The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
SBIR
Old 15th January 2012
  #241
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Jens I absolutely understand.

The question throughout here is one of level. I can't see one layer of plasterboard 2 Metres away as 'High gain'

You are good with Arithmetic and such.

How about you calculate the combined level of the 3 reflections caused by equidistance from floor, wall, wall.

Then calculate the level of the next loudest reflection, from a wall 3 Metres away. Diminish it's reflectivity slightly due to the plasterboard and insulation layer.

Lets see how they stack up.

In Noise Control measurement, the guideline is to ignore any factor 10dB lower than the others as it will not contribute significantly to the log sum.

I have an open mind here. If you wish to convince me of the power of all these far away and/or weakly reflective boundaries, do it by numbers.

Without that we are both guessing aren't we?


DD
Dan, again; there are 7 reflection paths from a single isolated corner (even if in free space), not only three, and these will have different arrival times relative than the first. Adding all of these will change the response and the fc of the dip (at the source) will drop as a result. Again, assuming very solid walls (not the normal real world condition unfortunately). If not, complex impedance and thus a phase change (more or less at some or more frequencies) is likely to occur and the whole situation suddenly becomes very complex and had to predict.
Old 15th January 2012
  #242
Lives for gear
 
akebrake's Avatar
 

Quote:
Originally Posted by cborg View Post
Thanks akebrake!
You don't know if there is any review or larger summary of that original paper available on the net?
Christian, your’e welcome!
Good question! Unfortunately I have no clue. I haven’t searched actively for it, but I will. This thread made me curious !
Thanks cborg and DD as the OT.

I like to expand a little on my claim: ”SBIR and LBIR are two different animals” from # 190 post.

The ”animals” I refered to was SoundPower Response and Frequency response respectively. Sorry for being vauge.

As I understand it, booth Allison (1975) and Berger (1983-84) focused on sound power response of the source. And Berger did the prediction program and he chosed to include three boundaries close to the source. Eg. within 2 ? and up to 500 Hz.
He dubbed it SBIR. It’s restricted to the source and it’s coupling to the nearby ”corner” are reflected (pun intended) in the radiated power curve. (rigid boundaires assumed)

Many years later, Cox & D’Antonio developed these original ideas furter. Early research was of course included in this (more involved) prediction program, the Room Optimizer.
The authors, Cox & D’Antonio have referenced to earlier work by Allison, Waterhouse etc. as scientists usually do.

Quote:
”We refer to this as the speaker-boundary interference response. This issue has been examined by Allison [9,10], Waterhouse [11,12] and Waterhouse and Cook [13]. ”
This paper was also referenced by Jens in an earlier post

http://www.rpginc.com/news/library/ropt_wp.pdf[/QUOTE]

But Cox/D’Antonio chosed to include more boundaries (later in time) in the computation and to the resultant frequency response. (”LBIR”)
Their SBIR term was refered to as the”short term spectrum” in the Room Optimizer program. A 35ms time window (30-65ms) which of course included several boundaries in most rooms e.g. the back/rear wall.

(A ”long term spectrum” was also included in the total spectrum involving the influence from room modes, which needed more time to build up.)

My conclusion:
The original SBIR term a la Berger relates to the source only. (Power)
The modified SBIR term Cox/DAntonio relates to the
frequency response (at listener pos) but within a short time window 30-65 milliseconds.

(E.g some psycho acoustic weigth of the perceived early reflections. My com.)

These are still prediction programs (rough model of reality)

Can we measure SBIR?
Toole suggested spatial avaraging of several LF response curves to see the influence of the loudspeaker position. (Adjacant boundary effect)
E.g. Average out the listeners postion dependent variations.

Cox& D’Antonio:
Quote:
”To illustrate the effect in Figure 3, the speaker-boundary interference is averaged over listening positions [14]...
And later, their conclusion:
Quote:
”By moving the loudspeakers and listeners to optimum positions in the room, the coloration produced by the room transfer function can be greatly
reduced.”

See also Jeffrey Headback's post re. REW measurement software.
Quote:
Originally Posted by Jeffrey Hedback View Post
Spurred on by some of Jens' great posts and links, there is a way (I'm suggesting) within REW to separate the Boundary Interference & Modal aspects of a response.

Eventually some more light?
Old 15th January 2012
  #243
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by Rod Gervais View Post
I have always assumed that the effect was caused by the summing of the signals at the speaker, and although I have been curious about the possibility of room pressure affecting cone travel - I really do not have a strong feeling that this is the case with SBIR.
I don't think most reflections would be strong enough to affect cone displacement, and certainly not when the reflections come from behind the speaker, which I think is what we're mostly considering here. The main reason SBIR exists at all is because most "box" type loudspeakers start to radiate omnidirectionally below around 300 Hz. This is also why the SBIR peak and null frequencies are related to the spacing between the rear of the speaker cabinet and the boundary, rather than from the cone in the front of the box. The rear of the box is actually the sound source location.

--Ethan
Old 15th January 2012
  #244
Lives for gear
 
akebrake's Avatar
 

Acoustic center

Quote:
Originally Posted by Ethan Winer View Post
... This is also why the SBIR peak and null frequencies are related to the spacing between the rear of the speaker cabinet and the boundary, rather than from the cone in the front of the box. The rear of the box is actually the sound source location. --Ethan
I don't quite follow you here...?
Measured with a software measuring instrument like REW, Fuzz Measure, TEF etc. the "starting time" (acoustical center) of a loudspeaker is often behind the front baffle at lo frequencies when seen from the listener / measuring position. E.g. Low frequencies are "late". This will also be observed on the "relative phase curve" as a steeper slope for the lowest frequencies.

IMHO If you measure the 1/4 wavelength distance from the rear of the speaker cabinet to the front wall (control room window) you will end up with to short distance. E.g too short PLD (path-length-difference) and too high frequecy of the cancellation dip.
Old 15th January 2012
  #245
Gear Addict
 

Time Gated Overlay Studies

What you are seeing, using ARTA for post processing, is the same Right Spkr Impulse Responses Over-Layed onto themselves of two rooms. The short time gate (usually 45ms-60ms depending on room size) gives enough time for the boundaries to impact the measurement at low frequencies. The longer time gate is representative of what's heard in the room.

Image one is pre-data of a new client's small listening room. The short time gate in this case is 50ms as the room is small (~10'x13'x8'). The speaker is nearly 2' off the front wall (corresponding to null at 150Hz). The mic was 3.5' off the rear wall (which corresponds to 80Hz null in short gate). Comparing the short time gate to the long time gate, you can immediately see how the room resonances "over-power" the immediate boundary interference effects.

The second graph is a treated room where we are working on final tweaks to speaker/mix positions (before integrating a pair of Rythmik Audio subs to the JBL LSR 6332's which have F3 of 54Hz). You can immediately note how the short time gate (60ms in this case) and the long time gate (300ms) are nearly identical. This indicates essentially no modal resonance. So the response is all boundary interference and the actual speaker response.

I've found this approach very helpful to know what battle to fight.
Attached Thumbnails
SBIR-untreated-room-pre-data-_overlay-time-gated-study.jpg   SBIR-treated-room-post-data-_overlay-time-gated-study.jpg  
Old 15th January 2012
  #246
Gear Guru
 
Ethan Winer's Avatar
 

Lightbulb

Quote:
Originally Posted by akebrake View Post
Measured with a software measuring instrument like REW, Fuzz Measure, TEF etc. the "starting time" (acoustical center) of a loudspeaker is often behind the front baffle at lo frequencies when seen from the listener / measuring position. E.g. Low frequencies are "late".
Sure, but that's for listeners in front of the speaker. The "lateness" is also due to inherent delays added by the crossover. But I'm pretty sure that SBIR from a boundary behind the speaker is measured from the edge of the box.

I sure wish someone would measure this already. heh

--Ethan
Old 15th January 2012
  #247
Lives for gear
 
boggy's Avatar
Quote:
Originally Posted by Rod Gervais View Post
.......
For what it's worth - I have always assumed that the effect was caused by
the summing of the signals at the speaker, and although I have been curious about the possibility of room pressure affecting cone travel - I really do not have a strong feeling that this is the case with SBIR......
......
Nearby boundaries can affect loudspeaker (electrical) impedance, because boundaries are acoustical impedances which load loudspeaker driver, in similar way as loudspeaker box. Basically, an ordinary loudspeaker driver has bidirectional behavior (like electrical transformer), so we can get information about acoustical impedance (of box, room boundaries, etc) through measured electrical impedance.
Also, in any normal use of loudspeakers, non-resonant boundary interferences (SBIR), can't significantly affect loudspeaker driver cone movement (and stop it), because sound pressure level decrease with increasing distance from the source, so we can easily get 50dB of SPL drop between SPL of direct and reflected wave at cone position. So resulted measured SPL will be clean "nearfield loudspeaker SPL measurement" which is widely known and used because its immunity from boundary interferences.
So we can measure SPL at 5mm from cone of bass driver, and we may be pretty sure that nothing from room boundary effects can't affect our measurements significantly.

At the end, it's possibly easier for someone to try this, but try to keep away your measurement microphones from high SPL, because at 5mm from dust cap, loudspeaker driver can be pretty LOUD.

Quote:
Originally Posted by Ethan Winer View Post
.........This is also why the SBIR peak and null frequencies are related to the spacing between the rear of the speaker cabinet and the boundary, rather than from the cone in the front of the box. The rear of the box is actually the sound source location.
....
Rear of the box can't be sound source location because ludspeaker cabinet (box) is much heavier than driver cone (5-25kg vs. 15-150g for ordinary two way loudspaker), so radiation (air movement) which originate from the box may be easily neglected, even if box surface is greater than loudspeaker cone surface.

To be clear, radiation from loudspeaker box walls exists, but in this particular case it is absolutely negligible (if we talk about room boundary interferences).
Old 15th January 2012
  #248
Gear Guru
7

Quote:
Dan, again; there are 7 reflection paths from a single isolated corner (even if in free space), not only three, and these will have different arrival times relative than the first.
Sorry Jens, I simply don't get this. Let us simplify further just to see if we can get on the same page. Lets hypothesise a totally unreal situation.
Three solid boundaries in free space ( with air though!) A tiny LF source, say equally 1 Metre from all three.
We are looking for cancellation at source, so the mic is coincident or very nearly so with the speaker.

At the mic (or speaker) I can visualise only 3 reflections, but I have been Geometrically blind in the past, please do explain your other four.
Without changing the mic/listener position.

To the frequency issue. We typically see SBIR from the Front Wall and another distinct one from the Side Wall. Two notches. They do not combine to form another frequency.
So how would three (or 4, or 7) SBIR paths combine to form a single null?

Phasors set to Stun? ;-)



DD
Old 15th January 2012
  #249
Lives for gear
 
akebrake's Avatar
 

Quote:
Originally Posted by Ethan Winer View Post
Sure, but that's for listeners in front of the speaker. The "lateness" is also due to inherent delays added by the crossover. But I'm pretty sure that SBIR from a boundary behind the speaker is measured from the edge of the box.

I sure wish someone would measure this already. heh --Ethan
Did I miss some thing? I didn't check for the listening position. Where was it? Still confused .

What counts is when the reflected sound hits the boundaries. If sound is "late" the physical measure (feet /metres) from the cabinet front doesn't not correlate with the acoustical path length for low frequencies.

RE the rear wall of the speaker cabinet.
If this is a closed box and the rear wall of the cabinet is vibrating (emitting sound) you'd better buy a real loudspeaker cabinet. heh
Old 15th January 2012
  #250
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Sorry Jens, I simply don't get this. Let us simplify further just to see if we can get on the same page. Lets hypothesise a totally unreal situation.
Three solid boundaries in free space ( with air though!) A tiny LF source, say equally 1 Metre from all three.
We are looking for cancellation at source, so the mic is coincident or very nearly so with the speaker.

At the mic (or speaker) I can visualise only 3 reflections, but I have been Geometrically blind in the past, please do explain your other four.
Without changing the mic/listener position.

To the frequency issue. We typically see SBIR from the Front Wall and another distinct one from the Side Wall. Two notches. They do not combine to form another frequency.
So how would three (or 4, or 7) SBIR paths combine to form a single null?

Phasors set to Stun? ;-)



DD
SBIR-ism.gif




And if only taking the first (arriving) three sources into consideration, the fc of cancellation would be about 86 Hz (assuming 1 meter to each boundary), right?

Now, below you see the output of Room Optimizer (from RPG) for the source and receiver in the corner (1 meter) in a very large room, the biggest size the program will let you enter: 49 x 49 x 49 meters, so reflections from other boundaries will have negligible addition at the receiver position in relation to the reflections (a total of 7) from the nearby boundaries:

SBIR-ro-sbir.gif
Note that the centre frequency of the dip is not at 86 Hz as I’ve tried to explain. The addition of all the 7 sources is needed to find the centre frequency of cancellation.


If placing the source and receiver in the middle of this huge room and 1 meter above the floor, then the fc will be 86 Hz since there will only be a single first order reflection with enough gain to add to the direct sound. If the receiver point is moved only in the Z –plane (height), this frequency of cancellation will stay unaffected but as soon as you move to the side as well, it will change, even if only by tiny amounts depending on position:

SBIR-ro-sbir-1-wall.gif


Did I make myself clear enough (not being sarcastic now, I’m actually trying to explain this stuff)?


/Jens
Old 15th January 2012
  #251
Gear Guru
Comprendo

Thanks for the effort Jens.
However, we don't know how the Optimiser program is doing this.
Does it allow the reciever position to be coincident with the speaker? Why would it?
It may be being simply pragmatic and blurring lots of close nulls together. Who knows.
The Wall Bounce Calculator seems to do the same. Smoothing.
However I remain convinced that each reflection gets it's own distinct null.
In extremis, 7 non equidistant images, 7 nulls.
We see this all the time. As in my last post. Front Wall SBIR and Side Wall SBLIR, two distinct nulls, unless the geometry coincides.

Consider this for a moment. A very light coned omni speaker driven by a relatively high impedance amplifier.
Take note of Boggy's posts.
Place it equidistant from three boundaries. Now can you imagine restricted cone movement?

DD
Old 15th January 2012
  #252
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Thanks for the effort Jens.
However, we don't know how the Optimiser program is doing this.
I have posted a link (in previous post in this thread!) to the PDF outlining exactly how the program works ... !

So, my efforts will never be enough for you I guess ... I think I’ll refrain to waste more of my energy on it!
Old 15th January 2012
  #253
Gear Guru
Waste

You efforts are appreciated Jens.

However I think we are constantly talking at cross purposes.
Albeit hypothetical and probably of no real use, as my test will probably show. For the moment my, Ethan, and Rod's, Classic SBIR has the transmitter coincident with the receiver.
I am testing as to whether energy is robbed FROM the source AT the source.
This would be reflected by a change of impedance in the speaker/crossover at the null frequency. As boggy has alluded to the speaker gets 'loaded' by the room.

Unless I am in error here, I think everyone else here understands that proposition?

I doubt that Room Opt allows the Listener and Speaker positions to be coincident.

I cannot figure how two distinct nulls caused by two distinct reflections from two distinct boundaries at distinctly different distances could possibly sum to form a single null at a new frequency.

Again, anyone else? Do the reflections from different distances combine to form a single null or not?


DD

Last edited by DanDan; 15th March 2012 at 03:13 PM.. Reason: REVISED
Old 16th January 2012
  #254
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Thanks for the effort Jens.
However, we don't know how the Optimiser program is doing this.
Does it allow the reciever position to be coincident with the speaker? Why would it?
It may be being simply pragmatic and blurring lots of close nulls together. Who knows.
The Wall Bounce Calculator seems to do the same. Smoothing.
However I remain convinced that each reflection gets it's own distinct null.
In extremis, 7 non equidistant images, 7 nulls.
We see this all the time. As in my last post. Front Wall SBIR and Side Wall SBLIR, two distinct nulls, unless the geometry coincides.

Consider this for a moment. A very light coned omni speaker driven by a relatively high impedance amplifier.
Take note of Boggy's posts.
Place it equidistant from three boundaries. Now can you imagine restricted cone movement?

DD
Quote:
Originally Posted by DanDan View Post
You efforts are appreciated Jens.

However I think we are constantly talking at cross purposes.
Albeit hypothetical and probably of no real use, as my test will probably show. For the moment my SBIR has the transmitter coincident with the receiver.
I am testing as to whether energy is robbed FROM the source AT the source.
This would be reflected by a change of impedance in the speaker/crossover at the null frequency. As boggy has alluded to the speaker gets 'loaded' by the room.

Unless I am in error here, I think everyone else here understands that proposition?

I doubt that Room Opt allows the Listener and Speaker positions to be coincident.

I cannot figure how two distinct nulls caused by two distinct reflections from two distinct boundaries at distinctly different distances could possibly sum to form a single null at a new frequency.

Again, anyone else? Do the reflections from different distances combine to form a single null or not?


DD

So … Don’t ask me why, but I’ve made a simple Excel version of the Room Optimiser software and with a resolution of 255 points for the frequency response graph … it should be enough or what do you say Dan?

I’ve also made it possible to “gate” the response in order limit the reflections taken into consideration. If no limit is set, the calculated response is made up of all image sources ranging from -5, -5, -5 to +5, +5, +5: a total of 1330 reflections.


So, let’s compare it to the original program (Room Optimiser) shall we; all parameters in terms of position is exactly the same:

Room: 49 x 49 x 49 meters.
Source: X = 1, Y = 1, Z = 1 (meters)
Receiver: X = 1,01, Y = 1,01, Z = 1,01 (meters, one cannot set the source and receiver in the exact same position).

Here’s the output from Room Optimizer:
SBIR-ro-sbir-modal.gif

And here’s the output from my quickly put together excel sheet if using the maximum 1330 reflections:
SBIR-jens-sbir-1330.gif

And if only the first 7 reflections (or “all” if only a single corner in free space) arriving within 10,1 ms after direct sound:
SBIR-jens-sbir-7.gif

And if excluding all but the first three (arriving within 5,8 ms):
SBIR-jens-sbir-3.gif


Now Dan (and Rod); do you see what I’m talking about now?
Old 16th January 2012
  #255
Gear Guru
Impressive

Thank you Jens. Again, I really do appreciate your efforts, and obvious passion much more so. If the extra four reflections do indeed change the frequency, as your Math have predicted, I will be humbled and pleased to be 'wrong' and to to have learned something new to me. You may indeed have proven that SBIR, as in the pedantic notion that the source level is simply altered by the nearest boundaries, in reality doesn't exist at all!
i.e. That there is no stable single SBIR frequency possible, certainly over a wide area of the room. Hopefully my test will shine a light on that.
Although I suspect .... well let's see.....

I am curious as to the very broad bandwidth of the dip. I thought real world SBIR notches were very very narrow. Three coincident ones being narrow but much deeper.
I suppose my test will shine a light on that at least.

I am quite open to the notion that strong influences may change a frequency when the resultant nulls are very close. Strong effects such as Modes etc. can 'suck in' nearby anomalies. Resonant frequencies can certainly change with the addition of treatments. e.g. A tuned trap can alter a modal frequency.
However, I do believe level and closeness in frequency matter big time here.
May I ask, did your calculations take into account the LEVEL of the reflections. i.e. What were the levels of the 7 reflections?.

Very very broadly speaking, lets say a doubling of distance in a reflective room does say -3dB per doubling of distance.
So in the case of my 'clean' room, lets make a wild guess and say the extra four reflections are say -8, to -12 dB down.
Are those levels or something similar in your algo?

Boggy has pointed at this level vs distance to say that there is no way the boundary effects could be measured at the cone.
This makes sense to me. Maybe there is no Speaker Boundary Interaction. Maybe there is only SLBIR?
DD

Last edited by DanDan; 16th January 2012 at 11:16 AM.. Reason: Broad
Old 16th January 2012
  #256
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Thank you Jens. Again, I really do appreciate your efforts, and obvious passion much more so. If the extra four reflections do indeed change the frequency, as your Math have predicted, I will be humbled and pleased to be 'wrong' and to to have learned something new to me. You may indeed have proven that SBIR, as in the pedantic notion that the source level is simply altered by the nearest boundaries, in reality doesn't exist at all!
i.e. That there is no stable single SBIR frequency possible, certainly over a wide area of the room. Hopefully my test will shine a light on that.
Although I suspect .... well let's see.....
May I ask, did your calculations take into account the LEVEL of the reflections. I am quite open to the notion that strong influences may change a frequency when the resultant nulls are very close. Strong effects such as Modes etc. can 'suck in' nearby anomalies. Resonant frequencies can certainly change with the addition of treatments.
A tuned trap can alter a modal frequency.
However, I do believe level and closeness in frequency matter here.

Very very broadly speaking, lets say a doubling of distance in a reflective room does say -3dB per doubling of distance.
So in the case of my 'clean' room, lets make a wild guess and say the extra four reflections are say -8, to -12 dB down.
Are those levels or something similar in your algo?

Boggy has pointed at this level vs distance to say that there is no way the boundary effects could be measured at the cone.
This makes sense to me. Maybe there is no Speaker Boundary Interaction. Maybe there is only SLBIR?
DD
The calculation assumes perfectly rigid boundaries causing no additional energy loss or phase change opun reflection. The only thing affecting the gain of the reflections is the distance travelled. I think the extreme similarity of the original program and my excel sheet proves that there's nothing wrong in my calculations. I just wanted to show you the result of phasor addition using a high resolution frequency graph in an tempt to illustrate that what you thought was an effect of "smoothing" is simply the result of adding phasors. If I still haven't succeeded, I don't know how to do it. Please understand that there naturally will be situations where you end up with many different cancellation frequencies but for the example that we have discussed, the fc of the dip simply moves as you add more reflections on top of the first three.

I once again feel the need to remind anyone trying to measure the effects of SBIR that the boundary condition in a real room seldom is as rigid as assumed when talking theoretically about these things. Also, there's the problem of analysing the effects of a limited amount of reflections at low frequencies since the gate times needs to be quite long in order to calculate the response at low frequencies. If you have access to a big hall/warehouse with thick and solid stone/concrete walls, perhaps you can do something useful but why assume that it hasn't been done before?
Old 16th January 2012
  #257
DAH
Lives for gear
 
DAH's Avatar
damn. stop assumptions and TEST it!
Old 16th January 2012
  #258
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DAH View Post
damn. stop assumptions and TEST it!
Do you seriously believe that these things have not been tested and are not fully understood already? We are talking about sound here, not the Higgs boson or M-theory …

Also, again:
https://www.gearslutz.com/board/7436973-post214.html
Old 16th January 2012
  #259
DAH
Lives for gear
 
DAH's Avatar
Quote:
Originally Posted by Jens Eklund View Post
Do you seriously believe that these things have not been tested and are not fully understood already? We are talking about sound here, not the Higgs boson or M-theory …
I do seriously believe that, if this discussion is only about the necessity of the term diferentiation between LBIR and SBIR and not the testing (or providing data of such test done in the past) that SBIR effect remains at a constant frequency independently of the receiver position, then the discussion is highly senseless.
Old 16th January 2012
  #260
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DAH View Post
I do seriously believe that, if this discussion is only about the necessity of the term diferentiation between LBIR and SBIR and not the testing (or providing data of such test done in the past) that SBIR effect remains at a constant frequency independently of the receiver position, then the discussion is highly senseless.

ok ...

So RPGs software is wrong, my excel sheet is wrong and the Java app from Falstad / Harvard (post 220) is wrong …
Old 16th January 2012
  #261
DAH
Lives for gear
 
DAH's Avatar
Quote:
Originally Posted by Jens Eklund View Post
ok ...

So RPGs software is wrong, my excel sheet is wrong and the Java app from Falstad / Harvard (post 220) is wrong …
Its not about the Java applests, spreadsheets and RPG.
"We are talking about sound here" (c) Jens Eklund
If the null frequency remains independently of the receiver position, then your backpressure cancellation hypothesis is wrong, if it does not exist when LBIR is eliminated (measurement done at the cone) and moves when the receiver`s position is not ath the speaker`s cone and changed, then you are right.
The only necessity I find for the test is to eliminate the modal response (hence the room itself).
Old 16th January 2012
  #262
Gear Maniac
 
cborg's Avatar
 

Quote:
Originally Posted by Jens Eklund View Post
What do you mean?
I mean that I've provided one situation in several posts. In post 218 I've tried to clarify it. Pictures and all.
And a simple geometric path calculation won't provide the answer for what's being measured.
As simple as that. I thought I was being very clear by now.
Old 16th January 2012
  #263
Lives for gear
 
boggy's Avatar
Quote:
Originally Posted by Ethan Winer View Post
Sure, but that's for listeners in front of the speaker. The "lateness" is also due to inherent delays added by the crossover. .......
Crossover filter is not needed for loudspeaker's low frequency delay. Any construction with high pass filter behavior (boxed loudspeaker driver without crossover is one of it) introduce delay at low frequencies in area around and below cutoff frequency, as is displayed on attached graph.
Attached Thumbnails
SBIR-graph01.gif  
Old 16th January 2012
  #264
Gear Guru
Despair

Don't despair Jens! I love to see Mathematics worked well in the hands of a good practitioner.
And again, I do understand what you are saying and have done.
However your view, seemingly based on arithmetical predictions, doesn't concur with mine which derives from first principles.
The calculations regard a thin plasterboard ceiling, 3 times the distance away, as an equally strong source of reflection as a foot thick concrete wall or a concrete floor on foundations and earth.
At 90Hz, not realistic.
Furthermore, why is the bandwidth so wide, the best part of an octave, with three coincident nulls?
We have seen the single floor reflection null do -15 or even -20dB without smoothing, diminishing to -2dB when viewed third octave.

DAH, the test will happen, but I am fairly sure we well get an ambivalent answer. i.e. the frequency will remain fairly constant throughout the room, but not perfectly so.

All please remember, my main quest is the Front Wall Back Wall issue. Why would the one further away have more effect, as Ethan's tests seem to show?

DD
Old 16th January 2012
  #265
Lives for gear
 
akebrake's Avatar
 

Quote:
Originally Posted by Jeffrey Hedback View Post
What you are seeing, using ARTA for post processing, is the same Right Spkr Impulse Responses Over-Layed onto themselves of two rooms. The short time gate (usually 45ms-60ms depending on room size) gives enough time for the boundaries to impact the measurement at low frequencies. The longer time gate is representative of what's heard in the room.

Image one is pre-data of a new client's small listening room. The short time gate in this case is 50ms as the room is small (~10'x13'x8'). The speaker is nearly 2' off the front wall (corresponding to null at 150Hz). The mic was 3.5' off the rear wall (which corresponds to 80Hz null in short gate). Comparing the short time gate to the long time gate, you can immediately see how the room resonances "over-power" the immediate boundary interference effects.

The second graph is a treated room where we are working on final tweaks to speaker/mix positions (before integrating a pair of Rythmik Audio subs to the JBL LSR 6332's which have F3 of 54Hz). You can immediately note how the short time gate (60ms in this case) and the long time gate (300ms) are nearly identical. This indicates essentially no modal resonance. So the response is all boundary interference and the actual speaker response.

I've found this approach very helpful to know what battle to fight.
This is most interesting! Thanks for shearing, Jeff!

As I understand it from your "Time Gated Overlay Studies" above

1. "SBIR" gating time is adjusted depending of room size
2. The modes are important to take care of.
2. Then, fine tune the speaker & listener position.

Some further questions:

Are room number two much larger than room one?
Are their monitors flush?
What frequency resolution was used in the measurements?

And finally: Is the gating time taken from "start", e.g.0ms-50ms?

Cheers
Old 16th January 2012
  #266
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Don't despair Jens! I love to see Mathematics worked well in the hands of a good practitioner.
And again, I do understand what you are saying and have done.
However your view, seemingly based on arithmetical predictions, doesn't concur with mine which derives from first principles.
The calculations regard a thin plasterboard ceiling, 3 times the distance away, as an equally strong source of reflection as a foot thick concrete wall or a concrete floor on foundations and earth.
At 90Hz, not realistic.
Furthermore, why is the bandwidth so wide, the best part of an octave, with three coincident nulls?
We have seen the single floor reflection null do -15 or even -20dB without smoothing, diminishing to -2dB when viewed third octave.
If you think about the setup, the explanation is trivial: The receiver is located extremely close to the source and the reflection (or image source) is very far away (in relative terms) so the contribution of it is extremely small.


If we where to move the receiver away from the source and thus decreasing the difference in gain between the direct sound and the reflection(s), the normal shape you are familiar with is appearing:

SBIR-sbir-notch.gif

Room: 1000 x 1000 x 1000 meters
Source: X = 500, Y = 500 and Z = 1 (meters).
Receiver: X = 500, Y = 500 and Z = 500 (meters).
Number of reflections included in prediction: 1


Understanding the relation of relative levels of direct sound and reflections, and their arrival time relative to direct sound is key.
Old 16th January 2012
  #267
Gear Guru
Good

Meant to say it earlier, great post Jeff.

Jens, thank you.
Does either Room Optimiser or your .xls have a possibility of entering levels, reflectivities, or absorption coefficients, for the boundaries?

Even if not, the spreadsheet you have written seems to be a development on Thomas Barefoots 2D Wall Bounce Calculator.
You could probably even sell it!, but if not you might consider putting it on your website.

DD

Last edited by DanDan; 16th January 2012 at 08:02 PM.. Reason: Found the website
Old 16th January 2012
  #268
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by cborg View Post
I mean that I've provided one situation in several posts. In post 218 I've tried to clarify it. Pictures and all.
And a simple geometric path calculation won't provide the answer for what's being measured.
As simple as that. I thought I was being very clear by now.
Can you please explain to me why you feel that the illustrations in post 220 (now 218) is incorrect or not valid?

The image source model is the most commonly used method to calculate the effects of early reflections but also modal behaviour. The results generated by the standard frequency based modal calculation has often been compared to that of the image source method and assuming enough images is taken into the calculation, the results are very similar. If you have doubts about the result of the ISM, you also question the result of frequency based modal calculation ... do you? If you feel there’s a better method to simulate these things, I would love to hear about it so that I can add it to my spread sheet …

If the results from ISM is incorrect, how do you justify the existents of software using it?
Old 16th January 2012
  #269
Lives for gear
 
Jens Eklund's Avatar
Quote:
Originally Posted by DanDan View Post
Meant to say it earlier, great post Jeff.

Jens, thank you.
Does either Room Optimiser or your .xls have a possibility of entering levels, reflectivities, or absorption coefficients, for the boundaries?

Even if not, the spreadsheet you have written seems to be a development on Thomas Barefoots 2D Wall Bounce Calculator.
You could probably even sell it!, but if not you might consider putting it on your website.

DD
We’ll se what the future brings … It’s always a matter of time. There are so many things I want to do! I did however make room for the things you mentioned in the sheet so it could easily be implemented if I can find the time. Perhaps an app for Android …
Old 16th January 2012
  #270
Gear Maniac
 
cborg's Avatar
 

Quote:
Originally Posted by Jens Eklund View Post
Can you please explain to me why you feel that the illustrations in post 220 is incorrect or not valid?

The image source model is the most commonly used method to calculate the effects of early reflections but also modal behaviour. The results generated by the standard frequency based modal calculation has often been compared to that of the image source method and assuming enough images is taken into the calculation, the results are very similar. If you have doubts about the result of the ISM, you also question the result of frequency based modal calculation ... do you? If you feel there’s a better method to simulate these things, I would love to hear about it so that I can add it to my spread sheet …

If the results from ISM is incorrect, how do you justify the existents of software using it?
Once again (and I'm soon done repeating myself), I'm discussing the example in post 218 (and what I posted previously).

Why is it so hard to refer to that?
If you don't understand then please just ask. What is it that you don't agree with in that scenario?
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump