 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Natural Science Forum / Physics / Acoustics / February 2005Reverb room versus ETF
Folks, I often need to assess the effectiveness of different absorbers at low frequencies. Above 100 Hz absorption can be measured reliably in a lab's reverb room, but I'm much more interested in testing at frequencies below 100 Hz. I don't need absolute absorption in Sabins, but I do need to know which material or bass trap design works better than another. Recently I've been using the ETF software in a normal size room (about 16 by 11-1/2 by 8 feet). So far the results seem much more useful, and certainly more repeatable, than a reverb room. Therefore, the purpose of this post is to get the opinions of the experts in this group. I'll explain briefly what I see as the pros and cons of using ETF versus a reverb room in a lab. Comments from all will be most welcome. A standard reverb room test requires a room large enough to develop a reverberant field to the lowest frequency you hope to measure. Most US labs are large enough to test 100 Hz and higher. So in theory you could test at any frequency and any bandwidth from 100 Hz on up, and be able to measure the change in decay time with reasonable accuracy. But under 100 Hz a reverb room is dominated by "pockets" of energy. It seems to me that even with a lab's moving microphone and multiple tests averaged together, the results still vary too much to be useful. Indeed, I've seen *negative* Sabin values at low frequencies in IBM's lab I use, even when testing large sample sizes. This waterfall plot shows a recent test I did using ETF with my test room completely empty: www.ethanwiner.com/misc-content/lab-ringing-empty.gif Unlike the coarse 1/3 octave results I get from IBM for their reverb room tests, ETF can resolve decay bandwidth to finer than 1 Hz. Looking at the graph linked above you can clearly identify each room mode and see it's individual decay time. The downside is the only frequencies that can be tested in this particular room are those that resonate. So while I can measure the change in decay at 42 Hz and 70 Hz and 96 Hz perfectly, I can't test at 60 Hz or other in-between frequencies. To do that I'd need to find another room. The lack of controlled temperature and humidity seems unimportant to me. Those won't change much between tests, they're less significant at low frequencies, and I'm not looking for certified Sabins values anyway. It seems to me that for a given room's mode frequencies, ETF gives a much more accurate reading of decay time than a reverb room below 100 Hz. And possibly better in the bass range above 100 Hz too. I'll also mention that I recently tested a large number of different materials using ETF this way, and the results were not only repeatable but also gave me the numbers I expected. As opposed to very similar tests I ran at IBM last year which yielded vague and inconclusive results. I see another big advantage to ETF, at least for testing the effectiveness of different bass absorbers. Besides showing exactly how each mode's decay time is reduced, ETF also shows how the absorption lowers the Q of those resonances. In my experience, this feature of bass trapping is at least as important as flattening the LF response and reducing modal ringing time. With a lower Q the peaks and nulls are less intrusive, so instead of individual bass notes sticking out like a sore thumb, a much broader range is emphasized. This graph shows the same room after adding a bunch of bass traps: www.ethanwiner.com/misc-content/lab-ringing-trapped.gif The improvement of a lower Q is quite obvious at all frequencies, but especially above 100 Hz. So what do the experts think? Do you agree that ETF (or an equivalent system) can give more reliable results than a reverb room, at least for those low frequencies available in a given test room? Thanks *very much* in advance for your insight. --Ethan > So what do the experts think? Do you agree that ETF (or an equivalent > system) can give more reliable results than a reverb room, at least for > those low frequencies available in a given test room? There's no question that you have depicted is a simple graphic manner more clear than ever (though still not totally comprehensive, if that ever can be!) the regime of low frequency response of a room. We all have been aware that frequency sweeps in a room produce marked data; the question has been; "How do we interpret such 'data'. The subjective syndrome ever-present in practical acoustics is the unction to produce a single number rating for everything, viz. The NRC of absorbers. And that is where the tort occurs, as it may be clear here.... "OK, Ethan, now tell us your result by giving us but one number! Was it a 10? Was it a 30? A .5? What?!!!" Put another way, I believe you should continue to evolve this multidimensional approach to low frequency room analysis. Apparently, you depict the sound pressure response at but one location, perhaps at a key listener location. The next natural step is to depict it at a second critical location, say the speaker location to predict how well a room accepts sound from it. Another extension is to go clear down to lo-lo frequency (below the first room mode). Obviously, zero Hz is not needed, or is it??? I can see value in knowing it, as it indicates the room's capability to "breathe" as in response to a sub woofer. Keep it up. I think it's high time we achieve quantification of this long-standing ETF curiosity. my 2 cents. Angelo Campanella  Signature--------- www.CampanellaAcoustics.com --------- "I have simply studied carefully whatever I've undertaken, and tried to hold a reserve that would carry me through." - Charles A. Lindbergh. "As for background noise level; 35 dBA is a good classroom; 45 dBA is a sound masking system!" - Anthony K. Hoover "Every day, we perform on the stage that we set yesterday." AJC. Angelo, Thanks very much for your comments. > "OK, Ethan, now tell us your result by giving us but one number! Was it a 10? Was it a 30? A .5? What?!!!" < Right. The point is to be able to do relative comparisons mainly. > Apparently, you depict the sound pressure response at but one location, perhaps at a key listener location. < Yes, as long as the microphone doesn't move between tests the comparisons should be valid. > Keep it up. I think it's high time we achieve quantification of this long-standing ETF curiosity. < Thanks for the encouragement! --Ethan Ethan, You wrote: > I often need to assess the effectiveness of different absorbers at low frequencies.< Are you trying to assess for a single room, or are you trying to assess the generic behavior of a device or devices below 100 Hz? For the former, there is nothing seriously flawed with using any sort of "off-the-shelf" analysis package. For the latter - assessing device behavior/characteristics/properties, and presumably quantifying same - I hope the reasons why it will be difficult are obvious. You pointed out yourself that you cannot reliably measure decay at all frequencies in a single small room. I would add to that the fact that the change in decay at the modal frequencies that you were able to measure could (and probably will) vary significantly depending on the placement of the devices under study. In other words, you'll get different results when you put them over a corner versus standing them up near a wall versus a mix of the two. This immediately raises the question of which set or sets of data are "correct"? My guess is none and all. :-) I think the research has merit. No question. However, you must give serious thought to what you will or won't be able to deduce from the research. It's very likely that - as with anything - you will come away with 10 questions to answer the 1 you started with!!! You can evaluate the behavior of a device, but you cannot really draw conclusions about the absolute behavior of a given device based on the approach you are using. It's one thing to say: "When device A is used in location P of room G, the change below 100 Hz was X." That sort of statement has merit and could prove to be quite useful. But it's a huge leap of faith to look at the same test results and say: "When device A is used in any room, the change below 100 Hz will be X, as shown by tests conducted in room G." It may be fair to assume some universality of a device's performance. But probably only after performing several rounds of tests in different rooms, with different placements of the devices, etc., etc. You should also consider loudspeaker positioning (as Ang mentioned), mic positioning (which it seems like you are), and loudspeaker performance in the range of consideration (very often all over the map). Good luck! Best regards, Jeff D. Szymanski Chief Acoustical Engineer Auralex Acoustics, Inc. Hi Jeff, As always, thanks for your valuable comments. You've probably seen the results of my first round of tests, but for anyone else interested the report is here: www.ethanwiner.com/density/density.html These tests assess how the low frequency absorption of rigid fiberglass varies with density, and also with the addition of an FRK backing. > You pointed out yourself that you cannot reliably measure decay at all frequencies in a single small room. < Yes, understood. But for my purposes, I'm glad to get reasonably useful and repeatable test results at a few select low frequencies, rather than what I see as mostly vague results from a standard reverb room. As I mentioned before, I've seen negative values at 40 Hz in a reverb room test, even with a very large sample size. With the plots on the page linked above, all of the decay times are clear and unambiguous. > the change in decay ... vary significantly depending on the placement of the devices under study. < Of course! In this first round of tests I was very careful to place all of the samples in exactly the same places in the room. > you must give serious thought to what you will or won't be able to deduce from the research. < Agreed. I haven't even decided how to derive numbers from this! One problem is that ETF won't export data for 3D graphs. If I had that I could import it into Excel and derive the change in decay time at each frequency. I asked Doug Plumb, ETF author, to add this in a future version and hopefully he'll do that Thanks again. --Ethan is there any reason to believe that the traditional sets of interacting equations using in assessing, report, and discussing damping of mechaniacl resonances would not also be applicable to acoustical resonances? for example, the basic relations: half power bandwidth = the frequency higher than a resonance that is 3dB below peak, minus the frequency lower than the same resonance which is 3dB below peak loss factor = half power bandwidth/resonance frequency decay rate = loss factor * resonance frequency * 27 so via the above, Rt60=2.2/frequency*loss factor where frequency is the resonance frequency in question these, save Rt60 perhaps, are applied as a matter of routine to mechaniacl resonances. Is there any reason they should not also be applied to room modes? i am only somewhat familiar with the general calculations intrinsic to room acoustics, but could one take an FFT at a location that represented maximum expression of a modal peak, and provided you had a reliable FFT, utilize the above (or similar, but derived for this specific purpose) to estimate the damping of the room mode. ? Hi Brian, Thanks. Your comments, here and elsewhere, are all greatly appreciated. > i am only somewhat familiar with the general calculations < As you may or may not know, I'm not much of a math guy. So I have to leave it to people like you and Jeff and the other experts here to determine how to derive Sabins of absorption from decay. I'm not even sure how to get decay directly from my tests, since ETF will not output numbers for the waterfall plots. Maybe one could overlay a labeled 3D grid in a graphics program to better identify the various points on the graph? Then there's still the problem of some peaks being hidden behind others. I have a series of plots I took in the same room at five different locations, all with the room empty. In some room locations, modes that had been hidden behind others elsewhere in the room now show clearly, and vice versa. So it seems to me the next step is to try to convince Doug Plumb (ETF author) to add the ability to export the 3D data from those plots. One final point: Even though the peaks change amplitude a lot at the different places, the relative decay appears to be more or less constant. That is, modes that decay slowly at one place decay slowly elsewhere too, and likewise for modes that decay more quickly. So I take that as further proof that the basic premise of assessing absorption by measuring modal decay is valid regardless of where the measuring microphone is placed. --Ethan i realize here that my 3dB FFT concept has some significant flaws. The frequency response of the mic, the speaker, and whatever reflection effects were occuring at that measuring position would have to be accounted for, so... if one wanted to attempt to scrutinize a single room mode and the effect of some absorbing material on it, perhaps my idea was poor. left with decay rates, which isn't so un-satisfying from a pondering perspective. in any case, i wouldn't hazard to describe me as an acoustics expert :) , i have some background in modal analysis in mechanical structures, and i do find the analogy of in-room absorption tests to that interesting. Ethan, your thought about the graphical overlay is pretty ambitious, but ineteresting. i wouldn't have a clue how to attempt that. Ethan Winer said this: "Then there's still the problem of some peaks being hidden behind others. I have a series of plots I took in the same room at five different locations, all with the room empty. In some room locations, modes that had been hidden behind others elsewhere in the room now show clearly, and vice versa" Yes, that makes great sense. A common means of taking a peek at mechanical modes is with a clamped free beam, like this: http://hyperphysics.phy-astr.gsu.edu/hbase/hframe.html Except in some potential extreme circumstance, the modes are very well spaced in the frequency domain, and a single accelerometer position can reveal the lot of 'em. But at other times, one does impact or steady-state stimulation of more complex structures, and this overlap can/will happen then, especially in situations where damping is very high, and the modes are very broad. In those situations it can be useful to find some location on the structure in question where some given mode is maxmially expressed. Sort of like finding the place in the room where some peak is maximum, (i think). And then try again for another mode in the band you want to look at, and so forth. Maybe you room acoustics folks have models/sketches so sort of predict where these minima and maxima will occur. it can get tedious, but it's a necessary part of so many projects in industry, and it's done all the time. The nice thing about your project, i think, Ethan, is that you seek understanding of how those things compare, and not (per your comments in the past) a formal sabin number. Simplifies things alot. that's part of the beauty of a reverb room and measuring the average decay of many modes. This topic got me thinking about the possibility of measuring a "diffuse" modal field in a mechanical panel. what a pipedream that is, i'm sure the panel would be larger than most labs. lol perhaps it would be interesting to get a copy of ETF, i've never worked with it. Brian Brian, > The frequency response of the mic, the speaker, and whatever reflection effects were occuring at that measuring position would have to be accounted for < I don't see why the raw response matters, which is why I made the point that moving around the room changes the response but (apparently) not the decay rate. So even if the mike/speaker/room/whatever is down 10 dB at 40 Hz, the difference between the peak level at t=0 and at t=1 second should still reflect the correct decay rate. At least that's how it seems to me. > Ethan, your thought about the graphical overlay is pretty ambitious, but ineteresting. i wouldn't have a clue how to attempt that. < I imagined creating a grid with a transparent background that has the same angled shape and aspect ratio as the ETF graph. Then in a graphics program it would be laid on top of the graph to provide a "floating" 3D grid. It wouldn't be that difficult. But it's even easier to ask Doug Plumb to add an Export feature. :->) --Ethan > Brian, > I don't see why the raw response matters, which is why I made the point that > moving around the room changes the response but (apparently) not the decay > rate. So even if the mike/speaker/room/whatever is down 10 dB at 40 Hz, the > difference between the peak level at t=0 and at t=1 second should still > reflect the correct decay rate. At least that's how it seems to me. For your tests, it doesn't matter at all. I had proposed estimating room damping via the "3dB" method, and for that it would matter. So, no, for your tests i haven't any concerns about it. > I imagined creating a grid with a transparent background that has the same > angled shape and aspect ratio as the ETF graph. Then in a graphics program > it would be laid on top of the graph to provide a "floating" 3D grid. It > wouldn't be that difficult. But it's even easier to ask Doug Plumb to add an > Export feature. :->) Send that guy a link to all these discussions, maybe he'll get motivated and put some new features in...? :))) Brian Brian, > Send that guy a link to all these discussions, maybe he'll get motivated and put some new features in...? :))) < Doug Plumb keeps hinting to me that he's working on something new, but it's been a few months now with no further news. And it's not like I don't have plenty of other things to do! :->) --Ethan life is a busy thing, isn't it? so many projects get sort of shelved, and then forgotten. i wonder what the average person who is in an exploratory field has as a list of should-haves and wish-i-would-haves when they retire? |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.