Measuring

[leonski]

Lots of activity on the DIRAC thread.    People set up microphones in various positions.....I hope following whatever the instructions say to do.   Than the problems start.   Thread is now nearing 150 pages!     All sorts of issues with outcome.  

Some have wonderful luck and others don't seem to be able to get it right, or have a mix of good and bad results.     Some question still about SOFTWARE version and good V bad.     I have no idea.   

But I do know about measuring and measurement systems.     For example?  Any tool used to measure, in this case a microphone with curve will have Variation and a range of output for the same input.    This means if you 

set the mic up and run the curve half a dozen times, you will get a range of output.    This is NORMAL.    If I gave you a simple ruler, calibrated to 1/64 of an inch and something to measure?    I'm sure that you'd get different answers and if I gave 10 pieces to measure 

to 3 people and had them measure the pieces 3x each, I'd get some very interesting data.    All sorts of possibilities.

Point?   When measureing the room response with a system like DIRAC you have several variablies NOT under the control of the guys that came up with the system in the first place.     

Now?   Keep in mind that I haven't read the instructions.   But this much is true.     Things like number of points measured matter.    More points would intuitively seem to be better with larger spaces needing more points.....

HEIGHT of microphone is important.   Sound not only bounces the 'long way' in the room, but Up /Down and also with multiple walls.....Front / Side / Back is one possibility.  

If you can do an IMPULSE measurement?      Keep in mind that sounds which occur too closely in time to one another may sound like 'one' sound.     I think the limit is around 10ms.    That's why the first reflection points are
important.     Get out your tape measure and you'll see that from where you sit to first reflection point to the speak MAY add 5 feet or more to the distance.   Potential smear in the time domain.  

But bottom line?     When doing your DIRAC and other room measures KEEP CAREFUL TRACK of everything.    I'd make a 'map' of the room showing dimensions and locaiton of major furniture pieces.     Make it TO SCALE.

Make 20 copies...and keep the computer file for later IF you move anything around or make more measurments than the number of copies you made.....OUCH!

Keep track of each measurement location.    Measure to nearest wall, maybe, Side and Front or Back.  Record HEIGHT of mic and direction it faced.    What do the instructions say about Tripods?   Does the mic have a 1/4-20 socket
which is standard on tripods?  

Make all your measurements THAN attach whatever output you can from the software.      Add some listening notes.    Image?    Extension of bass?   Clarity?    Other?
A good system and care will go a long way to making it easy and fairly repeatable.   Variation is NORMAL in these kinds of situations.   



All systems have inbuilt errors with the measuring tool, the sofware and the Operator all interacting.     I've seen this many times.    A trained operator on a fairly simple machine......Install the sample and focus/ press measure and
STILL getting statistically different results from his or her partner.    I kept my OWN samples which I used from year to year for this particular test.  

[ttocs]

Tolerance in measuring is something that goes missing too much, so I totally agree with you that you can get a different outcome from multiple people measuring the exact same thing.

"I put the mic in the exact same spot every time" can mean something different when others say it vs when "I" say it. To get the mic in the "same" spot for either testing or Dirac First Mic Point, I use a laser measuring device which is accurate to within 1/16" in 100", get the mic to be within 1/8" measured to each Front Speaker using the exact mirror-image spot on both, and at the same height using marked locations in the room.

Then there are the multiple settings involved, like the processor volume, REW output level, whether an HDMI output is used vs an outboard audio interface with its own volume level input and output, etc. It all matters.

The tools at our disposal can sometimes be in want, meaning, they might not do everything in the manner that's helpful to getting the type of measurements needed. Some tools however do almost everything needed. To that end I, just tonight, loaded up an app for the ATV4K which I think is a must have. The Studio Six Digital ATMOS Surround Signal Generator. It'll measure 9.1.6, and if you're like me and only have 9.1.4 then what happens is the Right Top Front and Right Top Rear both play for matrixing the Right Top Middle. In usage, when initiating a test sound it waits for the processor to change to the correct audio mode being tested - then it plays the sound, very nice! Thanks marcl for the suggestion.

[leonski]

Thank YOU for a good response. That you go to so much trouble to get Proper and Repeatable results is something others should take to heart.

My first experience with trying to calibrate a system was using an Audio Control Equalizer which came with a mic. I think it was 6 band, something like 30/ 60 / 120/ 250 and 2 HF bands....like 6k and 12k.
This was a LONG time ago, so my memory may be faulty.
You'd play a 'warble tone' and adjust. One thing I know is that my 12" 3-way studio monitors.......VERY Good copy of JBL 4311 had an expected mid-bass hump which equalized out fairly well but nothing could help lack of low bass...
but than my knowledge base at that time was very limited.

I'm well beyond wanting a calibrated stereo. I spend more time worrying about my computer monitor, proper grey scale and color space.

[marcl]

Feb 3, 2021 22:01:48 GMT -5 ttocs said:

Tolerance in measuring is something that goes missing too much, so I totally agree with you that you can get a different outcome from multiple people measuring the exact same thing.

"I put the mic in the exact same spot every time" can mean something different when others say it vs when "I" say it. To get the mic in the "same" spot for either testing or Dirac First Mic Point, I use a laser measuring device which is accurate to within 1/16" in 100", get the mic to be within 1/8" measured to each Front Speaker using the exact mirror-image spot on both, and at the same height using marked locations in the room.

Then there are the multiple settings involved, like the processor volume, REW output level, whether an HDMI output is used vs an outboard audio interface with its own volume level input and output, etc. It all matters.

The tools at our disposal can sometimes be in want, meaning, they might not do everything in the manner that's helpful to getting the type of measurements needed. Some tools however do almost everything needed. To that end I, just tonight, loaded up an app for the ATV4K which I think is a must have. The Studio Six Digital ATMOS Surround Signal Generator. It'll measure 9.1.6, and if you're like me and only have 9.1.4 then what happens is the Right Top Front and Right Top Rear both play for matrixing the Right Top Middle. In usage, when initiating a test sound it waits for the processor to change to the correct audio mode being tested - then it plays the sound, very nice! Thanks marcl for the suggestion.

View Attachment

This is the only way I've found so far to measure frequency response of Atmos channels (i.e. channels other than 7.1).   To do this, Use the RTA in REW and check the Peak Hold.  Then in Surround Signal Generator use the 7s Sweep or the LARSA 0.5s for each channel.  REW will capture the peak response across the full spectrum.  It's not as precise as the regular REW sweep with all the other features, but it allows you to see that your corrections  and crossovers on the Atmos speakers are basically working.

Also ... beware, ttocs and I have found the Sub channel in this app measures +10db higher than it should.  It seems either the app or ATV4K is adding the +10db to the LFE and then our processor is adding it again.

[donh50]

Measuring sound waves can be tricky and there are a lot of variables folk might not think about:

• Small changes in mic placement matter (at 1 kHz the wavelength is about 1' but changes can be heard at a fraction of a wavelength; 1/8th wavelength is one rule of thumb and that is about 1.5"). At the very least get a boom mic stand and you can mark placement with tape on the floor and the stand.
• Similarly small changes in the room can changes the readings, from moving (or removing) a pillow on the MLP couch, to adding a picture on the wall, moving a lamp, etc. One common issue is a coffee table in front of the MLP; changing what (or where) you have on it can change the response.
• Where you stand whilst making measurements can have a significant impact. I usually leave the room.
• External noise, particularly in the bass, can be insidious depending upon how well the room is isolated. Furnace or AC fan running or not, nearby appliances can couple sound, lights can emit audible noise, etc. My room uses a mini-split for HVAC so no ducts to the rest of the house, and I turn it off when measuring (along with the lights -- as I leave the room). Low-frequency rumbles may not really be noticed by you but the mic hears all. For my room, I could not completely float the ceiling, so heavy footfalls above will transmit and corrupt my calibrations. I usually wait until the house is empty, or people are settled and such.
• What you set the mic on matters. A mic stand is best; sitting it on the back of the couch or chair is bad, and holding it is a sure way to invalidate the measurements.

Etc.

Anyone else remember the Soundcraftsman "warble" record for setting up their equalizer?

[leonski]

ALL good points from someone who has apparently been thru the loop of measurement.

I'd add 2 things.

I used a warble tone years ago that was a built in feature of an Audio Control 5 or 6 band equallizer. 4 bass bands and 2 HF.....

And as for mic placement? I'd try to figure a way to isolate the mic / suspend the mic. While I have a VERY sturdy (think car stand grade) photo tripod by Bogen, it is rigid and using as a mic stand, I'd suspect a resonance.

[marcl]

Feb 6, 2021 0:42:56 GMT -5 donh50 said:

Measuring sound waves can be tricky and there are a lot of variables folk might not think about:

• Small changes in mic placement matter (at 1 kHz the wavelength is about 1' but changes can be heard at a fraction of a wavelength; 1/8th wavelength is one rule of thumb and that is about 1.5"). At the very least get a boom mic stand and you can mark placement with tape on the floor and the stand.
• Similarly small changes in the room can changes the readings, from moving (or removing) a pillow on the MLP couch, to adding a picture on the wall, moving a lamp, etc. One common issue is a coffee table in front of the MLP; changing what (or where) you have on it can change the response.
• Where you stand whilst making measurements can have a significant impact. I usually leave the room.
• External noise, particularly in the bass, can be insidious depending upon how well the room is isolated. Furnace or AC fan running or not, nearby appliances can couple sound, lights can emit audible noise, etc. My room uses a mini-split for HVAC so no ducts to the rest of the house, and I turn it off when measuring (along with the lights -- as I leave the room). Low-frequency rumbles may not really be noticed by you but the mic hears all. For my room, I could not completely float the ceiling, so heavy footfalls above will transmit and corrupt my calibrations. I usually wait until the house is empty, or people are settled and such.
• What you set the mic on matters. A mic stand is best; sitting it on the back of the couch or chair is bad, and holding it is a sure way to invalidate the measurements.

Etc.

Anyone else remember the Soundcraftsman "warble" record for setting up their equalizer?

I went from a low back sectional couch with soft fabric, to a high back leather couch with two recliner sections.  BIG difference!  So I settled on reclining the two MLP seats to get them lower and away from the mic positions, and I toss a blanket over the back and a weighted blanket on the seat.  I refer to those as simulating the "Marc and Elise transfer function".  I go ahead and do my post-Dirac measurements with the blankets in place and mic at the MLP ... just to see if there are any Dirac-related issues ... fully aware (and proven) that when I remove the blankets and raise the seats the response will change.   What I observe is that the change is most evident with the center channel, and not so much with the L/R.

The coffee table has a piece of foam on it, and of course a copy of Toole's book

The bottom line on measurements is consistency.  I now have a very long boom stand that can reach all the measurement positions without moving the base, which sits on carpet/concrete.  You may have room related anomalies, but if you measure consistently as you make small changes, the data can still lead you to a solution that improves sound in normal use.

[megash0n]

Feb 6, 2021 9:49:14 GMT -5 marcl said:

Feb 6, 2021 0:42:56 GMT -5 donh50 said:

Measuring sound waves can be tricky and there are a lot of variables folk might not think about:

• Small changes in mic placement matter (at 1 kHz the wavelength is about 1' but changes can be heard at a fraction of a wavelength; 1/8th wavelength is one rule of thumb and that is about 1.5"). At the very least get a boom mic stand and you can mark placement with tape on the floor and the stand.
• Similarly small changes in the room can changes the readings, from moving (or removing) a pillow on the MLP couch, to adding a picture on the wall, moving a lamp, etc. One common issue is a coffee table in front of the MLP; changing what (or where) you have on it can change the response.
• Where you stand whilst making measurements can have a significant impact. I usually leave the room.
• External noise, particularly in the bass, can be insidious depending upon how well the room is isolated. Furnace or AC fan running or not, nearby appliances can couple sound, lights can emit audible noise, etc. My room uses a mini-split for HVAC so no ducts to the rest of the house, and I turn it off when measuring (along with the lights -- as I leave the room). Low-frequency rumbles may not really be noticed by you but the mic hears all. For my room, I could not completely float the ceiling, so heavy footfalls above will transmit and corrupt my calibrations. I usually wait until the house is empty, or people are settled and such.
• What you set the mic on matters. A mic stand is best; sitting it on the back of the couch or chair is bad, and holding it is a sure way to invalidate the measurements.

Etc.

Anyone else remember the Soundcraftsman "warble" record for setting up their equalizer?

I went from a low back sectional couch with soft fabric, to a high back leather couch with two recliner sections.  BIG difference!  So I settled on reclining the two MLP seats to get them lower and away from the mic positions, and I toss a blanket over the back and a weighted blanket on the seat.  I refer to those as simulating the "Marc and Elise transfer function".  I go ahead and do my post-Dirac measurements with the blankets in place and mic at the MLP ... just to see if there are any Dirac-related issues ... fully aware (and proven) that when I remove the blankets and raise the seats the response will change.   What I observe is that the change is most evident with the center channel, and not so much with the L/R.

The coffee table has a piece of foam on it, and of course a copy of Toole's book

The bottom line on measurements is consistency.  I now have a very long boom stand that can reach all the measurement positions without moving the base, which sits on carpet/concrete.  You may have room related anomalies, but if you measure consistently as you make small changes, the data can still lead you to a solution that improves sound in normal use.

Curious more than anything, but when you compared the "difference between center & LR" above, where the (center) tweeters, or mids/tweeters at the same height as the LR? Sounds like an axis or dispersion difference based on the location of the source sound. I could totally be way off here as well. Just a thought. ☺

[marcl]

Feb 6, 2021 10:14:01 GMT -5 megash0n said:

Feb 6, 2021 9:49:14 GMT -5 marcl said:

I went from a low back sectional couch with soft fabric, to a high back leather couch with two recliner sections.  BIG difference!  So I settled on reclining the two MLP seats to get them lower and away from the mic positions, and I toss a blanket over the back and a weighted blanket on the seat.  I refer to those as simulating the "Marc and Elise transfer function".  I go ahead and do my post-Dirac measurements with the blankets in place and mic at the MLP ... just to see if there are any Dirac-related issues ... fully aware (and proven) that when I remove the blankets and raise the seats the response will change.   What I observe is that the change is most evident with the center channel, and not so much with the L/R.

The coffee table has a piece of foam on it, and of course a copy of Toole's book

The bottom line on measurements is consistency.  I now have a very long boom stand that can reach all the measurement positions without moving the base, which sits on carpet/concrete.  You may have room related anomalies, but if you measure consistently as you make small changes, the data can still lead you to a solution that improves sound in normal use.

Curious more than anything, but when you compared the "difference between center & LR" above, where the (center) tweeters, or mids/tweeters at the same height as the LR? Sounds like an axis or dispersion difference based on the location of the source sound. I could totally be way off here as well. Just a thought. ☺

It's way more complicated.  The L/R have 5ft long ribbon tweeters standing vertically, but bouncing off the side walls at a fairly steep angle toward the MLP.  The center is also a Magnepan dipole, but the voice coil is horizontal and only two feet wide, and "quasi-ribbon", not ribbon.   

What I think is happening is the center speaker plays directly at the couch and the sound bounces off and hits the mic with a very short delay.  But it also goes past the mic to the back of the room, reflects and returns.  I can see that reflection in the ETC around 22ms ... the time for the round trip to the back of the room!  Those reflections don't cause a big problem ... just some ups and downs around 300-500Hz.

[monkumonku]

Feb 6, 2021 0:42:56 GMT -5 donh50 said:

Measuring sound waves can be tricky and there are a lot of variables folk might not think about:

• Small changes in mic placement matter (at 1 kHz the wavelength is about 1' but changes can be heard at a fraction of a wavelength; 1/8th wavelength is one rule of thumb and that is about 1.5"). At the very least get a boom mic stand and you can mark placement with tape on the floor and the stand.
• Similarly small changes in the room can changes the readings, from moving (or removing) a pillow on the MLP couch, to adding a picture on the wall, moving a lamp, etc. One common issue is a coffee table in front of the MLP; changing what (or where) you have on it can change the response.
• Where you stand whilst making measurements can have a significant impact. I usually leave the room.
• External noise, particularly in the bass, can be insidious depending upon how well the room is isolated. Furnace or AC fan running or not, nearby appliances can couple sound, lights can emit audible noise, etc. My room uses a mini-split for HVAC so no ducts to the rest of the house, and I turn it off when measuring (along with the lights -- as I leave the room). Low-frequency rumbles may not really be noticed by you but the mic hears all. For my room, I could not completely float the ceiling, so heavy footfalls above will transmit and corrupt my calibrations. I usually wait until the house is empty, or people are settled and such.
• What you set the mic on matters. A mic stand is best; sitting it on the back of the couch or chair is bad, and holding it is a sure way to invalidate the measurements.

Etc.

Anyone else remember the Soundcraftsman "warble" record for setting up their equalizer?

As you said, there are a lot of variables influencing measurements. Many of these variables change all the time, so how can you ever get a really valid measurement? I guess the answer is most of them make only small differences.

I realize that where you stand while taking measurements has an effect, but if you would normally be listening in the room to music, why would you leave the room when taking measurements? Wouldn't it be better to sit in your normal or preferred place?

[donh50]

Feb 6, 2021 11:29:32 GMT -5 monkumonku said:

As you said, there are a lot of variables influencing measurements. Many of these variables change all the time, so how can you ever get a really valid measurement? I guess the answer is most of them make only small differences.

I realize that where you stand while taking measurements has an effect, but if you would normally be listening in the room to music, why would you leave the room when taking measurements? Wouldn't it be better to sit in your normal or preferred place?

My media room is dedicated so I don't have constant changes. As to differences being small or large, that is one of the variables. Some may be small but others very large. Much depends upon your speakers and how they radiate and interact with the room. And of course room treatments (or lack of) are a major effect. Then there is what "small" means to different people...

Changing the carpet changes floor bounce. Adding or removing pillows from the couch can change local reflections. Moving a floor plant, shifting a chair, adding or removing window coverings, etc. can all have an effect. As marcl said it is all about consistency. Small things (physically small) usually chane higher frequencies and may be less noticeable. Of course, if you change something significant (move things around, add or take away furniture, etc.), it may be worthwhile re-running the calibration.

Room modes are largely a function of the room's size, but things can change there too. Adding or removing bookshelves can change the volume. Opening or closing a door or window. Etc.

My previous speakers were planar dipoles (Magnepans) that did not radiate significantly to the sides or top and bottom, but they did have a large reverse wave, so I treated heavily behind them to reduce comb filter effects at the MLP. I did not consider those small, they drove me crazy (a short drive), but others might not notice until I pointed it out.

The microphone is placed where I normally sit, at ear level, per the recommendation of virtually every room analysis and/or correction program I have ever used. The idea is to measure the response where I would be, without me in the way. Since the mic will pick up the effects of where I stand, thus reflections and absorptions due to my body and clothing, that is a huge variable I choose to eliminate.

Yes, there are many variables, some may change often, and the effect (and effect on the listener) varies. I minimize what I can, strive to make my measurements as consistent as possible, and hope for the best. I know what things bug me based upon decades of listening and measuring, and try to fix those, but by and large we (or at least I) am pretty tolerant of errors -- at least until I compare to something better, then I want that! I've found I'll tolerate a lot without knowing how bad it is until I fix it and discover how much better it can be. A statement that applies to things far beyond the listening room...

IME/IMO/FWIWFM/my 0.000001 cent (microcent) - Don

[leonski]

Feb 6, 2021 10:14:01 GMT -5 megash0n said:

Feb 6, 2021 9:49:14 GMT -5 marcl said:

I went from a low back sectional couch with soft fabric, to a high back leather couch with two recliner sections.  BIG difference!  So I settled on reclining the two MLP seats to get them lower and away from the mic positions, and I toss a blanket over the back and a weighted blanket on the seat.  I refer to those as simulating the "Marc and Elise transfer function".  I go ahead and do my post-Dirac measurements with the blankets in place and mic at the MLP ... just to see if there are any Dirac-related issues ... fully aware (and proven) that when I remove the blankets and raise the seats the response will change.   What I observe is that the change is most evident with the center channel, and not so much with the L/R.

The coffee table has a piece of foam on it, and of course a copy of Toole's book

The bottom line on measurements is consistency.  I now have a very long boom stand that can reach all the measurement positions without moving the base, which sits on carpet/concrete.  You may have room related anomalies, but if you measure consistently as you make small changes, the data can still lead you to a solution that improves sound in normal use.

Curious more than anything, but when you compared the "difference between center & LR" above, where the (center) tweeters, or mids/tweeters at the same height as the LR? Sounds like an axis or dispersion difference based on the location of the source sound. I could totally be way off here as well. Just a thought. ☺

Magnepan uses a RIBBON tweeter, considered by many to be among the finest of its type.   It is Vertical and long.    Not like a small cone or dome....or even one of the small EMIT tweeters......which are more representational of a point source.   
Magnepan will even sell you replacement tweeters, since the ribbon has a reputation as a fuse protector.    Easy DIY swap....and a core charge!        
Maggies are what is called a Line Sourde and being dipoles, interact with room somewhat differently than cone speaker-guys are accustomed to.   

The stuff that marc noted about furniture / backs and material should be common to all measurements and rooms.    Leather reflects.   An open weave cloth may absorb more.  
I'm just not sure about the efficacy of modifying the room FOR measurements, than swapping back for listening / use.

[marcl]

Feb 6, 2021 14:18:00 GMT -5 leonski said:

Feb 6, 2021 10:14:01 GMT -5 megash0n said:

Curious more than anything, but when you compared the "difference between center & LR" above, where the (center) tweeters, or mids/tweeters at the same height as the LR? Sounds like an axis or dispersion difference based on the location of the source sound. I could totally be way off here as well. Just a thought. ☺

Magnepan uses a RIBBON tweeter, considered by many to be among the finest of its type.   It is Vertical and long.    Not like a small cone or dome....or even one of the small EMIT tweeters......which are more representational of a point source.   
Magnepan will even sell you replacement tweeters, since the ribbon has a reputation as a fuse protector.    Easy DIY swap....and a core charge!        
Maggies are what is called a Line Sourde and being dipoles, interact with room somewhat differently than cone speaker-guys are accustomed to.   

The stuff that marc noted about furniture / backs and material should be common to all measurements and rooms.    Leather reflects.   An open weave cloth may absorb more.  
I'm just not sure about the efficacy of modifying the room FOR measurements, than swapping back for listening / use.

To the last point .... Dirac uses the first set of measurements to get distance and time alignment.  I add my baffles only for that measurement.  Then I take 8 more without the baffles.  I have found - through experiment and post calibration measurement - that doing it this way ensures accurate time alignment, while the Dirac algorithm for combining all the measurements results in proper frequency response compensation.  I wasn't sure it would work ... but I tried several variations and found a process that does work.

[leonski]

What you have is what I call an Artifact. It would be difficult or maybe impossible to transfer what you learned to others. And few will go to that kind of trouble.....

My question remains, but could probably be answered by watching a movie or listening......

What is a good reference movie? Any fans of Burce Willis in 'The Fifth Element'? Or even a good copy of 'Forbidden Planet'? I'd even settle for one of the horse race scenes from 'Seabiscuit'.....

Several other entries in the 'good sound' and effects sweepstakes. You choose!

[ttocs]

Bladerunner has the best ATMOS soundtrack!

[leonski]

Feb 6, 2021 16:38:45 GMT -5 ttocs said:

Bladerunner has the best ATMOS soundtrack!

original or 'part II'?    

[ttocs]

Feb 6, 2021 18:09:29 GMT -5 leonski said:

Feb 6, 2021 16:38:45 GMT -5 ttocs said:

Bladerunner has the best ATMOS soundtrack!

original or 'part II'?    

The best ATMOS immersive audio I know of. The rain storm with thunder is absolutely real!

[leonski]

I never realized how many 'versions' of the original there are. I count 7, but maybe off either way.

[megash0n]

Feb 6, 2021 10:26:44 GMT -5 marcl said:

Feb 6, 2021 10:14:01 GMT -5 megash0n said:

Curious more than anything, but when you compared the "difference between center & LR" above, where the (center) tweeters, or mids/tweeters at the same height as the LR? Sounds like an axis or dispersion difference based on the location of the source sound. I could totally be way off here as well. Just a thought. ☺

It's way more complicated.  The L/R have 5ft long ribbon tweeters standing vertically, but bouncing off the side walls at a fairly steep angle toward the MLP.  The center is also a Magnepan dipole, but the voice coil is horizontal and only two feet wide, and "quasi-ribbon", not ribbon.   

What I think is happening is the center speaker plays directly at the couch and the sound bounces off and hits the mic with a very short delay.  But it also goes past the mic to the back of the room, reflects and returns.  I can see that reflection in the ETC around 22ms ... the time for the round trip to the back of the room!  Those reflections don't cause a big problem ... just some ups and downs around 300-500Hz.

View Attachment

good detail. We're on the same page. I didn't quite dive all the way there in my thought, but what you depicted is in line with my general, 1 inch deep thought process about it.

[leonski]

line source / Maggies don't radiate much to the ceiling OR floor......or the sidewalls. They are fairly 'beamy' as frequency goes up. The ribbon probably doesn't interact with the side walls at more than 20db down from primary response.

Front / rear waves of such panels CANCEL basically at the plane of the panel edges. You can easily TEST this by moving your ear front / back and note the 'null' right at the panels edge. CLOSE the other ear, please!

Here is the STEREOPHILE measurement panel from the new LRS speakers. Others should behave in a similar manner. Magnepan hasn't doesn't anything 'big' in 40 years, but has instead
refined and improved.

www.stereophile.com/content/magnepan-lrs-loudspeaker-measurements