple subs can reduce modal ringing decay rates around a room, until I see it for myself I'll continue to assume they can't do that even 1/4 as well as bass traps can, even at very low frequencies.

--Ethan

 

[Ethan Winer]

BTW, to be perfectly clear, below is my write-up of the series of tests I have asked three different proponents for, but all were too busy to do. So I put this out here for all, still hoping someone will do these tests.

--Ethan

Even though logic, physics, and my own tests with Audyssey and Dirac concluded that EQ and multiple subwoofers cannot reduce ringing enough to be useful, some proponents insist otherwise. Over the past several years I have three different proponents to provide measurements proving their point. Not just at the one location where the measuring microphone was placed during calibration and optimization, but also three feet away. If reducing ringing is to have merit, it must hold for all seats, not just a tiny area smaller than a person's head. So here is my challenge for anyone who believes that EQ or multiple subwoofers can reduce ringing a useful amount all around a home-size room:

Please measure your room using the Room EQ Wizard software as described below, then send me the file using the email address on my home page. I need two sets of six measurements if you have four subwoofers, otherwise four measurements per set if you have only two subs. Measure the first set with the microphone at the main listening position:

1) One subwoofer active, no EQ (baseline)
2) Two subwoofers active, no EQ
3) Four subwoofers active, no EQ

4) One subwoofer active, with EQ
5) Two subwoofers active, with EQ
6) Four subwoofers active, with EQ

Then after the above measurements were made with the microphone at the main listening position, move the microphone about three feet to the left or right (one seat over) and measure again. Please do not change or re-optimize the balance or phase etc between subwoofers, or change the EQ settings:

7) One subwoofer only, no EQ, (baseline for the new location)
8) Two subwoofers active, no EQ
9) Four subwoofers active, no EQ

10) One subwoofers active, with EQ
11) Two subwoofers active, with EQ
12) Four subwoofers active, with EQ

Please also include a photograph of your room showing the main speakers and any subwoofers that are in the front part of the room. If you're able to run a camcorder while you're testing, please send me the video too!

 

[AJ Soundfield]

I've tested two different EQ systems and neither improved ringing.

The only thing I can find on your website(s) is Kal Rubinsons (single sub?) 2007 test: http://realtraps.com/art_audyssey.htm
Direct links to your multi sub + Audyssey and Dirac tests please

 

[Mark Seaton]

What do you mean by "natural" room ringing?
I walk into a room, where is the natural ringing? Must an acoustic source be involved?
Let's say I place a pressure source 3' out from both walls in one corner, place a pressure mic at the LP, measure the pressure/decay, will the "natural" ringing show?
Ok, now I place a velocity source (gradient) with the same free field response as the pressure source, in the exact same spot in place of the pressure source. The axial, tangential and oblique modes are now differently excited, the pressure/decay response at the LP mic is quite different.
I rotate the gradient by 45 degrees in the exact same spot. The modes are once again excited quite differently, the pressure mic once again show a different response at LP.
Of the 3 different responses, which will properly show the "natural" ringing of the room?

cheers,

AJ

btw, your measurement examples do show the effectiveness of "midbass" traps.

I hesitate to jump into such a long thread only scanning the last few pages, but to possibly reduce some wasteful debate, I believe what Wayne was referring to more so is not ringing/excitation related to sub positions, but the rate of decay once the room is energized. This would be akin to being past the critical distance in a larger room where reflections are stronger than the direct sound. While different subwoofer radiation patterns can reduce the energy into certain modes, if you have enough energy to deliver the frequency to the listening position, that sound will hit the back wall and have to decay based on the room's boundary interactions just as any sound would.

What you will likely find is that differences from directivity could affect initial decay or excitation, but on a longer time window, sound will decay at a similar *rate* for most locations. This effectively defines a minimum rate of decay that electronics, placement, and directivity will be hard pressed to reduce. The conundrum is that in a small room at lower frequencies it's near impossible to separately analyze modal ringing vs decay rate like you could in a large space. This conundrum is also why EQ has such a dramatic effect in reducing the ringing in most instances.

 

[Mark Seaton]

BTW, to be perfectly clear, below is my write-up of the series of tests I have asked three different proponents for, but all were too busy to do. So I put this out here for all, still hoping someone will do these tests.

--Ethan

Hi Ethan,

I tend to have plenty of subwoofers around, but I need to work on more that are easy to move around for such experiments (actually in the works).

I expect "proof" of an FIR based room EQ providing better results than more classic PEQ and IIR filters will be tricky at best, and in my experience results would vary a good deal with the user, procedure, and manual adjustments made.

On the matter of multiple subwoofers having the *potential* to offer less ringing over a larger listening area, I don't think this there is much debate. We know that smoother response will result in less ringing, which we could also define as more even decay of sound vs frequency. If we start including rooms that have multiple front-to-back rows of seats, you would be looking at some rather extraordinary efforts to acoustically make the response uniform below 50-100Hz. You are of course correct that EQ can only make global adjustments that apply to all locations. In the future I expect we'll see this change as electronic manipulation of multiple devices at lower frequencies can make changes across a space.

I set up the subs in a large theater that I had posted a bit of data from on AVS (not as much as I would have liked to if I had more time and measurements, but I do have products to ship). The measurements are all TDS magnitude sweeps from my TEF25 system, which is highly consistent with high noise immunity. These measurements do not directly include time domain data, but the response changes observed are so significant it should be obvious that ringing is improved. Yes, when playing with delays it is good practice to keep an eye on a waterfall plot to insure you haven't created some strong ringing that isn't obvious in a magnitude plot.

Unfortunately in taking so many measurements I didn't save more starting point measurements from the front row showing the starting point with just the front subs or no delay, but I did have some comparisons uploaded on the main vs back row. The front row was actually served very well by a set of 4 subs placed across the front wall. With the 4 front subs operating, you can see the magenta, yellow, and red curves indicating middle, right and left seats in the middle row. The green curve was the response from the same 4 subs in the middle of the rear row:

I did a lot of testing of available sub locations prior to the setup, and found the rear right corner to be a very efficient compliment to the front subs. We added 2 identically performing subs at the rear right corner (vs 4 at the front wall). Here is a measurement from the center of the middle row with the front 4 subs from above in yellow and 2 subs at the rear in green. All 6 units are producing the same energy and getting the same signal.

This looks promising, but the relative distances and how they interact with the room resulted in severe cancellation seen below in the red curve. The blue curve was achieved by simply applying a 7ms delay on the rear 2 subs. While not the most efficient, it was close, and proved a good match in the front and rear rows:

Remember that green curve from the back row with just the front 4 subs? With rear subs delayed 7ms we now have the measurement below from the center of each of 3 rows. The middle row is now green, and the rear row is magenta, with the front row also shown in yellow:

My question to you then is what sort of acoustic treatment plan would be needed to get this sort of uniformity across 3 rows of a 28' deep home theater? I'm excluding the sledgehammer and open air to the entire back wall as a viable solution.

Another view of the above measurement referenced to the middle row:

 

[Ethan Winer]

We know that smoother response will result in less ringing, which we could also define as more even decay of sound vs frequency ... These measurements do not directly include time domain data, but the response changes observed are so significant it should be obvious that ringing is improved.

It's not obvious to me. Ringing and a flat response are different and can easily be independent. This is why I find myself asking for proof again and again and again. EQ and multi-sub proponents claim this is "settled science," but for some reason they never have proof of reduced ringing at multiple locations. As soon as I see this proof I'll gladly change my opinion!

My question to you then is what sort of acoustic treatment plan would be needed to get this sort of uniformity across 3 rows of a 28' deep home theater? I'm excluding the sledgehammer and open air to the entire back wall as a viable solution.

Yes it takes a lot of bass trapping to counter reflections in a room. The smaller the room, the worse the problems are. My living room is 25 feet long by 16 feet wide and I have many bass traps. I have one rear "row" but nobody every sits there so I never measured.

--Ethan

 

[Duke]

I have a dog in the fight, so disclaimers and grains of salt all around.

My understanding is that the rate of decay at low frequencies is largely a function of room damping, and that the relative decay times across the bass spectrum are largely a function of the smoothness of the frequency response.

So if we want everything to decay faster, we add damping (well obviously Helmholtz absorbers would be able to target a specific frequency). And if we want everything to decay more evenly, we'd want to smooth the frequency response.

It is my understanding that the ear has very poor time domain resolution at low frequencies. I recall a study published in the AES Journal some years ago (sorry my membership has expired so I can't cite it) that looked at group delay, and found that group delay typical of a bass reflex system was barely audible on test tones and statistically inaudible on program material. They used digital signal processing to separate group delay from any changes in the frequency response. Once the frequency response changes associated with a reflex box are back in the picture, the differences become audible. This is because the ear is very good at hearing SPL vs frequency in the bass region.

As evidence of the ear's heightened sensitivity to frequency response anomalies down low, consider a set of equal-loudness curves:

Notice how the curves bunch up at low frequencies. This means that a small change in SPL at low frequencies makes a disproportionately large change in perceived loudness. In the 80 to 100 dB region, a 5 dB change at 30 Hz sounds like a doubling of loudness, whereas over most of the spectrum it takes a 10 dB change in SPL to sound like a doubling of loudness. This is why peaks in the bass region stick out like sore thumbs - the ear is literally extra-sensitive to changes in SPL down there. This is also why it takes so darn long to dial in the gain control on your sub(s)... a small change in SPL makes a disproportionately large change in perceived loudness.

So if we combine these two pieces of information (the ear is relatively insensitive to time domain issues at low frequencies, and the ear is extremely sensitive to SPL-vs-frequency at low frequencies), then it becomes apparent that the higher priority is, fixing the frequency response. This is NOT TO SAY that there isn't worthwhile improvement from increasing the damping in the room at low frequencies! I am certain that there is.

I would not claim that a multisub system reduces the rate of decay in the bass region. But I would claim that, to the extent subwoofer(s) + room constitute a minimum-phase system (as claimed by Geddes and Toole), improving the response in the frequency domain simultaneously improves it in the time domain, so the decay times are improved (because a peak takes longer to decay since it starts out louder). I would also claim that, perceptually, the frequency domain is what matters most (even though both are improved simultaneously). And I would argue that an intelligently distributed multisub system exhibits significantly less spatial variation with respect to frequency response than a single equalized sub, so the perceptual improvement holds up throughout the room.

Last month at Axpona the Essential Audio room showed a distributed multi-sub system in a small hotel room whose low-frequency damping was no doubt every bit as bad as all the other small hotel rooms. A fair number of visitors remarked that it was the most natural-sounding low end they had heard at the show, and many made a point of including in that statement large rooms with many-times-the-price systems on display. From a post on Robert E. Greene's forum: "The bass was terrific; easily the best I heard at Axpona and some of the best reproduction of low frequencies I've heard anywhere, ever". So there you have my self-serving anecdotal pseudo-evidence that a distributed multisub system can give good results even in rooms with poor low-frequency damping. And if that ain't science-y enough for ya, then I'm on the wrong forum...

 

[dallasjustice]

Amir, I need to like a post more than once. I need about 10 more likes for Mark and Duke's last couple of posts.

 

[amirm]

Amir, I need to like a post more than once. I need about 10 more likes for Mark and Duke's last couple of posts.

.

Great to have so many experts in this thread!

 

[Mark Seaton]

It's not obvious to me. Ringing and a flat response are different and can easily be independent. This is why I find myself asking for proof again and again and again. EQ and multi-sub proponents claim this is "settled science," but for some reason they never have proof of reduced ringing at multiple locations. As soon as I see this proof I'll gladly change my opinion!

Yes it takes a lot of bass trapping to counter reflections in a room. The smaller the room, the worse the problems are. My living room is 25 feet long by 16 feet wide and I have many bass traps. I have one rear "row" but nobody every sits there so I never measured.

--Ethan

You not measuring the rear row is akin to our not measuring the low frequency decay. Unless you've created something really peculiar, it's pretty rare for a mostly smooth frequency response to result in a very resonant, non-minimum phase resonance. These are much more common near deep/narrow nulls in the response which have been boosted. Note the measurements I posted have ZERO PEQ applied. This is effectively just using multiple locations and phase adjustment.

In more recent calibrations I have more often worked with REQ and a UMIK-1 or Earthworks mic for convenience, and waterfall plots are much quicker to generate with REW than my TEF25. At some point I'll take some measurements and post them, or someone else is likely to. I just don't have a relevant set of data saved.

Mind you I am a big proponent of thick, broad band acoustic treatments in a room, especially below 500Hz. Some very audible gains can be had reaching easily down to 50-100Hz. The reality you skimmed past in your response of "a lot of bass trapping" is that moving beyond a moderate improvement in frequency response balance and toward more uniform response over multiple rows generally requires FEET of treatment over significant area, not just a few more loosely defined, "bass traps." In dealing with the wavelengths and dimensions involved, multiple sources with different boundary and modal interactions and a sprinkling of electronic correction pose a dramatically more elegant solution than treatments alone.

 

[Mark Seaton]

So then I tested Dirac just last summer. It reduced ringing in my large home recording studio a very tiny amount - certainly nothing like the improvement you get from even a few bass traps. It also boosted my main null around 100 Hz too much, making the sound wonderfully full but unnatural. This wasn't just my subjective opinion. I also had my friend Ed Dzubak visit and listen to some of his own music. Ed is composer with four Emmy awards for his work on many TV shows. Ed said the same: Dirac added too much bass, without Dirac the sound is more accurate.

--Ethan

The null in your room around 100Hz is one of those cases where intelligence in the correction system and the method for collecting enough measurement data need more polishing. To some degree some software can determine what is or is not minimum phase behavior. This determination, combined with measurements at other locations can help determine what should or should not be fixed, and to what degree.

None of the correction systems do a great job with a single point of measurement data, just as a good calibrator will want info from more than one microphone location. The issue is that it is also easy to include data that is not useful, and the system can't easily determine which is which. This is where the manual input on systems like Dirac and Trinnov can help get some exceptional results. A few moments in editing the target curve and examining a decay plot should have helped you quickly found the best balance of correction vs. overall energy around that 100Hz null. It's very easy to quickly make the Dirac curve trace the null and not try and correct what it should not. Dirac's ability to show the measured average, individual responses, and target is quite useful. In many cases I find the best results by approximately tracing the upper frequency response shape of a speaker rather than completely skipping the correction of that range. There are many things that happen in the overall power response that we will always have trouble capturing with a finite set of measurements, which is why having manual intervention or more appropriately, points of input and direction, will still make for the most consistent results.

 

[Ethan Winer]

My understanding is that the rate of decay at low frequencies is largely a function of room damping, and that the relative decay times across the bass spectrum are largely a function of the smoothness of the frequency response.

Decay times are wildly uneven in rooms because the resonant frequencies (modes) store much more energy than all the other non-mode frequencies. Yes, peaks are related to extended decay times, but using an EQ to reduce a peak doesn't guarantee that the decay time will be reduced. For example, an EQ setting with a very low Q can reduce a high-Q room peak, but that won't affect the decay time. A linear phase EQ can reduce a peak's level but will probably have no affect at all on the decay time.

It is my understanding that the ear has very poor time domain resolution at low frequencies.

It's true that frequency response is probably the most noticeable of the four audio parameters, but decay times are important too. If it takes one second for a bass note to decay to silence, everything else within an octave of that note's frequency will be masked or at least muddied during that one second.

From a post on Robert E. Greene's forum: "The bass was terrific; easily the best I heard at Axpona and some of the best reproduction of low frequencies I've heard anywhere, ever". So there you have my self-serving anecdotal pseudo-evidence that a distributed multisub system can give good results even in rooms with poor low-frequency damping. And if that ain't science-y enough for ya, then I'm on the wrong forum...

You are correct, that is not nearly "science-y" enough for me.

--Ethan

 

[Ethan Winer]

You not measuring the rear row is akin to our not measuring the low frequency decay.

Not really. As I already explained, nobody ever sits back there. The only reasons I still have that two-seat piece of our larger couch there at all is 1) I have no other place to put it, and (more important) 2) if I get rid of it my wife said she plans to put a dining room table there and start inviting people over for dinner. I'm serious. When people visit us now we all go out for dinner, and I prefer it that way!

waterfall plots are much quicker to generate with REW than my TEF25. At some point I'll take some measurements and post them, or someone else is likely to. I just don't have a relevant set of data saved.

Yes, please do! See my post 225 above. I'd love for someone with multiple subs and the knowledge to set them optimally to do these measurements and share the REW file with me.

Mind you I am a big proponent of thick, broad band acoustic treatments in a room, especially below 500Hz. Some very audible gains can be had reaching easily down to 50-100Hz. The reality you skimmed past in your response of "a lot of bass trapping" is that moving beyond a moderate improvement in frequency response balance and toward more uniform response over multiple rows generally requires FEET of treatment over significant area, not just a few more loosely defined, "bass traps."

My Early Reflections article lists the surface area of my living room and all the acoustic panels combined. Here's the summary:

Total room size is 1,579 square feet.
Total square footage of 44 acoustic panels = 280 square feet

So yeah, I have a lot of treatment. And it really helps.

--Ethan

 

[RayDunzl]

I have no other place to put it

Makes sense to me (moving in picture. It might still be there, but the little table isn't)

 

[Fitzcaraldo215]

Makes sense to me (moving in picture. It might still be there, but the little table isn't)

I get it! I need to put an auxiliary sub in the bathroom. Thanks, Ray.

 

[Mark Seaton]

My Early Reflections article lists the surface area of my living room and all the acoustic panels combined. Here's the summary:

Total room size is 1,579 square feet.
Total square footage of 44 acoustic panels = 280 square feet

So yeah, I have a lot of treatment. And it really helps.

--Ethan

Sorry for the quick post and lazy language there. I meant it takes treatments that are *multiple feet thick* over significant area to have dramatic impact on subwoofer frequencies such as what multiple subwoofer locations can do.

Much of the debate here comes down to specific language and formal definitions vs colloquial terms. Much info has been posted in various places online showing how a properly fit PEQ can reduce modal ringing and decay time. The key is the shape of the filter where you do need to see the decay to get the best match. While Duke mentioned that we are most sensitive to amplitude, I would say it's important to correct this to saying we are hearing sound power, not just maximum intensity. Power is defined as intensity over time, so this means the subjective loudness is directly tied to how long the sound sticks around or how quickly it decays. This is exactly what you have described and stated with the importance of the time domain.

The question is then what is the rate of decay after we use any number of methods to make the initial intensity more similar vs frequency. For those reading along, a good way to view this is what does the balance of sound look like at say 250ms or 500ms after the initial sound? This was actually the basis of Meridian's early room EQ where PEQ was used by looking at the spectrum of energy still present after some time window (300-600ms IIRC).

 

[Duke]

Decay times are wildly uneven in rooms because the resonant frequencies (modes) store much more energy than all the other non-mode frequencies.

Frequency response is wildly uneven in rooms for exactly the same reason! Fix the frequency response and you've fixed the decay times. And vice-versa. It's the same problem!

This is a fundamental property of minimum-phase systems. According to people far smarter than me (Floyd Toole and Earl Geddes), sub(s) + room constitute a minimum-phase system... and if so, this correlation between the frequency and time domains is valid.

You are correct, that is not nearly "science-y" enough for me.

Ha! At least you admit that I was right about something!!

While Duke mentioned that we are most sensitive to amplitude, I would say it's important to correct this to saying we are hearing sound power, not just maximum intensity. Power is defined as intensity over time, so this means the subjective loudness is directly tied to how long the sound sticks around or how quickly it decays.

Agreed!