Corner Bass Traps - Always beneficial?

[Bill Hart]

In the case of the subject at hand, bass frequencies, the room is absolute control. You can see that in this measurement:

So there, changing speakers other than absolutely power, will make little difference otherwise.

In frequencies above transition, as the chart shows, the speaker response dominates. At the risk of stating the obvious, changing speakers will most definitely change the sound in the room. That aside, there are design considerations that are specific to speakers. For example, for surround channels, the room may have been designed around a speaker with wide directivity. Changing that to narrow may somewhat change the response. As someone noted already though, in general a well designed room will work most speakers.

Thank you Amir. When I get to the 'next room' project, this is a key issue, among others.
I do want the flexibility to change speakers. And I'm thinking about BIG horns for the bass.

 

[Jeff Hedback]

quoting from self (and Nyal)

I thought this might be appropriate to reference the White Paper Nyal and I wrote last year. The breakdown of LF issues might be helpful to some. I particularly like the juicy quote from Dr. Toole that surfaced during a review session between Dr. Toole, Nyal and myself.



F: Low Range Frequency Response
Standard: In room low frequency (LF) response measurement at listening position should be:
• Within +/- 10dB at 1/24th octave resolution from 20Hz to 250Hz for both speakers measured together.
• Within +/-5dB at 1/3rd octave resolution from 20Hz to 250Hz for both speakers measured together.
Two measures are required to properly validate the quality of a room’s low frequency response. 1/3rd octave studies provide a view into overall spectral trends equating closely to how the human ear perceives loudness whilst 1/24th octave allows discrimination of individual resonances that cause timbral distortion.
The low frequency (LF) response of a stereo pair of speakers in a room is first dictated by the speaker and listener positions and then by the acoustical control that exists within. Boundary interference patterns are created within ~65ms when the LF energy of the speakers makes contact with the room surfaces and recombines with the direct energy. After 65ms the room modes will impart distortions if not properly damped (reference Section D).
To obtain the best possible LF response:
• The speaker and listener initial locations should be carefully selected based on proper study of room dimensions and practicality. The left and right speakers should create a spread of 45 to 60 degrees from the center point of the listening position.
• After a subjective review of the initial response, acoustical measurements should be taken to begin the fine tuning of both speaker and listener locations. These changes would typically involve movements greater than 6”.
• The remaining boundary interference issues can be tougher to address. Varying the fixed distances from “speaker to boundary” and “listener to boundary” will reduce strong cancellations. It is a balancing act as one location that may offer a smoother LF response may not provide the optimal midrange and treble response. Note: as speaker distances from the front wall may often be 4’-6’, a speaker boundary interference related null is likely to appear between 70Hz to 45Hz respectively. There would also be similar type null based on the relationship of the listening position to the rear wall.
• The frequency response should always be correlated to a resonance study as both are crucial to balanced and natural perception of LF transients.
• More evolved levels of analysis would factor single speaker data as well as the phase aspects of the in-room response (excess group delay).
At this point, it is appropriate to look closely at use of parametric EQ or digital room correction options to flatten the LF response. On this topic from Dr. Floyd Toole, “I try to be careful to say ’at low frequencies (subwoofer region below 80 Hz) small rooms behave as minimum phase systems (as it applies to equalizing resonances).’ When pushed I go on to say that this is mostly true for resonant peaks that stand above the average spectrum level, but as that level recedes below the average spectrum level, we clearly have a "signal to noise" (minimum phase-to-non-minimum phase) problem, because there are obviously non-minimum-phase activities in the region - interference dips being an example. This is relevant because the best evidence of an audible resonance is a highly visible peak, which, if attenuated to the average spectrum level (i.e. flat) results in a severe attenuation of ringing - which is the desired situation.”
In the larger sense, everyone desires a “flat” LF response and no modal ringing. Simply, this is a tough achievement. The absurdly large collection of interrelated variables between two fullrange speakers and the room (speaker design, speaker/listener location, room size/construction and acoustical control within) makes this so. It is up to the individual to determine what their limits are as regards placement and acoustic treatment.

 

[DonH50]

Don,

The room dimensions are 3.8x9.35x2.4 meters (13 x 30.5 x 8 feet). The walls are 2 feet stone - the worst possible in terms of bass damping. I have borrowed a Martin Logan Descent I subwoofer and I am going to "scan" my room with it using REW this weekend. The unfiltered sub is linear up to 200 Hz. From previous measurements I have found that the peaks are dominated by the main longitudinal axial modes, but could not relate mathematically the nulls with the dimensions or placement, something that bothers me.
As the room is very long, I am building a full wall membrane bass trap at the rear wall, but would like to understand what is going on before tuning it.

That is strange. If I decrease the length by 20% or so (since it is the largest dimension it most affects the bass modes) I get much closer to your 24 and 64 Hz measurements, making me think something(s) in the room is making the room acoustically "shorter". Or it could go the other way, match. This would vertainly vex me some until I figured out what is going on!

Here is what my simple program generates for room modes (Hz) in L, W, and H using your dimensions:

18.418 45.319 71.755
36.837 90.638 143.51
55.255 135.957 215.265
73.674 181.276 287.02
92.092 226.595 358.775
110.51 271.914 430.53
128.929 317.233 502.285
147.347 362.552 574.04
165.766 407.871 645.795

 

[DonH50]

The opposite of that is also interesting. As wavelengths get much smaller relative to the size of your head, then what the two ears hear is different. Your head also then presents a 0.4 millisecond delay to the ear further from the source (e.g. a reflection). Translating, nothing will seem like it is anymore! Psychoacoustics gets involved and the picture gets far less intuitive. I will post longer articles on this but for now, this is a good teaser .

True 'dat. Not a reflection, though, a time difference that impacts both the direction and potentially amplitude of the sounds, direct and reflected, making analysis, um, "challenging".

 

[Jeff Hedback]

Sharing example of corner trapping



This room features a beautiful set of Vandersteens. Analysis showed the basic location of speakers and listening position is very good for the space. Treatments were design based on room data and predictive tools. In the front corners are RealTrap Mega and in sidewall/ceiling junctures are GIK Monster. The additional panels behind the speakers are in fact RPG 4" BAD panels which have a very good LF absorption ability. All of these devices reflect or diffuse mid/hi freq's...the goal is balance after all.

I'm sharing freq response graphs and Resonance Studies (Burst Decay) Pre & Post. In all cases the woofer amp is at same volume and the EQ is not engaged at all. After the final measurements you see, the Vandy EQ specialist came in and did his thing (I have no data after that stage but trust that the 42Hz peak was tucked down just a touch).

The LF response graphs have an overlay of the very same Impulse response but the smoothing is changed. So there is a Pre graph that shows both 1/24th oct and 1/3rd oct. Same with the Final (or post treatment).

 

[Ethan Winer]

From previous measurements I have found that the peaks are dominated by the main longitudinal axial modes, but could not relate mathematically the nulls with the dimensions or placement, something that bothers me.

Yes, most peaks are modal, but nulls can occur almost anywhere. If you move the measuring microphone even a few inches the response changes drastically. The response at any given cubic centimeter location is the sum of the direct sound from the speakers plus many competing reflections coming from many different surfaces.

--Ethan

 

[microstrip]

That is strange. If I decrease the length by 20% or so (since it is the largest dimension it most affects the bass modes) I get much closer to your 24 and 64 Hz measurements, making me think something(s) in the room is making the room acoustically "shorter". Or it could go the other way, match. This would vertainly vex me some until I figured out what is going on!

Here is what my simple program generates for room modes (Hz) in L, W, and H using your dimensions:

18.418 45.319 71.755
36.837 90.638 143.51
55.255 135.957 215.265
73.674 181.276 287.02
92.092 226.595 358.775
110.51 271.914 430.53
128.929 317.233 502.285
147.347 362.552 574.04
165.766 407.871 645.795

Don,

The room is a perfect shoe box with two 1.5 square meter windows and a solid wood door. I tried opening them completely and the changes in frequency response were minimal. I attach the measured response in my room with 1/3 and 1/24 octave averaging. It was measured using REW and I am sure the dips are not due to the measuring equipment - they do not show in another system and other rooms. Please see that all the peaks are at the frequencies you computed and agree with my calculations.

 

[microstrip]

Yes, most peaks are modal, but nulls can occur almost anywhere. If you move the measuring microphone even a few inches the response changes drastically. The response at any given cubic centimeter location is the sum of the direct sound from the speakers plus many competing reflections coming from many different surfaces.

--Ethan

Ethan,

Thanks. Should we conclude that peaks are deterministic and nulls are hazardous? I find it rather embarrassing and annoying, specially because a null seems much more difficult to deal with than a peak.

This would support the believe of some people who feel that all the rules about the best dimension ratios are over rated and its importance exaggerated - you can have the best ratios and still have some unpredictable nulls.

 

[Jeff Hedback]

Microstrip,

All rooms (of this type/size) have modes...everyone of them. Anytime you place a speaker in a bounded space the spectrum below ~200Hz-300Hz is going to be distorted: first by surface reflections combining with the direct sound and then by modal resonances (or modal cancellations).
Why not have the most evenly space modes (i.e. good room dimensions), place your speakers as expertly as possible to diminish distortions, treat/damp resonances and possibly employ some EQ or DRC? All of these factors are important. Not every one of these has to be perfection to have excellent audio, but breaking them down into these types of factors can sure help figure out how to improve a given experience.

As for unpredictable nulls, some nulls are predictable and others are very complex to pinpoint the cause. This is a fact of physics but has no connection to the benefits of good room dimensions.

Commenting on your two graphs: per my own paper's thinking, we don't see the data up to 250Hz but from what is shown you are right on the fringes of the paper's goals +/-5dB at 1/3rd oct smoothing and +/-10dB at 1/24th oct smoothing. That is very encouraging. It would next be great to see a resonance study (waterfall or spectrogram from REW).

On correlation of measurements to our hearing: we are most sensitive to resonant peaks and wide nulls. This is why Nyal and I like to look at data in various ways and various smoothings.

In the room I shared above, the post treatment graph rates per the 1/3rd oct smoothed and 1/24th oct smoothed graphs similar to yours (noting my data is shown out to 250Hz...not the point though). Even if you target the singularly outstanding remaining issue of the 42Hz peak, is is clearly evident that the traps (primarily Ethan's Mega Traps in this range) have brought the resonance to within "balance" of the LF spectrum. As mentioned, the Vandy's have LF EQ section. It's easy to imagine how "flat" this space would become with proper use of EQ.

Sorry for the very long reply. It's nice to read your input and see your graphs.

 

[Ethan Winer]

Should we conclude that peaks are deterministic and nulls are hazardous?

They're both hazards.

I find it rather embarrassing and annoying, specially because a null seems much more difficult to deal with than a peak.

Yes, and in most rooms nulls are the larger problem. Peaks are typically 6 dB or less, but nulls are often 20 or even 30 dB deep. If a null aligns with bass notes in the key of whatever music you''re playing, the result is weak bass whose timbre is also affected.

This would support the believe of some people who feel that all the rules about the best dimension ratios are over rated and its importance exaggerated - you can have the best ratios and still have some unpredictable nulls.

Yes, all rooms have peaks and nulls. As Jeff explained, the advantage of good ratios is it spreads the peaks out more or less evenly, which avoids having two peaks at the same or nearby frequencies. But I wouldn't necessarily make a small room even smaller just to improve the ratio, because small is bad too. Further, ergonomics are important. If making a room smaller to get a better ratio means you don't have enough room for your gear or projector etc, again I wouldn't make the room smaller. So the best answer is "it depends."

--Ethan

 

[dallasjustice]

In my room the transition frequency is 250Hz. I have a null from 120-140Hz. Here it is in 1/12 octave smoothing:


I also have a couple of small nulls 48-73Hz and 77-95Hz. I have to say that my bass sounds pretty articulate and dynamic.

My speakers have powered subs built in. My room has 10 membrane bass traps; not much more room for additional trapping.

Could a second pair of subs in the back of my room potentially smooth out these nulls?

 

[NorthStar]

----- Probably. ...Strategically positioned in relation to your main listening position (seat).

* Someone before mentioned (Ethan) that moving only few inches, from your main listening position, can totally alter the sound equation, including those nulls. ...And it's very true.
You've probably seen complex graphs with various room's dimensions and with all the peaks and nulls mathematically measured at all the listening spots.
So, only few inches can make all the difference on what we hear, and don't hear. ...And at various frequency responses of the audio spectrum, and particularly in the lower octaves.

In my own room, I sometimes change position, just for another sound change (even if only by six inches).
And in my Home Theater setup, my center channel speaker, in relation to my main listening position, is off center! ...And for a very specific reason; the center line of any room's dimensions is detrimental for the very best 'sound' balance.

The best part about sound in our hobby, is that even after we set everything perfectly, including all the loudspeaker's positions and subwoofers (and acoustic room treatments); we can still EQ that sound by simply moving a few inches from our main listening position. It only requires a simple head's movement.

 

[microstrip]

Jeff and Ethan,

Thanks for your comments. I will take some more measurements with REW (waterfall included) and will ask again for your opinions.
I have the space to build a large bass trap tuned to 24 Hz, but would like to understand a little better what is really happening.

The only model I can see for it is that as the speakers are 7m from the back wall and I am at 3.5m (middle distance) the speaker and the back wall are resonating at 340 / (7m x2 ) = 24 Hz and the listening position is at the pressure valley. In this case suppressing or attenuating the refection at the back wall with a very large bass trap covering the whole wall could help.

 

[audioguy]

Recommendation:

Using REW or equivalent, try moving the mic up and down to determine if the null is from the floor/ceiling bounce. You can also move left/right and forward/back to test. In my case, when I moved either left/right or forward/back the q of the null changed, the amplitude changed a small amount and the center of the null changed a small amount. However, when I moved vertically, I could completely eliminate the null. So I hung bass traps from the ceiling and NO MORE NULL!

If your null is from the floor/ceiling bounce, you can build corner traps until your broke and you won't really solve the problem !!

 

[microstrip]

Recommendation:

Using REW or equivalent, try moving the mic up and down to determine if the null is from the floor/ceiling bounce. You can also move left/right and forward/back to test. In my case, when I moved either left/right or forward/back the q of the null changed, the amplitude changed a small amount and the center of the null changed a small amount. However, when I moved vertically, I could completely eliminate the null. So I hung bass traps from the ceiling and NO MORE NULL!

If your null is from the floor/ceiling bounce, you can build corner traps until your broke and you won't really solve the problem !!

Audioguy.

It what I am planing to do - using a generator tuned at the frequency of the null and the sound-meter to get a 3D map of the intensity of the null frequency in my room.
What was the frequency of the null you were dealing with?

 

[audioguy]

I actually just ran frequency sweeps instead of the specific null frequencies to see the overall effect of moving the mic. If you don't do that, you might remove the null at one frequency and it may show up elsewhere!

My null was about an octave or two between about 150hz (plus or minus) to about 300hz (plus or minus) and about 10+db's deep! It made the frequencies above it more prominent and added to an already exaggerated midrange/upper midrange glare !

 

[AudioExplorations]

Micro, let us know how you get on. Some graphs would be interesting to see.

I too am stuggling with a significant null. Last night I re-measured and tried to boost it (due to comments earlier in this thread) but that was a bad idea... the speakers started distoring and the sound was terrible. IMO EQ should only be used to pull down peaks.

Note also that the speaker design/configuration plays a huge role with floor-to-ceiling room issues. Some speakers have their woofer halfway up in the room (e.g. TAD CR1), some have the woofer low and near the ground (e.g. Rockport Mira/Ankaa/Aquila), and some have dual woofers both at the bottom and top of the speaker (e.g. larger rockports or evolution acoustics). I imagine the dual woofer approach to be best for a smooth response, akin to dual subwoofers.

 

[edorr]

View attachment 4314 View attachment 4315 View attachment 4316 View attachment 4317 View attachment 4318

This room features a beautiful set of Vandersteens. Analysis showed the basic location of speakers and listening position is very good for the space. Treatments were design based on room data and predictive tools. In the front corners are RealTrap Mega and in sidewall/ceiling junctures are GIK Monster. The additional panels behind the speakers are in fact RPG 4" BAD panels which have a very good LF absorption ability. All of these devices reflect or diffuse mid/hi freq's...the goal is balance after all.

I'm sharing freq response graphs and Resonance Studies (Burst Decay) Pre & Post. In all cases the woofer amp is at same volume and the EQ is not engaged at all. After the final measurements you see, the Vandy EQ specialist came in and did his thing (I have no data after that stage but trust that the 42Hz peak was tucked down just a touch).

The LF response graphs have an overlay of the very same Impulse response but the smoothing is changed. So there is a Pre graph that shows both 1/24th oct and 1/3rd oct. Same with the Final (or post treatment).

Nice. I found the 34" wide (front) mega traps from realtraps 1 day after I ordered corning 703 to build a similarly sized floor to ceiling trap myself, and would have probably bought 6 megatraps myself had I known they existed.

What are the panels on the ceiling? I don't see these on the GIK monster site.

 

[Jeff Hedback]

Thanks edorr,

On the ceiling are Auralex Q'Fusors which are 3" profile folded QRD type diffusor. On the sidewalls are RPG Skyline diffusors. And the GIK Monster run down the sidewalls in the junction with the ceiling (with spacing between panels). These speakers really thrive when diffusing first reflections points. I've attached a T30 (decay time study) Pre & Post to see how the devices used resulted in control and balance.

I was very pleased with the combination of devices. As has been nicely pointed out in this thread, it's not easy to "trap" 40Hz...the Real Traps Mega definitely played a strong role this this effort. Obviously the traps didn't erase the 42Hz peak, but they certainly damped the resonance!

 

[rbbert]

I was just (again) looking at photos of mid- to high 5 figure (dollar) systems at the Steve Hoffman Forums with no signs of room treatments or even noticeable attention to speaker placement (sigh)??