I've been investigating calculated versus measured room modes, and the deeper I look the more suspicious I'm becoming!
In THIS thread, which Eric Desart links to repeatedly when he warns people away from my ModeCalc program, Eric claims that non-axial modes are very important and are critical for a complete mode assessment. I've seen others here argue the same point many times. Eric also discusses the RealTraps lab specifically in THIS thread at StudioTips, and concludes, "Clear is that there ARE tangential modes on those pictures."
But the mode display from Eric's calculator makes no sense given the dimensions of my room! In fairness, neither does the display from my own mode calculator. This is why I suspect that both of our mode calculators are missing something fundamental.
Since the very first mode that was actually measured by ETF does not align with either of our calculators, I'm suspicious of Eric's other interpretations, such as which peaks are non-axial. Here's the graph from Eric's mode calculator for the RealTraps lab:
And here's an ETF display of the same room with no bass traps (or anything other than the speakers and desk) when measured at the mix position:
According to ETF, the lowest frequency peak is at 41 Hz. Nothing even close to that is displayed by either of our calculators. I am absolutely certain of the room dimensions because I've measured many times. It is exactly 16'2" long, by 11'6.5" inches wide, by 8' high. There are no irregularities, all surfaces are sheet rock except for the cement floor, and the two doors are very nearly flush with the walls.
Since the very first mode is off by 20 percent (!), how can any of the subsequent frequencies be trusted? Further, the second mode displayed is way off too! Both of our calculators report the second mode (the lowest one for the width) at around 49 Hz. Yet ETF shows the second mode at either 54 Hz or 57 Hz, depending on which of the many different locations I measured at. Eric rightly questioned the results from his own calculator when he wrote this at StudioTips:
Quote:
The 39.4 Hz on the referred page noted as 40 Hz is a question mark. Logically it should be the 1,0,0 mode, but it lies between the 1,0,0 and 0,1,0 mode ... But why then does the 40 Hz is no 1-2 relation to the 70 Hz?
Even more relevant, Eric claims the mode at 157 Hz is tangential. But look at this ETF graph showing the decay time with the room empty:
We all agree the tangential modes should be less severe than axials, yet the mode shown here at 157 Hz is as strong as any of the others, and decays for nearly as long as any. I simply cannot accept that this very dominant mode is non-axial without a more compelling explanation.
Finally, I'm not the only person to observe calculated modes disagreeing with measured results. This was posted earlier today in another forum I frequent:
Quote:
In my room, for instance, a room mode calculator (not Ethan's) comes up with 24.5 Hz as my first axial mode, but a measurement with ETF comes up with about 30 Hz.
Notice that in this fellow's room the mode as measured is also about 20 percent too high.
Eric, have you, or anyone else here, ever actually measured the response in a room and compared what was measured to what a mode calculator program predicts? I really doubt ETF is broken because I've compared against sine waves whose frequencies are known. Here's that proof:
So unless someone can point out an obvious error, I'd say Eric and I both have some explaining to do.
Are you suggesting that Lord Rayleigh's differential equation solution is wrong?
I recently did my room and there was a very high degree of correlation between the predicted and measured values with all of the major offenders being axial.
eric_desart Recording Org Pro Audio Forums
Joined: May 23, 2003
Posts: 816
Location: Belgium Antwerp
This shift in the low frequency mode is described in the thread and in the Hartman papers linked in that thread.
And by definition room modes are based on a rigid reflective room.
Just read the thread.
Ethan, I measure for 25 years. You have ETF as your first measurement tool for how long? Until 3/4 weeks ago you even hadn't idea about the relative y-axis scaling and the influence on the interpretation of your own measurements.
What Ethan tries here is provoking me, but he's that incredibly sly that hardly anyone will notice what I'm talking about.
But if I should become angry Ethan gladly will play the mature innocent victim again.
I for one definitely don't need convincing that the equation for mode calculation is correct; I have studied some advanced calculus and the solution is correct without a shadow of a doubt. What I thought was going on here was some kind of modal shifting, but I don't know anything about the science behind this phenomenon; rather, I have just heard people talking about it. Do you have any links for more info on how modes can be shifted from their predicted values? Do you think that this is what's causing Ethan's 20% error?
p.s. - please... just a polite reminder to everyone to keep cool and to use good science
i've little to inject, just a couple of idle thoughts:
the speed of sound is not constant, and can vary a bit with atmospheric conditions (temperature, to a lesser extent humidity, altitude)
absorbing materials have slower speeds of sound, so that might cause some changes too (though i don't konw how much)
Dennis Erskine made a comment about a room mode appearing in a room with poor TL walls that corresponded to a concrete wall behind one of the poor TL walls once.
2 neighboring modes might "add up" to create a solitary dip that appears to be something else entirely. Same thing is easily observed with mechanical resonance on panels of something or other. move an accelerometer and the impact or stimulation point around, and you can get various shifting behaviors, especially on highly damped structures. might look like one extremely well damped mode spanning 400-800hz in one spot, and 3 modes if measured in another spot...
it's worth mentioning that along time ago Philip de Haan noted a similar phenomenon when measuring a car with MLSSA
_________________ All posted information copyright Brian Ravnaas
Technical Director, Audio Alloy
z60611 Recording Org Pro Audio Forums
Joined: Dec 29, 2003
Posts: 820
Location: Ontario, Canada
It's not just the softness of walls (absorbtion within the wall + transmission through the wall) that can change the modes from the predicted.
Adding absorbtion, and objects, inside the room has been reported to move the modes too.
The lower the frequency, the more likely that the acoustical dimensions of the room will be different from the physical dimensions.
The more rigid the walls, the more energy is reflected back into the room, and the more likely that the acoustical dimensions will match the physical dimensions.
Nevertheless the physical dimensions will always contribute, but they needn't always be domanant.
Not all predicted modes are significant nor problematical.
I would guess, that for the kinds of soundproof walls we're supposed to build, that the predicted would be relatively close to the observed. Otherwise the experts wouldn't devote pages and pages to the topic in their books.
And again, that for the kinds of soundproof walls we're supposed to build, that tangental and oblique modes would be significant. To find all modes measurements from multiple locations are required.
If you have a built room, it's always best to measure.
(so says a fellow who has yet to build a soundproof room)
bpape Recording Org Pro Audio Forums
Joined: Feb 21, 2005
Posts: 23
Location: St. Louis (Wildwood), MO
I agree totally with the comment about the wall flexure. This is one of the reasons I don't recommend to people to go after flexible walls. While they can certainly offer some advantages in the way of bass absobtion, their predictability becomes very difficult.
A perfectly rigid wall (which the calculations are based on by the way along with a perfectly rectangular room) should be closer to measuring what the calculations predict than a flexible wall will be.
I also read Dennis' comments about finding a peak corresponding to an outer wall in a dual wall arrangement. In this case, I don't know the construction of the room in question nor would I expect the shift to be HIGHER in frequency. Logically, if the outer wall were the culprit, the shift would be LOWER due to the larger dimension.
1. It is my understanding that the Blackman-Harris FFT window is the least accurate with regards to frequency. A rectangular window will provide the best frequency accuracy, but the amplitude accuracy will suffer. Use of a Hanning or Hamming window is usually the best "compromise."
2. As Brian noted, the speed of sound affects the modes. I typically find that 1130 ft/s is a good value, but I have seen cases where this can vary by ±10 ft/s depending on temp, etc.
3. Any absorption in the room can "move" the modes. Remember that the solution to the Rayleigh equation we're all so fond of using completely disregards damping (i.e., absorption). No damping in the prediction means that the prediction almost never agrees 100% with the measurement.
4. The ~41 Hz peak in the specific measurement shown above cannot be assumed to be solely the room mode. The loudspeaker may have a resonant peak at that frequency. Boundary loading could boost the low-end in general. And, as far as I can tell, a ~35 Hz modal excitation could very well be included in the peak centered at ~41 Hz in the ETF graph. Knowing how the other predicted modes correspond to the ETF measurement requires more information about placement of mic and loudspeakers.
5. Whether tangential modes are "as strong as" axials is all extremely relative. I don't think the collective "we" have agreed on what Ethan implies we have (at least I don't agree - so that rules out one of the "we" ): A tangential mode can be more influential in some circumstances, less in others, the same in others. This assertion of tangentials (or even obliques) being "not as strong" as axials really needs to stop.
6. Finally, trying to verify the accuracy of any computer prediction with one and only one example is pretty ridiculous. But y'all probably knew that, right?
_________________ ---lovecow---
"You've got to be very careful if you don't know where you're going, because you might not get there." - Yogi Berra
Ethan Winer Respected Past Moderator
Joined: Mar 19, 2001
Posts: 3209
Location: New Milford, CT USA
Those are all great comments, and exactly what I was hoping for.
David:
> I recently did my room and there was a very high degree of correlation between the predicted and measured values <
If you measured the room, do you have something concrete you can post?
> with all of the major offenders being axial <
That makes sense to me.
Eric:
> by definition room modes are based on a rigid reflective room. <
Agreed. In this case the room has standard sheet rock, but what's behind the sheet rock on all sides is very far away. This room is in our factory. On three sides there are large areas on the other side of the wall. On the fourth side, which is in the Length dimension, there's another smallish room on the other side. That room adds another ten feet or so.
> I measure for 25 years. <
So surely you must have some hard data you can show, yes?
> What Ethan tries here is provoking me <
This is a legitimate scientific question. Why would you even think that?
Again, the real question is how can you claim that the apparent mode at 157 Hz is non-axial, when nothing else predicted by your program agrees with what was measured?
Brian:
> along time ago Philip de Haan noted a similar phenomenon when measuring a car with MLSSA <
Thanks, that now makes three examples of measurements not agreeing with what's calculated.
BP:
> Logically, if the outer wall were the culprit, the shift would be LOWER due to the larger dimension. <
Exactly my point. And if there were absorption in the room, which there wasn't, that too would explain a shift downward. Probably nowhere near 20 percent, but some amount of shift.
Jeff:
> It is my understanding that the Blackman-Harris FFT window is the least accurate <
Okay, but could it really be off by 20 percent?
> 1130 ft/s is a good value ... can vary by ±10 ft/s <
Okay, that's almost 1 percent...
> Any absorption in the room can "move" the modes <
Understood, and that's why I made sure to point out that the room was totally empty except for the speakers, desk, and computer.
> The loudspeaker may have a resonant peak at that frequency <
Good point. Though I really doubt it because the decay is reduced when traps are added.
> This assertion of tangentials (or even obliques) being "not as strong" as axials really needs to stop. <
ROF,L.
Jeff, you've been doing this a lot longer than I have, so I'm sure you must have lots of data measured in various rooms. Do you have any waterfall plots (or even just raw response) you can post that show the various modes and how much each contributes? This is the crux of my investigation: 1) To understand why measured modes can vary so much from what's predicted, and 2) to find relevant data proving or disproving the importance of non-axial modes. Not what was read in a book, or what some "expert" claimed in a forum post, or any other anecdotal "evidence."
Please understand that I'm not taking a position either way! I just want to get to the bottom of this using believeable and repeatable measurements and data.
Folks:
What I'm looking for is hard data showing the predicted versus measured response of a "typical size" room. As in, "Here is a room X by Y by X feet in size, and here is a graph of the measured response. The microphone was placed here, proving the measured peaks were not caused by constructive boundary interference." And so forth.
Why don't you ask Terry Montlick?
He measures and measures, ah let me restate, he measures room dimensions with a tape measure, and he measures acoustics of a room, that's better, different homes every day of his life. Terry has the data you seek. And probably a lot more of it than Jeff or Eric.
Auralex help phone desk may have a bunch of data, but the acoustic measurements won't be as good as Terry's.
> It is my understanding that the Blackman-Harris FFT window is the least accurate <
Okay, but could it really be off by 20 percent?
First, I didn't say it could. I ignored the "20%" comment 'cause anytime someone uses percentages in relation to sound, I generally tune them out. The difference you're talking about is about 6 Hz. Sure, that's almost 20% in the range you're talking about. The same difference at 1000 Hz is a negligible 0.6%, though. And the FFT window doesn't care whether it's 10 Hz or 1000 Hz; the shape is the same.
So, could it really be off by that much? I suppose it could.
My main point - a point I've tried to make before - is that the combined effect of all these seemingly minor issues might help explain the observably large variations.
Quote:
> Any absorption in the room can "move" the modes <
Understood, and that's why I made sure to point out that the room was totally empty except for the speakers, desk, and computer.
All of which absorb sound. As do the walls, the ceiling, the floor, etc. All of which will affect the modal response characteristics. Yes, by as much as "20%" if that's how you'd like to look at it.
Quote:
> The loudspeaker may have a resonant peak at that frequency <
Good point. Though I really doubt it because the decay is reduced when traps are added.
You don't know for sure, though. Correct?
Quote:
> This assertion of tangentials (or even obliques) being "not as strong" as axials really needs to stop. <
ROF,L.
Jeff, you've been doing this a lot longer than I have, so I'm sure you must have lots of data measured in various rooms. Do you have any waterfall plots (or even just raw response) you can post that show the various modes and how much each contributes?
I guess I could come up with something, but I don't see the point. There's nothing universal to conclude, so it would be a pretty pointless exercise. What I can or cannot show for one room would hardly be applicable to your room...or vice versa.
Quote:
This is the crux of my investigation: 1) To understand why measured modes can vary so much from what's predicted,
Which is what we're trying to help you with.
Quote:
and 2) to find relevant data proving or disproving the importance of non-axial modes.
Which you have. Just figure out what the modal response should look like based on your room size - assuming it's rectangular - your loudspeaker position(s), and your mic position. Compare the two. See how well they agree. See what modes dominate. See what else might influence things. You might check this thread at Studiotips where Scott talks about observing and measuring a resonance that had nothing to do with any room geometry. Could be that your ETF measurement isn't even measuring a room problem at all. Could be... Might be... Etc...
Quote:
Not what was read in a book, or what some "expert" claimed in a forum post, or any other anecdotal "evidence."
But...isn't that what you're asking for? A "claim" from an "expert" (me) in a "forum" (RO)? ( ... Couldn't resist.)
Quote:
Please understand that I'm not taking a position either way! I just want to get to the bottom of this using believeable and repeatable measurements and data.
Well, I don't know that there's much to "get to the bottom" of. Comparing predicted modal response to measured response is something you can do on your own, or with existing software packages. In some situations, you'll be able to correlate everything. Other situations may present bigger challenges. Welcome to the club.
_________________ ---lovecow---
"You've got to be very careful if you don't know where you're going, because you might not get there." - Yogi Berra
David French Respected Past Moderator
Joined: Jun 19, 2002
Posts: 2852
Location: Indiana
If you measured the room, do you have something concrete you can post?
Ethan, remember when we were emailing back and forth about FRK data and you told me you wouldn't let me see your data because you spent money to obtain it? Becuase of this I now feel reluctant to share my data with you for free. Whatever you learn in this thread will likely be put to commercial ends anyway, so perhaps all of us should think twice before participating.
Terry has the data you seek. And probably a lot more of it than Jeff or Eric.
Auralex help phone desk may have a bunch of data, but the acoustic measurements won't be as good as Terry's.
The part about data quantity is neither here nor there. (How would we know, anyway?)
As for quality, I have every bit of faith that any in-depth small room measurements Auralex has performed are every bit as top-notch as those of Terry Montlick. Who is an exceptional individual, I might add.
But, IMO, it's worth going back to my point that presenting more data may unnecessarily cloud this issue even more. Just a thought.
_________________ ---lovecow---
"You've got to be very careful if you don't know where you're going, because you might not get there." - Yogi Berra
avare Recording Org Pro Audio Forums
Joined: Feb 12, 2004
Posts: 339
Location: Hamilton, Ontario, Canada
What I'm looking for is hard data showing the predicted versus measured response of a "typical size" room. As in, "Here is a room X by Y by X feet in size, and here is a graph of the measured response. The microphone was placed here, proving the measured peaks were not caused by constructive boundary interference." And so forth.
"Acoustics" L. Beranek, 1957. Pretty graphs with measured SPL and calulated room modes.
But then again, I recall citing this same reference to you previously.
At the risk of being hypocritical, I've provided links below to something I put together back in November 2003. Both studies are measured responses compared to the calculated modes and (in the case of the newer link) predicted modal response for a customer's control room. The old PDF has already been linked here (in RO) before, so I figure it's as good as anything to look at. And perhaps familiar? Regardless, please ignore the red arrows in the graph for the first PDF. They were part of a (now infamous) thread that has since been put to rest.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
For more information on advertising, investing ,
merging
or any other ideas you may have for this community" Feedback
Pro Audio forums, Pro Audio Dealer, audio reviews and all the moderating here is
volunteer. Please remember no-one is being paid to be here or deliver
hot coffee. Play Fair, be polite, patient and considerate to others. Title your
topics properly and do not slander anyone, ever online. Read this before your post here:Recording
Org Disclaimer
This site can be translated into 13 languages. 錄音工作室幫助下，新聞和信息，數位專業音頻論壇, Opname studio helpen, nieuws en informatie, digitale pro audio forums, Studio d'enregistrement à l'aide de nouvelles et d'information, forums de l'audio numérique pro, Tonstudio helfen, Nachrichten und Informationen, digitale Pro-Audio-Foren, Estudio de grabación ayuda, información y noticias, foros de audio digital profesional. help, pro tools, cubase, nuendo, DAW, AD/DA, microphone, preamp, compressor, equalizer, Music Education, Arranging, Composing, collaboration des musiciens,
professionelle Musiker und Ingenieure,
colaboración de músicos profesionales y los ingenieros lo que pensamos acerca de una banda
Our new beta site is now online for testing The New Recording.org CMS
Also or Non Respected Past RO Acoustics & Design Moderator? 
): A tangential mode can be more influential in some circumstances, less in others, the same in others. This assertion of tangentials (or even obliques) being "not as strong" as axials really needs to stop. 
