Measure, Arrange, Absorb - Where to start & where to end? - Home Theater Forum and Systems -

Thread Tools
post #1 of 8 Old 02-05-12, 01:30 PM Thread Starter
Join Date: May 2011
Posts: 5
Measure, Arrange, Absorb - Where to start & where to end?

OK, I'm officially doing the OCD thing now. I've read way too many contradictory posts about what and when and where to absorb. I'm going nuts.

I have a roughly rectangluar room, with one closet occupying a corner.

18'4" x 11'7" x 9'

I have a pair of Quad ESLs. They are panel dipoles. I'm starting with the 2 channel setup, then planning to layer in the other channels later. No sub to contend with at the moment.

I've built 24" superchunks (OC703), but have a door in each corner, so can't do full wall height. I've built 1x4 frames to house 2'x4' OC703 2" panels for absorbtion. I've even built tube traps with 15" rigid fiberglass 1" pipe insulation.

But, where do I really begin? I've been taking measurements with REW. Trying to understand the results. My frequency response is anything but smooth. nearly 10dB peaks around 60 and 120 Hz. Over about 500Hz, i get 30+ dB spikes and dips extremely close together (comb filtering?)

My Waterfalls are "decent." 300 Time Range, 300 Window, a 30dB drop happens within 200ms. (Is this "good") My 60Hz drops 30dB within 350ms, then seems to go on forever.

Do I need to do more work (if so, what?) to address these massive spikes and dips in the mid-high range? Is it an artifact fo my measuring technique? Tools? Is it caused by too many reflections?

I have from time to time experienced a nearly holographic experience with the stereo imaging, but don't know exactly how I got it. Sometimes, very minor shifts of my head from left to right skew the center image in a big way.

I've moved the speakers and seat to address the 60Hz issue, but lost my soundstage in the process.

When are we "done"? I mean, what sort of "ranges" are good enough? When do I get to stop obsessing on the numbers, and room treatemnts, and measurments, and actually get to listen to the music?

kdubious is offline  
Sponsored Links
post #2 of 8 Old 02-05-12, 09:48 PM
Join Date: Dec 2009
Location: CST
Posts: 516
Re: Measure, Arrange, Absorb - Where to start & where to end?

The first thing you do is stop.

And then, very systematically, approach the issue focusing on the complete analysis of specific behaviors rather than the hodgepodge jumping around that is occurring… The process will go much quicker and be much productive if taken sequentially and completely addressing each aspect rather than making changes that interactively adversely affect other factors due to their being conducted out of sequence...

And I hope that you do not expect a comprehensive answer in a paragraph...Additionally, to do this correctly, it will not simply take 30 minutes. But you are free to do as much or as little as you like - realizing that the result will be a compromise of what is optimal and possible.

First, does the proposed listening position offer significant left-right symmetry?
If not, select a position that does.

A position that lacks significant LR symmetry will not be able to be treated so that optimized accurate imaging and localization is possible. This is critical.

The following assumes also that you have already determined the optimal crossover for the mains and sub and the the adjustments necessary to signal align the main and sub are understood and able to be implemented. Otherwise the measurements of modal behavior will also reflect errors in the signal alignment of the speakers and it will be difficult to separate what errors are due to the speakers and what is a result of room interaction. Thus, speaker crossover and signal alignment adjustments (including SBIR related positioning) will want to be resolved first. This procedure does not address those specific issues.

Remove whatever treatment you have installed.

Once such a symmetrical LR orientation is identified, set the system up (along with the proper alignment of the mains and sub):

Generate a (edit: full range) unsmoothed frequency response (edit: driving one speaker) and convolve the unsmoothed frequency response into a waterfall plot. Using the "Limits" settings window the response from 0-250 Hz and set the Z axis time display to about 1 second. Adjust the Y axis gain setting to display about 50 dB from signal peak. (Edit: You might want to toggle the "freq axis" tab in order to select a 'linear' frequency display (as opposed top logarithmic). )

At your assumed listening position, you will want to map the region for nulls.
You can either make multiple measurements mapping the region around the listening position for the presence of nulls - or, you may use the signal generator in REW to slowly sweep the frequencies around where the waterfall indicated nulls to identify the modal resonances and nulls at the listening position.

The most basic adjustment that can be made to minimize the nulls (as you cannot effectively treat them), is to measure the positions both forward and behind the current listening position in order to relocate the listening position (forward or back) to a region in neither a peak nor a null while maintaining LF symmetry.

One the listening position is formally determined, the mic position will effectively be frozen and maintained in the exact same position for ALL remaining measurements You may wish to develop a plumb bob or similar means to locate the capsule of the mic for future reference as well. Without this critical point being maintained consistently, you will not be able to accurately compare... measurements.

You will then want to apply corner 'Superchunk' style bass traps made of fluffed 'pink fluffy stuff' Fiberglass insulation, preferably with an asymmetric orientation with the front wall leg being ~4-6" longer (~30" total) than the side wall leg (24"). The face will be covered with a minimum of 6 mil. plastic that will be effective in reflecting most of the energy above ~600 Hz.

You will then repeat the waterfall at the listening position. Depending on the response, you will decide whether you can apply corner traps to all of the vertical corners of the room and/or to also add horizontal corner traps to the wall/ceiling vertices. If you have adjacent rooms who opening are on the front or rear walls, you may want to experiment with the effect that opening or closing these doors have on the modal distribution. Doorways on the L/R side walls will want to be closed for reasons of specular symmetry.

At this point, if you have a suitable parametric EQ unit, you may want to also employ REW's ability to generate filters to moderately EQ the LF peaks below about 80 Hz.

After the modal issues are addressed... and maintaining the exact same microphone measurement position throughout...

Utilizing a properly configured REW (with hardware propagation delay loopback properly enabled…

You will want to generate individual Envelope Time Curve (ETC) responses for EACH main speaker.

This is done by generating a full range (5Hz to ~20 kHz) UNsmoothed frequency response, convolving the impulse response, and then convolving the ETC response.

From these you will want to window the ETC response for maximum visibility with the Limits settings of approximately minus .0005s to ~.0035s on the X axis time scale, and from -30 dB to .5 dB on the Y axis gain scale. This should provide an approximate view that will allow for maximum detail of the main room response.

From this you will want to identify the high gain early reflections as well as anomalous sparse later reflections that exhibit a higher gain than their surrounding arrivals.

You will then resolve these indirect energy arrivals into their vector components of direction while identifying and marking their point of boundary incidence with blue painter’s masking tape using either the string method or the blocking method.

You will do this for each ETC plot, labeling the plot and the boundary point with a label allowing you to correlate each point to a specific energy arrival peak.

Then, based upon the preferred acoustical response model desired, appropriate treatments will be selected and applied addressing the various boundary incident points.

Typically, early arriving high gain energy is treated with absorption and later arriving sparse high gain energy is treated with diffusion. But the specific treatments will depend on the desired response and the location of the points of incidence relative to the other points…

After applying each treatment, a repeat ETC will be generated allowing for the proof of performance of the treatment and any adjustments as necessary.

Note also that the 2" thick 'broadband" panels are not broadband - as they will not mitigate the low mid specular reflections, effectively EQing the reflected energy as a low pass filter.... You will want 4" thick panels placed 4" from the wall for this. The result of using absorption that is not adequate to address the full range of specular energy ill have the effect of EQing the reflection, rolling of the highs while allowing the low-mid energy to persist, resulting in the coloration of the direct sound as well as a detrimental effect upon localization, imaging and intelligibility when it superposes (combines) with the direct signal at the listening position. The result will be simply a ‘moving about’ of the problem rather than a solution.

Note also that significant control of the rear wave of the ESL dipoles will be required as this will result in significant destructive early arriving high gain signals. thus either substantial application of broadband absorption on the front wall, or substantial redirection (controlled reflection) of the rear wave energy past (but not incident with) the listening position will be required. Redirection is the preferred method substantial benefit that offers - IF possible.

I realize that this is a 30,000 foot synopsis of the procedure and that the details are simply not able to be specified as they are not only myriad, but situation specific.

But if you systematically follow this, we can help with specifics along the way. Also, when referring to measurements please post the raw REW .mdat data files that will allow us to convolve views of the data.

Last edited by SAC; 02-06-12 at 05:12 PM. Reason: clarity/details
SAC is offline  
post #3 of 8 Old 02-06-12, 07:03 AM
HTS Senior Moderator
Wayne A. Pflughaupt's Avatar
Join Date: Apr 2006
Location: Katy, Texas
Posts: 8,809
Send a message via Yahoo to Wayne A. Pflughaupt
Re: Measure, Arrange, Absorb - Where to start & where to end?

kdubious wrote: View Post
But, where do I really begin? I've been taking measurements with REW. Trying to understand the results. My frequency response is anything but smooth. nearly 10dB peaks around 60 and 120 Hz. Over about 500Hz, i get 30+ dB spikes and dips extremely close together (comb filtering?)

My Waterfalls are "decent." 300 Time Range, 300 Window, a 30dB drop happens within 200ms. (Is this "good")
As SAC noted, your speakers present a challenge for acoustical measurements and/or treatments. The comb filtering you’re seeing could well be caused by them. It can also be caused by measuring multiple speakers simultaneously, so be sure and measure only one at a time. It would be helpful if you could post a graph so we can see what you’re seeing.

Posting A Graph
Getting Graphs Ready to Post

Your waterfalls sound pretty good, but (again) a graph would be helpful. For a waterfall, lower the graph floor to 30-35 dB, and make sure the signal when you measure is 50 dB or so above the room’s noise floor.

My 60Hz drops 30dB within 350ms, then seems to go on forever.
That’s 60-cycle noise from an extraneous source, like a washing machine, refrigerator, A/C system, etc . Make sure all that stuff is turned off during measurements.

Wayne A. Pflughaupt is offline  
post #4 of 8 Old 02-06-12, 08:43 AM Thread Starter
Join Date: May 2011
Posts: 5
Re: Measure, Arrange, Absorb - Where to start & where to end?

Thanks Wayne and SAC.

I've cleared out the room, which now sounds like an echo chamber, and am about to start measuring.

I have been doing both speakers at once, and not using the "loopback", so that is changing now.

I'll be back shortly with data & questions.

kdubious is offline  
post #5 of 8 Old 02-06-12, 11:27 AM Thread Starter
Join Date: May 2011
Posts: 5
Re: Measure, Arrange, Absorb - Where to start & where to end?

The room is a rectangle, and I took a 12 measurements, sweeping from 40 - 250Hz moving down the center line of the long dimension. I'm not using a sub, and may not ever use one. I'm focused on my 2-channel setup for now. The "theater" may be another room.

Some positions have severe dips in the SPL graph. I'm not exactly sure how to follow this:

Generate an unsmoothed frequency response and convolve a unsmoothed frequency response into a waterfall plot. Using the limits settings window the response from 0-250 Hz and set the Z axis time display to about 1 second. Adjust the Y axis gain setting to display about 50 dB from signal peak.
I generated Waterfalls, and am looking at 0Hz - 250Hz. My highest peak is 80dB, so I am looking at 80dB to 30dB. I set this in the Limits menu. That view doesn't identify "nulls" as easily as the SPL view... unless the SPL view is not showing Nulls. Z axis was set using the "controls" menu "Time Range" setting... not the Window setting (it is still 300) nor the Z slider. It is 150.

Also, I'm not sure how to compare some of the graphs. I suppose "Higher Frequency Nulls" will be easier to "treat", and are therefore preferred over low frequency ones?

Thanks again, and I'm ready for more tips.

kdubious is offline  
post #6 of 8 Old 02-06-12, 06:26 PM
Join Date: Dec 2009
Location: CST
Posts: 516
Re: Measure, Arrange, Absorb - Where to start & where to end?

Looking at the waterfall plots...

A few suggestions about which I was not extremely clear... For all measurements, just make full range plots - do not band limit them. It hurts nothing to include more than less, and 'less' can potentially come back to bite you... It is easy to set the "Limits" in order to control the degree of 'zoom' and what you see. (This function is shown on the annotated plots...) Also, you may want to look at the linear plot rather than the logarithmic plot. (...for limited LF response plots i definitely prefer this...)

I am assuming that only one speaker is driver.

Plot 13 seems to exhibit the least severe nulls. We can use treatment and, to a degree, EQ to mitigate peaks, but not to minimize nulls, so selecting a location least effected by them is a good starting point.

Assuming all is 'good', for a square room there is an amazing lack of strong modal activity, noting that there is a distinct (and curious) lack of strong modal ridges extending from the display. In fact, the most prominent resonance seems to be a 60 Hz electrical hum which you will want to troubleshoot separately...

By moving the speaker in one dimension only - e.g., towards the front wall maintaining the side wall spacing; or separately. moving the speaker toward or away from the side wall but maintaining the front wall spacing...), and repeating the measurement with the mic is PRECISELY the same spot (this is critical for all measurements), if the nulls move then you have identified what is called SBIR as a culprit. This is a local cancellation where signal 180 degrees off one boundary superposes to create a null. This can be mitigated by speaker placement and/or by treatment.

If SBIR is not the culprit, then the nulls is room supported by way of the modal distribution in the room determined primarily by room dimension, boundary impedance, and to a degree by the location of the speaker in the room.

You will want to be considering large floor to ceiling Superchunk style 'pink fluffy stuff' corner traps for all corners as a given.

Seeing as how the ESLs do not have an excessively low frequency response, some EQ can be used to mitigate the peakiness...

I am also guessing that with the excessive degree of rear radiated sound, that not only are you are going to have a significant degree of SBIR as noted earlier that may very well be contributing to the higher end nulls, but that you WILL have excessive comb filtering in the higher specular frequencies. The nulls above ~200 Hz may also very well be caused in part by this. Note too that EQ will have little usefulness on this.

Anticipating future areas of concern:
You will want to pursue this aspect further using the ETC response and noting the behavior and interaction of the direct and indirect energy. This will be exacerbated by the fact that you will not only have indirect energy and reflections from the forward radiated energy, but you effectively have a '2nd speaker' facing rearward generating out of phase energy as well. While I personally admire the transients of an ESL speaker, the rearward propagated wave wreaks havoc (and that's as polite as I can be!) in a small to medium size bounded space.

Just be aware that SIGNIFICANT broadband absorption will be required - in addition to bass trapping, to control the behavior sufficiently to optimize the performance of these speakers. In other words, you will be tasked with essentially eliminating this non-minimum phase rearward oriented energy in addition to addressing the indirect pathways generated from the forward directed energy.

For this you will need broadband absorptive panels, the best best material density characteristics for which are either ~3lb/ft^3 Fiberglas or 4 lb/ft^3 mineral wool. The minimum configuration that should be considered for a broadband panel is 4" thick with a 4" boundary gap which will effectively give you the performance of an 8" thick panel. Any less and you will not adequately address the low mid frequency energy. This is important as lesser absorbers will effectively EQ the indirect signal by absorbing only the higher frequency energy and the result will be the coloration of the direct signal by the superposition of the low mid energy hat is unmitigated. You will also suffer from the loss of imaging, localization, intelligibility - which would be a rear shame, as these aspects are areas in which these speakers can really excel! So I would seriously suggest attention to this facet of the treatment as determined by the ETC.

Below is a capture of the freq response for plot 13 along with its waterfall.
Again, I will emphasize the conspicuous (and curious!) absence of serious modal resonance (in a square room!!!) aside from the 60 cycle electrical hum issue. (if the laptop and pre-amp are not running off batteries, if you can, see if you can try that and if the hum goes away...)

(I would personally bring in a sub, place it in the front right floor corner of the room and, with the aid of a boom mic stand, place the measurement mic diagonally across the room in the in the rear upper (ceiling-wall) left corner and run some sweeps in order to determine the lowest modes that the room supports in order to note any differences.)

Also, if you like, PM me and we can speak by voice if you like. Its infinitely easier and Much quicker to interactively pursue real issues rather than to abstractly attempt to account for every potential possibility!
SAC is offline  
post #7 of 8 Old 02-06-12, 09:13 PM Thread Starter
Join Date: May 2011
Posts: 5
Re: Measure, Arrange, Absorb - Where to start & where to end?

THanks again, SAC. I know it takes time to write thoughtful instructions.

The room is a rectangle (dimensions in my first post), and I didn't remove the 24"x24" superchunk triangles in each of the top tri-corners. (18" tall) So, maybe that's the reason that the bass response is "decent" as a starting point?

I hung my 2x4 broadband traps (4" OC703) off the wall 4" to catch the reflections of the speaker to "Its" side wall, and the speaker to "the other" side wall. That drastically reduced one band of spikes in the ETC. Curiously, the "next worse" spot was a tie between the first reflection off the ceiling and a 2-bounce reflection "ceiling-back wall". Same absorption setup there. Finally, I added the same OC703-4"-4"Off wall behind the speakers. ALl of this drastically reduced the peaks in the ETC.

The remaining spikes are more than 20dB lower than the direct sound. It that sufficient?

I hope to have some time tomorrow to spend building my ceiling trap and doing additional measurements.


kdubious is offline  
post #8 of 8 Old 02-06-12, 09:21 PM Thread Starter
Join Date: May 2011
Posts: 5
Re: Measure, Arrange, Absorb - Where to start & where to end?

It seems I need 5 posts in order to send a PM.
kdubious is offline  


absorb , arrange , end , measure , start

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