How to find cause room response problem

[Sliderule]

I have attached a picture of my system and 4 REW screen shots. Left and right speakers frequency and phase response (203 & 204) and group delay (201 & 202). The problem is the suckout about 200Hz. I had the same suckout with my previous speakers. The speakers are Soundlab 745's full range elctrostats. The base room is 12' ceiling 17' long by 23" wide. It is quite irregular with hallways, entry way, dining room and kitchen in common. All four graphs are with 1/12 smoothing. Center of room has carpeting with hardwood floors exposed elsewhere. The group delay peaks match the suckout dip. The phase reversals also match the suckout. I have reviewed the REW help files on Group Delay but did not gain any insight into the cause of the delay peaks. Any help appreciated.

 

[VinceHoffman]

How far are your 'stats placed off of the wall behind them? Your pic suggests they are quite close...

 

[Sliderule]

They are 3'. As much as I would like to place them out at 5' the waf (wife acceptance factor) would fall drastically. I am well aware I will have to treat the wall behind the stats. The left speaker, because it is near a sidewall. The group delay between the left and right speakers is inverted.

 

[VinceHoffman]

I would strongly suspect your cancellation notch is due to SBIR (Speaker Boundary Interference Response) related to the back wave reflection off the front wall. The distance traveled by the strong reflection when it arrives back at the loudspeaker will cause a cancellation notch at the frequency corresponding with one half wavelength of the reflection distance (as the reflection arrives back at the loudspeaker 180 degrees out of phase). As for the nearby left sidewall, it isn't likely to be an issue as your full range electrostat speakers along open baffle loudspeaker designs have a dipole radiation pattern. Dipole speakers manifest a null in their output immediately to the sides- so no SBIR issues with room boundaries to the sides of dipole speakers. An effective treatment solution would be treat that rear wall reflection. I would reach out to the ASC Tube Trap folks at info@acousticsciences.com as I suspect a 6' tall stack of 13" Tube Traps immediately behind each 'stat would be a great solution.

Unlike your dipole Soundlab 'stats, my conventional (monopole) loudspeakers radiate low frequencies omnidirectionally. So all nearby room boundaries can cause SBIR problems. With my setup the L & R loudspeakers are 3 feet from the sidewalls which caused a significant frequency response deviation comparable to yours. Here is a pic of the stack of 13" ASC Tube Traps immediately to either side of my mains which almost completely fill in that dip by absorbing the problematic reflection.

​

 

[Sliderule]

I had to look up SBIR (Speaker Boundary Interference. It usually refers to to non dipole 1/4 wave interference. I realize that there is comb interference with dipoles. As the right speaker is a long way from a side wall it has a longer path for the for the back wave to reach the listening position. If you look at the group delay for the right it is negative mS while the left is positive mS. Also the imaging from center to irght is detailed and solid while the left is mediocre an not continuous. I have not found the reason for the suckout yet. I have moved the mic to see if it affected the back wall or the divider reflections with no change. I also used a room modal analysis program https://www.bobgolds.com/Mode/RoomModes.htm My room is irregular enough to have no resonance issues even below 100Hz. I had this same 200Hz issue with a hybrid system and had assumed it was due to phase issues at the 200Hz crossover point.

 

[VinceHoffman]

Hey @Sliderule how about we arrange a quick phone call (assuming you are in North America) or a Facetime chat and I'll clarify exactly why your 200Hz-ish suckout problem is that strong reflection from the wall immediately behind your speakers...

 

[Sliderule]

Hi Vince, I am in South Carolina. I did some tests moving the mic ahead of the listening position so I could determine if reflections from the 1/2 wall or the back wall (17') were affecting the measurement. It is obvious that the reflected and the direct signal are interfering at the listening position and 2' ahead of it. If you look at the screen shots of the left and right group delay you will see that the left is positive delay and the right is negative delay. I am going to test the near field response and move away from the speakers until I can see the effects of the reflected sounds. Stereo imaging is another way to detect early reflections. The right speaker has excellent imaging due to the delayed reflection time while the left has gaps in the image. I'll get back to you in a day or two.

-david

 

[Mike-48]

With a speaker 3 ft from a wall, the expected null due to SBIR would be about 94 Hz. SBIR at 190 Hz would be from a wall about 18" from the woofers.

These suckouts are not always easy to attribute to particular dimensions, in my experience. I wish you good luck getting this one squared away.

 

[VinceHoffman]

With a speaker 3 ft from a wall, the expected null due to SBIR would be about 94 Hz. SBIR at 190 Hz would be from a wall about 18" from the woofers.

These suckouts are not always easy to attribute to particular dimensions, in my experience. I wish you good luck getting this one squared away.

Not in this case @Mike-48. Your calculation above assumes a monopole lodspeaker design with an essentially omnidirectional radiation pattern at low frequencies. In the OP's case we are dealing with a dipole loudspeaker where its rear radiation is 180 degrees out of phase with its forward radiation. So the reflection of that out of phase rear wave will arrive back at the loudspeaker still perfectly out of phase for a wavelength in the vicinity of 6ft long or 188Hz. Too much of a coincidence for this reflection not to be the causal factor of the suckout that @Sliderule has measured.

This is one of the reasons why best practice speaker positioning for full range dipole loudspeakers has always been "way out from the front wall, the further the better". 5+ feet ideally. If near-front wall speaker placement is a necessity, a full range dipole isn't all that great a choice. If you choose to do so anyway, you will need to employ acoustically absorptive treatment behind the dipole to address the now problematic strong reflection of the rear wave.

 

[Sliderule]

One of the things I did today was to make frequency sweeps with the mic @1ft, 4ft, and 8 ft from each speaker. I have the attached the screenshots 207 is 1ft and 208 is 4 ft. Note the 8 ft distance looks the same as the 4 ft. The 4 ft phase reversals show the reflection effects on the signal. Note that the smoothing was 1/48 octave. As I mentioned earlier I will try moving out the speakers out to 5' out to remeasure the speakers and check for stereo imaging. My speakers have 13 rows of cells and 7 vertical columns. The speaker is segmented in 3 7.5° steps (45° in total) from the center column. I did measure the speakers parallel to the back wall (not recommended) and 15° toe-in and 22.5° toe-in. I also intend to try toe-in of 25° tp avoid having any of the columns parallel to the back wall. This will keep the focal point of both speakers slightly behind the listening position. As the 4' response is fairly level between 40Hz and 100Hz and there no room nodes I am not sure about the best way to attenuate back wave reflections.

 

[VinceHoffman]

...I am not sure about the best way to attenuate back wave reflections.

As I previously suggested @Sliderule, reach out to Acoustic Sciences (Corporation) for some good advice and a possible solution. Art Noxon, ASC's founder built the very first Tube Trap prototypes with the specific intent to absorb the back wave of full range Magneplanar dipoles way back in the early '80s.

 

[skid00]

I use memory foam mattress material behind my Carver Amazing driver. It is, at the very least, a very cheap way to test.

 

[VinceHoffman]

I have an even cheaper proof of concept @skid00 & @Sliderule.

-Visit your local Home Depot or similar builders supply store and purchase one or the other of what I've pictured below.
-Keep your receipt! You'll need it later.
-When you get home DO NOT OPEN THE PACKAGES. Simply stack the sealed packages against the front wall immediately behind your loudspeakers as I've directed on each of the attached pics.
-While this is not the optimal way to use fluffy pink fiberglass to DIY bass traps, it works surprisingly well- Well enough that if you perform another REW measurement run, the reflection based null should be significantly reduced.
-Great, return the bales of insulation to the store for a full refund.
-Now you've proven where the problem lies and how to solve it at close to zero net cost (just some time and gas).
-Finally purchase or DIY the suitable broadband absorptive traps to do the job optimally.

In @Sliderule's case, I'd recommend a trap which offers an absorption coefficient of 1.0 or higher all the way down to 150Hz or lower. The 13" diameter stacks of ASC Tube Traps I previously suggested will do this handily.

 

[Sliderule]

Hey, some great suggestions. I have been making measurements and doing listening tests. By moving the speakers out 5' from the back wall and running REW tests the measurements stay the same but the stereo imaging on the left speaker fills in and becomes solid. Due to irregular back wall behind right speaker there is a longer reflection path that allows a solid image. The 10ms back wave delay wiil do the trick, However, as big as the speakers are 5' out from the back wall is overwhelming in the room. Imaging quality has become the best way for me to evaluate speaker performance. I have reached out to Acoustic Sciences tech support for their recommendations and am awaiting them. I will try the fibreglass rolls on the left speaker at 3 ft back spacing and check the imaging.