|
|
Post by bjmsam on Feb 20, 2022 23:49:29 GMT
I am struggling to eliminate just such an audible change in response when moving my head laterally at the MLP even when L and R are isolated. I will do my best to describe the situation and would appreciate recommendations!
Room configuration:
The side and rear reflections seem relatively insignificant:
Curtains cover sliding glass doors and windows. The corresponding primary reflection is -16dB at 22.9ms for left channel only. Wool rugs hang on the rear wall. The corresponding primary reflection is -19dB at 23.6ms for both channels.
The floor is carpeted wall to wall and an additional thick pile shag rug extends the width of the screen back to the MLP.
I used REW to identify a primary reflection at the ceiling of -7.5dB and applied absorption to mitigate.
Physical measurements:
50” from ceiling to between AR9 tweeter and upper-midrange dome
50” from ceiling to tip of Audyssey microphone
180” from between drivers to tip of microphone
206" - 180" = 26" = 2' 2" (within ~1" of impulse delay due to rounding)
Regular black "acoustic" ceiling tiles (.55 NRC) replaced with black Cinetiles (.95 NRC) at positions shown in darker shade.
Measurement confirms that treatment was effective (primary reflection reduced from -7.5dB to -18dB):
Sadly, the comb filtering effect remains annoyingly audible (mic positions before / after not precisely equal).
Could the very early impulse response below at ~5 inches and ~10dB cause such significant impact? All others within 20ms are -20dB or less.
Both measurements are with the speaker grilles removed, so perhaps the headrest of my fabric chair is more reflective than assumed.
Any other suggestions? Thank you!
|
|
|
|
Post by Hexspa on Feb 21, 2022 11:28:06 GMT
Home theater seems a little different than the kind of nearfield listening that I do. Still, the same principles should apply. Your speakers are so far away so I imagine that the room reflections will be a bigger part of the sound than if they were closer. If you want to minimize interference, you have to make your surfaces more absorbent. 2" foam is a decent start but 4-8" rigid insulation is better. Try to find your first/early reflection points and make sure they're absorbent. Also, make sure your chairs aren't causing interference as Ethan's were; deaden the couch (particularly behind your head) with blankets.
|
|
|
|
Post by bjmsam on Feb 21, 2022 13:00:30 GMT
Thanks for your reply. That 2" foam behind the AT screen covers 4" Roxul AFB.
I will tackle that early reflection at the MLP in hopes it is the culprit.
|
|
|
|
Post by Ethan Winer on Feb 21, 2022 17:49:42 GMT
I am struggling to eliminate just such an audible change in response when moving my head laterally at the MLP even when L and R are isolated. I will do my best to describe the situation and would appreciate recommendations! Your analysis is certainly thorough! And so is your treatment. To be clear, the problem is the left channel only, yes? At -16 dB that's not terrible, and shouldn't be a huge problem audibly. It looks like disabling Audyssey helps. Another thing that should help is angling your speakers to face the prime spot in the center of the couch. Speakers are always more flat on-axis, and that also sends less of the sound off to the sides. Also, does your couch back extend higher than your shoulders? If it does, raise the measuring microphone and see if that helps. |
|
|
|
Post by bjmsam on Feb 21, 2022 18:29:41 GMT
Thank you for weighing in!
The comb filtering is audible for both channels, when playing alone and when playing together, whether or not Audyssey is engaged. For all measurements above, XT32 was in FLAT mode with Dynamic EQ set to 10dB Reference Level Offset (green trace below), which sounds best to me.
I experimented with toe initially and was surprised that it seemed detrimental to frequency response but will try again to see if comb filtering is affected.
The fabric headrest does extend above my shoulders and now is the focus of my investigation this week of these very early reflections (virtually identical for L and R):
Thanks again!
|
|
|
|
Post by Ethan Winer on Feb 21, 2022 18:53:33 GMT
Okay, good, let me know what your experiments uncover.
|
|
|
|
Post by Hexspa on Feb 21, 2022 19:34:14 GMT
I didn't register that you had 4" rigid behind that wall. Sometimes people get too into the details so I was trying to look broad strokes and missed it.
Depending on how things go with your headrest, you can also consider diffusing more of the surrounding room. The EBU paper I keep mentioning seems to imply that high frequency diffusion over just a narrow band (4-8kHz iirc) is all that's needed. You seem technical enough to be able to figure out the diffuser specs to accomplish that. Diffusion should flatten the comb filtering.
|
|
|
|
Post by bjmsam on Feb 22, 2022 20:06:16 GMT
Thanks to you both for replying. I measured a single speaker on axis at 3', 4', 5', 6' to produce these graphs (animated for comparison):
I also used the mic on a tripod in place of the chair at the MLP (15' from each speaker) to measure both speakers with no toe and with toe in to produce these graphs (animated for comparison):
Diffusion might help if I had anywhere to put it (ceiling and floor are not options, side and rear walls are too distant).
|
|
|
|
Post by Hexspa on Feb 22, 2022 20:12:40 GMT
It seems like that 6' distance changes everything. Maybe you can try moving your speakers 1' or so closer. Back to general principles, it looks like your listening position is in the middle of the room. Have you tried moving the LP up toward the 38% mark?
Closer speakers plus a moved-up LP will definitely shift the balance of direct vs. room sounds. There's a concept called 'critical distance' and it's usually not very far. It refers to the distance at which the ambient level is equal to the direct one. Naturally with flat surfaces, you're going to have a lot of comb filtering once you're on the other side of that ratio.
I understand that your impulse plot shows close surfaces as being reflective. Maybe that's your table/ottoman (not sure what's in front of your chairs) and headrest. In any case, I've not yet used an impulse response graph to treat my room; maybe I'll have to try it.
|
|
|
|
Post by Ethan Winer on Feb 22, 2022 21:18:25 GMT
What Hexspa said.  |
|
|
|
Post by bjmsam on Feb 22, 2022 21:20:26 GMT
For these measurements I removed the chair leaving nothing between speaker and microphone.
It's interesting how the relative magnitude of that early impulse (consistently 455us / 6.15") increases with distance.
36" @ -12dB 48" @ -9.75dB 60" @ -7.50dB 72" @ -5.25dB
The same impulse response is relatively low (compared to those before and after it) at the MLP:
Maybe it's a standing wave. I'll try measuring again with a narrower frequency band.
Closer sounds better (presumably more about front geometry than rear distance) but does not work for three (sometimes four) people viewing the 120" screen. 'Tis a dilemma!
Thanks again.
|
|
|
|
Post by Hexspa on Feb 23, 2022 10:34:48 GMT
The last idea I'll throw out there, which I'm betting is DOA, is to orient the room 'up' or facing the apex. You could fill in that corner with fluffy, put your screen in front of it (which I assume is acoustically transparent), and redistribute the roxul as needed. I remember Ethan saying that you don't usually want to face a corner but, if it was my room, I'd definitely measure that layout before settling on anything.
|
|
|
|
Post by bjmsam on Apr 1, 2022 20:02:39 GMT
To help address the comb filtering effects caused by my phantom center setup (the "Toe L+R" chart above), I built a center speaker behind my screen that matches the top section of my left and right speakers ( project details on CSP).
It completely eliminated the audible change in response when moving my head laterally at the MLP, and I really don't seem to notice or be bothered by comb filtering effects over 1kHz from any individual channel when listening to content other than pink noise, so my focus now is on boundary-induced comb filtering under 1kHz as Ethan Winer described for lower frequencies in Do Room Modes Even Matter (i.e.- frequency proportional to quarter wavelength distance from the rear wall).
The traces below represent measurements taken with a calibrated UMIK-1 mic positioned on-axis at distances of 0.5m, 1.0m, 1.5m, 2.0m, and 4.25m from that C speaker in the front wall of my room (8.75m long with irregular sides as depicted in my first post). All C frequencies below 200Hz are routed to L+R. A few of my trend lines appear to be wrong, but it seems clear that, unlike Ethan's garage, my room appears to be dominated by BICF rather than modal response, correct? How does treatment differ?
Here is the spectrogram for a 4.25m trace:
Here is the corresponding SPL & Phase:
Here it is again with Audyssey XT32 Flat engaged:
I certainly would appreciate any insight you might offer, as a better understanding of what is going on will help inform a strategy for mitigation!
|
|
|
|
Post by Ethan Winer on Apr 2, 2022 16:12:01 GMT
This is a lot to take in, but I think the plot just below your photo above is more modal than boundary. Your dashed lines don't really follow the peaks and nulls correctly. For example, the bottom green null at 60 Hz is more likely the same as the orange null directly above. And the orange null at 40 Hz simply doesn't exist at the green distance. Modal peaks and nulls also change with position, so a peak at one location or frequency can be a null at another.
|
|
|
|
Post by Hexspa on Apr 6, 2022 13:56:28 GMT
If you want to figure out where each of your peaks and nulls are coming from, you can check out my Acoustic Analysis For Drunks pdf; it's on this site for free if you search [20230917 out of print]. I was analyzing people's responses here and eventually just wrote down how I was doing it. It seems to be accurate - no one has disputed it - so by all means work through it.
Maybe doing a home theater is different than a stereo home studio but if it's not then I think you might be way overthinking all of this. As far as I know, you just need two maybe three graphs: a smoothed broadband SPL response, an unsmoothed SPL response below 250Hz, and a decay plot showing initial response and 150ms.
SPL is going to have a lot to do with positioning and decay is a problem until you have sufficient absorption. Luckily you can kill two birds with each stone: every broadband absorber helps with bass and reflections; after all, they're really the same thing. Like I already mentioned, you've committed your arrangement to asymmetry - it's going to be hard to overcome that.
If you want a flatter response, you have to move your speakers closer or make your room more dead. All these issues you're observing are nothing more than the influence of your room being greater than that of the direct signal. Ethan has already summed it up for everyone in a small space: more absorption is better.
That's my understanding and it's worked ok for me so far. I hope that helps.
|
|
