How do room modes show up on a waterfall plot? - Page 2 - Home Theater Forum and Systems -

Thread Tools
post #11 of 17 Old 07-15-08, 11:43 AM
Posts: n/a
RT60s in a small acoustical space.....hmmmmm.

RT60s in a small acoustical space.....

As mentioned above, the CSD/waterfall plots are most useful for the identification and mitigation of LF room modes.

But what do we use to examine the behavior of a small acoustical space above the frequencies dominated by modal behavior where the wavelengths at issue become small relative to the obstacles they encounter?

Unfortunately, there persists the notion of using RT60/30 and their variants in small acoustical spaces. This was a rather common notion 35 years ago, but has since been eclipsed by a greater understanding of acoustical behavior in what is defined as a small acoustical space (as distinct from, yup, you guessed it, the large acoustical space!)

It is time we moved on to understand and to embrace the new acoustical model suggested by Schroeder that has been rigorously verified, for all of our mutual benefit.

In a small acoustical space, the use of RT60s (and any similar variant, regardless of weighting) is completely erroneous, and indicates that we are misapplying a measurement that has no correlation in physical fact in a small acoustical space.

Fundamental to this is an understanding of the distinction between a large and a small acoustical space. These terms are not used casually, as they have very specific conditions which define and differentiate the two. And a failure to distinguish between the two will result in some rather unfortunate consequences.

This distinction is not only the subject of much of Manfred Schroeder’s work, but it is a fundamental concept that underlies all of his work.

And to reduce the distinction to its most basic functional difference, a large acoustical space features a developed statistically random reverberant sound field where reflections at any location are equally probable to radiate from any direction. Conversely, a small acoustical space LACKS a statistically random reverberant sound field where reflections at any location are equally probable to radiate from any direction – and instead, is DOMINATED by focused specular reflections which are definitely identifiable as decreet phenomena as a vector with both direction and intensity and a discreet time of arrival and decay that an be measured and all characteristics identified and isolated via such measurements as the envelope time curve (ETC).

The RT60/30, etc. are ONLY suitable to measuring the decay times of a statistically random reverberant sound field. As such, they have no place in the small acoustical space. If one only has a hammer, all of the world begins to look like a nail, and that is precisely what has happened to the mis-application of the RT60! Instead, the proper tools that accurately identify the real phenomenon in the small acoustical space are to be utilized.

Someone might attempt to use a rock to drive screws, but few would be so bold as to suggest that this methodology is suitable for use in the art of cabinetry. And if you are going to expend the effort and the money to do the job, it is incumbent that we use the proper tools that afford us accurate atomistic insight into the actual phenomena of focused specular reflections, rather than simply using the wrong tool suitable for measuring that which does not exist in a small acoustical space.

Understanding this also leads one to use absorption surgically instead or randomly and with prejudice! As being able to see what is actually occurring has a profound impact upon one’s acoustical world view. And as we have had the tools to do exactly this now for over 30 years, it is time that more in the world of acoustics move up and embrace the current models. After all, even the most stalwart of nay saying classical physicists have begrudgingly admitted that there just might be something to quantum electrodynamics!

Its time to put the idea of using RT60/30/etc. in small acoustic spaces, to bed!

To quote from Sound System Engineering, 3rd Ed.[/B] Pp 178-9:

“Small Room Reverberation Times

To quote the late Ted Schultz (formerly of Bolt, Beranek and Newman)
“In a large room, if one has a large sound source whose power output is known, one can determine the total amount of absorption in the room by measuring the average pressure throughout the room. This total absorption can then be used to calculate the reverberation time from the Sabine formula. This methods fails badly in a small room, however where, a large part of the spectrum of interest lies in a frequency range where the resonant modes of the room do not overlap but may be isolated…In this case the microphone, instead of responding as a random sound field (as required for the validity of the theory on which these methods depend), will delineate a transfer function of the room… It does not provide a valid measurement for the reverberation time in the room.”*

What is often overlooked in the attempted measurement of RT60 in small rooms is that the definition of RT60 has two parts, the first of which is commonly overlooked.

1.) RT60 is the measurement of the decay time of a well-mixed reverberant sound field
Well beyond Dc (the critical distance).
2.) RT60 is the time in seconds the reverberant sound field to decay 60 dB after the sound source is shut off.
Since in small rooms, there is no Dc (critical distance), no well mixed sound field, hence no reverberation but merely a series of early reflected energy, the measurements of RT60 become meaningless in such environments.
What becomes most meaningful is the control of early reflections because there is no reverberation to mask them.”

What you have instead is a small acoustic space dominated by room modes (which are not reverberation!) and focused spectral reflections – which by definition are anything but a well-mixed reverberant sound field wherein the arrival of reflected signals from every direction is equally probable.

In the small acoustical space, there is no mixing, nor homogeneous, statistical reverberant sound field! In fact, in a small acoustical space, as the lowest frequency that can effectively develop across the largest dimension can easily increase to ~500Hz – compared to a large acoustical space where such frequencies are often below 30Hz!

In such environments, intelligibility can be degraded by specular reflections that must be isolated and corrected directly, not statistically.

*Note, It is equally proven that the fundamental form of Sabine’s expression cannot be modified so as to become correct for large absorption. Per Sabine’s Reverberation Time and Ergodic Auditoriums, Wm. Joyce, Journal of the Acoustics Society of America, Vol. 58, No. 3, pp. 643-655,Sept. 1975.

Imagined non-existent RT60s, or some fascinating “similar but scaled down analysis” has no basis in fact or physics! And they are of no valid use here. The simple persistence of acoustical energy is of little use in a small acoustic space!

An RT60 is useful in a large acoustic space as the statistical reverberant field is sufficient to mask specular reflections. But this is NOT the case in a small room where the specular reflections EASILY dominate over the all but nonexistent statistically reverberant sound field (where they exist only at frequencies of interest to UHF engineers, dogs and bats!).

The simple persistence of the acoustical energy fails to provide ANY insight into the specific nature of specific focused specular reflections which are the real issue of concern! Such a measurement fails to reveal the intensity, arrival time, or any information that aids in the identification and behavior of, let alone provides specific information regarding the effectiveness of the surgical treatment of any focused specular reflection – and by their nature such a measurement would be location specific as you are NOT in a statistical reverberant field! And an average over many locations would be completely nonsensical! In fact, such a concentration leads only to the shotgun application of absorption – precisely that which we are trying hard to avoid!

It is the persistence in the belief that completely invalid variations of a technique designed to evaluate that which does not exist in a small room that has held back a real understanding of acoustic behavior in small acoustic spaces! Regardless of how one squints and tilts their head when evaluating such nonsense in a small acoustic space.

I realize that this steps on the toes of a few here. But its time to move on to the new tools and to use the measurement techniques which we have at our disposal to better and more accurately understand the physical phenomena of acoustics and sound for all of our benefits.

I will be posting this in a separate independent thread as well (with a few additional comments as to the source of this common misconception), as this topic comes up in many threads...

Last edited by mas; 07-15-08 at 12:32 PM.
Sponsored Links
post #12 of 17 Old 07-15-08, 12:01 PM
HTS Senior Moderator
bpape's Avatar
Bryan Pape
Join Date: Sep 2006
Location: Wildwood, MO (St. Loui
Posts: 5,209
Send a message via AIM to bpape Send a message via Skype™ to bpape
Re: How do room modes show up on a waterfall plot?

I assume you're addressing me and that's fine.

I will agree that RT60 and RT30 do not apply to small spaces. One can, however, use the same calculations to determine RELATIVE persistence across the frequency spectrum with the end result only being used to properly address what portions of the spectrum may need more or less absorbtion (broadband or targeted). They do not present accurate target ranges specifically for the reasons you've addressed and should not be used for such.

That's all I'm going to say on the matter as I'll not have another thread turned into an uncivilized debate.


I am serious... and don't call me Shirley.

Bryan Pape
Lead Acoustician
GIK Acoustics
bpape is offline  
post #13 of 17 Old 07-16-08, 02:22 AM Thread Starter
Senior Shackster
Join Date: Jun 2008
Location: Vancouver, BC
Posts: 162
Re: How do room modes show up on a waterfall plot?

A lot of good comments and interesting points of view. Coolio!

I verified that my original waterfall plot was correctly taken except I forgot to apply the mic calibration My observation remains: the axial room modes do not show up as nulls which I would have expected given the centered mic position. However I'm not so fussed about this given the comments in this thread...clearly real world rooms behave in a much more complex manner than these simple models suggest.

Anyway, older and wiser, I will now take the advice to "just measure it" and go from there!

Although I'm not sure I ever got an answer on how room modes show up on a waterfall plot. Presumably peaks and dips with ringing are modal responses. But is that the criteria for identifying them? REW purports to find them...does anyone know what it looks for? Maybe no-one cares...but I would have thought non-modal induced peaks and dips would be handled differently (ie less inclined to EQ them, more inclined to absorb or diffuse them etc).
bjs is offline  
post #14 of 17 Old 07-16-08, 03:45 PM
Posts: n/a
Re: How do room modes show up on a waterfall plot?

The modes 'show up' in a waterfall plot as 'persistent peaks' - meaning that they have a greater amplitude and they persist in time - in other words they appear as a ridge. Also, as I am not sure of your expectations, you are going to see the gross summations of the total modal behavior! Which is what you need to address anyway.

Remember, just because you have a measurement, that does not mean that you are finished.
Instead, exactly 'how' the measurement is made, including the capabilities of the tool, have a great impact upon your final results, and such seems to be the case here.

Just a few quick cursory observations…{as I don't have MatLab on this computer and I quite literally don't have the time to do all of the manual trig to determine each effective reflective path length relative to the side wall and ceiling (and you cannot rule out the rear wall and possibly floor) to determine the corresponding time of travel from source to each first incident reflection point to microphone in 3space...}

(Besides, I have become spoiled as TEF and EASERA will display the time/distance variable breakdowns for various absolute and relative reflections if we just run a corresponding ETC of the same setup! ;-) )

Likewise, I have no idea of the window and resolution settings or limitations of your rig...


Below ~50 Hz I question the measurement, which I suspect is a direct function of the windowing/ resolution of the particular measurement system.

But you nevertheless appear to have resonances centered ‘about’ the 32, 37, 47, 63, 85, 110, 135, 160, 180, 223 Hz points. (Another comment here…using the log scale is great if your display covers a bandwidth of 20kHz, but for simply a 200 Hz window, use the linear scale! However in such a limited window, I would prefer not to compress 100 Hz of response in the display and obscure measuement…But that is a personal choice…)

Additionally, you appear to have severe comb filtering effecting your measurement due to the superposition of the reflection(s) from the ceiling and side walls as indicated in the severe notches at 120-145Hz and 190-210 Hz regions.

Again, not to belabor the point, but at the same time also acknowledging that you cannot ignore the facts; without the additional abilities of the ‘larger’ rigs such as TEF using time delay spectrometry, you lose significant ambient noise immunity with the lack of tracking filters to effectively adjust the windowing capabilities and to effectively 'remove' the destructive reflections. So again, your measurement rig and technique is going to become interactive with the response you are attempting to measure. (you will see again and again that the larger tool sets provide many capabilities and much additional derivative information – and hence this what also determines their additional cost… And to go one step further to illustrate this, in large halls, as running a million chirps tends to drive one crazy, with the TEF/TDS we are able to put something on that we like to listen too over the system under test while we simultaneously run sweeps; whereas with MLS and other techniques, noise immunity is a severe limitation. This is not to say that you absolutely need them, but when one is trying to make a living doing this and your goal is to get in and get out as efficiently as possible, versus having the luxury to take as long as you like as it is a hobby, the returns translate into real considerations ;-) )

For the wavelengths you are measuring, your window is going to have to be large, and this is also necessarily problematic for any measuring device, exacerbating your technique and control of associated variables which will vary (to a degree) with your tool.

Noting that you are not so much measuring the room’s response as you are measuring the locations response, the deviations from the old chart is not at all surprising. They will not be identical to the total summed room response. If you are desirous of comparing the calculator’s response, you want to drive the room and in order to do this I might suggest placing the microphone ‘in’ the pressure zone of the diagonally opposite corner facing away from the speaker.

In the listening position you are going to be measuring variations from that overall room measurement. For instance, if you are in an axial null, the tangential and oblique modes will dominate, etc. Hence why the process becomes iterative with modifications to the listening position and position of the sub in an attempt to minimize the summed destructive interference amidst a complex environment.

Anytime you make a measurement you are going to need to step back and determine if the results are reasonable. Obviously there is a quality/quantity that is unknown, or why would you be measuring it… But never cease to question things that look too good to be true. For example, if you are measuring the response of a 3 inch driver and you see a prodigious ‘bump’ in output at 60 Hz, would you reasonably expect a 3 inch driver to output significant energy at 60 Hz? If you are anywhere but Bose (sorry, couldn’t resist!) you might start looking for a source of 60 cycle hum in your measurement rig.

Another aspect to consider is the test’s repeatability. Can you reliably repeat the experiment and achieve the same results. Can another person, given the same unit under test?

I am not able to tell yo9u exactly what the performance parameters of your tool are. I simply don’t know, so this will necessarily remain a 50,000 foot analysis…

But a few issues to consider in any acoustic measurement environment that often show up to bite you in the posterior if you are not vigilant…

First is the uncertainty principle…When a measurement is taken, there is some limit to the resolution, or how much detail we can resolve. If we make measurements with a resolution of say 1ms, then we are going to be unable to see any events that occur in a shorter time frame than 1ms. Now this doe not necessarily mean that we can’t see them! In fact we may, but they will be imprecise and distorted – a blur, just like if you need glasses and try to read the fine print that is beyond your limits of visual resolution.

In audio, if we make measurements with a resolution less than say 1 kHz, then any details less than 1kHz will be blurred. Likewise, if we repeat the experiment with a resolution of say, 500Hz, then more details between 500 Hz and 1kHz will become clearer, but we will be unable to clearly see anything below 500 Hz.

Because time and frequency are reciprocals, as our acuity in one domain increases, the corresponding accutiy in the other domain decreases. This is a function of the relationship, and not our equipment. Thus, as our acuity in one domain approaches infinity, our acuity in the other approaches zero.
This relationship is unity, meaning that we are bound by this relationship. As we gain acuity in one domain, we lose it in the complementary domain.

To bring this to bear on the situation we are facing here… If we wish to measure, say the response of a speaker with a resolution of 20 Hz, our resolution is 1/20, or .05 seconds. At STP(standard temperature and pressure), this .05s corresponds to a distance of 56.5 feet. This means that any reflections within that distance will be included in the measurement an will contribute to a false reading relative to our initial goal. If the resolution is changed to say, 500Hz which corresponds to a time of .002s and a distance of 2.26feet, true anechoic measurements can actually be taken.

Technologies such as TDS (time delay spectrometry) exhibit adjustable tracking filters which track the stimulus and bound the reception of the response signals and filter out reflections outside of the bounded region, effectively enlarging our scope by filtering out destructive interference and enlarging our scope.
But nevertheless, we are all subject to the same fundamental issues – regardless of the technologies that may enhance one technique over another in a particular environment.

And I believe we are seeing aspects of some of them here. For instance, when 2 (or more!) signals arrive at the measurement microphone with nearly identical levels with one slightly delayed in time, we experience a phenomenon known as comb filtering. And depending upon our resolution, this may or may not be readily apparent, but it is nevertheless real! This is exactly the result when we set too large a time window and a reflection is included along with the response in which we are really interested. Such notching is far apart if the signals arrival times are close, and closely spaced in the arrival times are far apart. On a log scale this can be difficult to see. And hence one of the reasons linear scaling is becoming increasingly popular.

This type of interference is common whether the source be two speakers or a real and a virtual source such as is generated by diffractive or reflective sources.

When making frequency measurements, the frequency resolution must be considered. In order to measure a frequency, there must be sufficient time to measure at least one full cycle. Thus, if have, say, a frequency resolution of 500 Hz, the period of one cycle is 1/500 or 2ms. A general rule of thumb is that the data will be reliable from a frequency that is equal to ½ of the resolution frequency. Thus, with a resolution set at 500Hz, we can reasonably expect the data from 250Hz up to be sufficiently reliable. The problem arises when we do not recognize this relationship and we assume the additional data that will be displayed to be reliable – as far too often happens!!!

Noise is another critical issue. (TDS is highly immune to noise, so that unless you are using very high sweep rates in very noisy environments, noise is not a problem.) However for other platforms, and always for ETCs as a wider filter bandwidth is required, In many cases noise is interpreted as part of the system under test. Level adjustments can help.

{In the other thread it was asked to what TEF, TDS, EASERA, etc. referred. Here it might be useful to mention that one of the advantages to TEF/TDS is the ability to effectively and accurately measure down to dc- 0Hz – in addition to ‘negating’ or extending many of the heretofore mentioned limitations. And thus it is an excellent instrument to also measure the complex impedance of a device…which leads us to another entire realm of Nyquist and Heyser spirals displayed in 3space and an entirely amazing and feature rich environment that has yet to be fully explored!}

post #15 of 17 Old 07-17-08, 02:18 PM Thread Starter
Senior Shackster
Join Date: Jun 2008
Location: Vancouver, BC
Posts: 162
Re: How do room modes show up on a waterfall plot?

mas wrote: View Post
The modes 'show up' in a waterfall plot as 'persistent peaks' - meaning that they have a greater amplitude and they persist in time - in other words they appear as a ridge.
Given room modes are standing waves, one would expect they could also manifest as initial depressions that ring on into ridges. After all, the microphone may not necesarily have been placed in a peak of the standing wave.

...but for simply a 200 Hz window, use the linear scale!
I also prefer linear in this case, especially for room modes which are linearily related. However REW doesn't display waterfalls in linear scale properly so I have to use log.

I tried the standard scaling used here at HTS but found it poor for waterfalls. A 45dB floor is much too low creating a great deal of artifacts obscuring the key results. Especially in my noisy measuring environment!

Additionally, you appear to have severe comb filtering effecting your measurement due to the superposition of the reflection(s) from the ceiling and side walls as indicated in the severe notches at 120-145Hz and 190-210 Hz regions.
Quite correct, and front/rear walls too! Attached is the ETC plot for this same position. Unfortunately ETC plots are not perceptually uniform in frequency. Lower frequency reflections are harder to see and easily masked by higher frequency reflections. However since I tested the subwoofer with a wider than normal bandwith the reflections are visible. The flutter echos are perhaps reasonable given the room has nothing on the walls...just hard flat gyproc surfaces.
How do room modes show up on a waterfall plot?-rommodaltest_micincenterofroom_rightwooferincorner_etcplot.jpg

Another aspect to consider is the test’s repeatability. Can you reliably repeat the experiment and achieve the same results. Can another person, given the same unit under test?
Well, like Britney Spears...I did it again...! A week later and this time with higher frequency resolution. I probably put the speaker back within a few inches of the original test and the mic also (mic is probably within an 1inch). And I used a different mic this time... a cheap Radioshack SPL meter (it probably cost me $30 some thirty years ago). And I've learned how to properly save and upload my graphs so it looks a little different this time. Having said all that, the results look very similar.
How do room modes show up on a waterfall plot?-roommodaltest_micdeadcenterofroom_wooferinrightcorner_version2.jpg

Your comment about stepping back and checking the results is a good one. The ridges at 16hz and again at 24hz are due to heavy industrial air conditioning nearby.

Anyway, it seems I have enough tools ready and they make enough sense to me that I'm actually ready to start measuring and treating my room! I start a new thread for that and see where it goes. People can watch me learn "on the job" as I'm sure I'll do some daft things at times...but that is the great thing about this site...a lot of experience and expertise for the newbies!
bjs is offline  
post #16 of 17 Old 07-17-08, 03:25 PM
Posts: n/a
Re: How do room modes show up on a waterfall plot?

Just a comment and a suggestion that I hope will help in the future...

That response is looking much nicer...

I've not spent much time with equipment that displays the particular format of the ETC that RoomEQ seems to show - of an averaged line connecting the amplitude peaks of the various reflected specular signals as opposed to also displaying the individual constituent reflections.... as discreet events in time and amplitude...

In order to identify (and verify) the actual sources of each peak and null, you can take a small piece of OC703 or equivalent, and standing in a spot behind the mic, preferably in its null so as to avoid contributing to the measurement with your reflection or shadow - verified by repeating a known good measurement - and holding the absorptive material such that it intercepts what is assumed to be the signal pathway. If you have assumed correctly, subsequent measurements with the absorptive material impeding the pathway will show a reduction in the specific ETC response gain corresponding to the particular arrival time of the reflection.

I suspect this has already been suggested elsewhere in detail, but if not, I offer it simply as a helpful suggestion...

And once you have addressed the room modes, you will want to switch your point of view to the ETC and the time/amplitude perspective rather than the frequency perspective.

Have fun!

Last edited by mas; 07-17-08 at 04:26 PM.
post #17 of 17 Old 09-19-08, 06:54 AM
Senior Shackster
jerome's Avatar
Join Date: Apr 2007
Location: Norway
Posts: 233
Send a message via MSN to jerome
Re: How do room modes show up on a waterfall plot?

Very interesting reading mas! I'm not sure I got the more complex details but please continue explaining us newbies how things work

Completely OT but I was just wondering: in short, what is your background? Should we be aware of which company your work for?

Klipsch (THX2) | Crown | SVS | SlimDevices | Sony | Pioneer | Yamaha
jerome is offline  


modes , room , waterfalls

Quick Reply

Register Now



Confirm Password
Email Address
Confirm Email Address
Random Question
Random Question #2

User Name:
Please enter a password for your user account. Note that passwords are case-sensitive.


Confirm Password:
Email Address



Activation requires you reply to an email we will send you after you register... if you do not reply to this email, you will not be able to view certain areas of the forum or certain images... nor will you be able download software.


See our banned email list here: Banned Email List

We DO NOT respond to spamcop, boxtrapper and spamblocker emails... please add @hometheatershack DOT com to your whitelist prior to registering or you will get nowhere on your registration.

Email Address:


Human Verification

In order to verify that you are a human and not a spam bot, please enter the answer into the following box below based on the instructions contained in the graphic.

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML is not allowed!
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


For the best viewing experience please update your browser to Google Chrome