Unexpected Audyssey measurement results

[Sunsanvil]

First, it is acknowledges that my room is very small and of unfavorable dimensions. It is however the only space I have. I've been in this room for over two decades and despite its imperfections have survived and enjoyed many a movie. Still, there is allways room (no pun intended) for better understanding. The basic room mode calc is as follows:

As you can see its almost (but not quite) a square. Speakers/TV are on one of the short walls, love seat opposite. Unfortunately, given the compact dimensions of the room, there is zero flexibility in the seating position and only minor latitude in speaker placement. The room has a good amount of Auralex studiofoam, 3", which performs fairly well down to around 300Hz (below that the NRC drops well below 0.5). There is a large opening to the side of the loveseat which leads to a more or less infinit space so I disregard the width data (for now).

Whats puzzling me is what Audessey (Denon AVR-X3800H using the iPad app) shows:

Of course what jumps out is that massive suck-out at just over 100Hz, which does correlate with what the spreadsheet is showing as the 2nd order mode in the length. Interesting is that the presumed corrected result still has a wedge in it, presumably because Audyssey wisely limits just how much boost it will apply (to protect the speaker). This dip is consistent for all three front speakers as well as the sub. All four are positioned such that their front baffle is right around 2.5' out from the wall (years ago using only my ears this produced the best overall result/compromise). Conversely our heads end up right around 2' out from the opposite wall. That puts both the speakers and our heads at or very close to the nulls of the 105Hz mode... which theoretically should mean that we dont experience the null at all... and yet it is the most prominent feature of Audessey's measurement. I'm just trying to understand why, and if maybe I should deviate from standard mic-position practice.

I've tried their standard array of 8 mic positions: 3 across the listener plane, 3 somewhat in front, and 2 behind. I also tried using only 6 in a tighter spacing: 3 a little in front of the listener plane, 3 just a hair behind (being mindful to never take a measurement closer than 20" to the back wall). The results, while varying slightly, all exhibit the same dramatic 100Hz dip though (which again one wouldn't expect given the spreadsheet data).

Appreciate any thoughts you might share.

 

[NotInYourHead]

Going off of the top of my head here. Small rooms tend to have huge issues with bass, and cubic/square-like rooms are the worst. You can check the 2nd tab of that Harman spreadsheet, where it shows you room mode distributions across the bass frequencies. For small rooms, room modes are few and far between and they tend to be very strong too.

You mentioned all three speakers, L/R and sub, exhibiting a similar dip. I think that's one of the main issues, because common FR peaks/dips can compound and become infinitely worse. You can (and should) place speakers for best stereo reproduction, and the sub for best bass. This can be done with a measurement mic.

Subwoofer settings should also be considered. Is the subwoofer at too low of a volume relative to the mains, or crossed over at a very low frequency? What about the phase switch? Is it being used correctly? A common strategy for 2.1 setups (and one used by Genelec/Neumann) is to high-pass the mains right before the dip and setup the subwoofer to basically fill it in, or something among those lines.

 

[Sunsanvil]

On the whole, post Audyssey (XT32), it actually sounds pretty good. Well, it sounded good even without (thanks in no small part to the the abundance of absorption). I just really want to understand why that one 105Hz mode is manifesting so strongly, disproportionately so I think. My understanding was that when a speaker is placed at the null of a mode, it will fail to excite that mode, but perhaps I've misunderstood how that plays out in-room.

 

[jcarys]

Have you verified that Audyssey is correct? You said it sounds good post, so it might be. You can run a frequency sweep without Audyssey, checking either by ear or another mic if you have one. If it's really down 15db at your listening spot, you're going to notice it. If it is correct, my first guess would be that the side space that you are skipping in your calculations is the culprit, or perhaps the acoustic foam treatment.

 

[Sunsanvil]

Just did some old fashioned sweeps with Audyssey off and there is definitely a suck-out around 100Hz, so thats real. Does sound better with Audyssey on, but still able to hear a bit of a dip at that frequency with the sweep (so, in this case at least, the Audyssey is actually in the ballpark). I know the Auralex foam wedges I have are pretty much benign at that frequency. I was looking at EcoCell NRC numbers and they are actually still close to a 1.0 NRC at 125Hz so I wonder if that would help. Auralex's Aurapanels are based on this stuff but are stupidly expensive (only dealer is Long & McQuaid). I'd make some DIY panels to try but cannot find anyone in Canada who'll sell a pack of batts. lol

 

[janbth]

On the whole, post Audyssey (XT32), it actually sounds pretty good. Well, it sounded good even without (thanks in no small part to the the abundance of absorption). I just really want to understand why that one 105Hz mode is manifesting so strongly, disproportionately so I think. My understanding was that when a speaker is placed at the null of a mode, it will fail to excite that mode, but perhaps I've misunderstood how that plays out in-room.

If you place a speaker at a null of a mode, it will lead to a dip in the response at that frequency. It will indeed fail to excite the mode, and that means there will be no audible sound at that frequency because adjacent frequencies respond normally. However, narrow dips are not necessarily an audible problem when playing music or movies. It seems that there is no problem in your case, because, as you say «it actually sounds pretty good»?

 

[kemmler3D]

One simple thing to try if you can get this connected to your system (a phone with Bluetooth works): https://www.szynalski.com/tone-generator/

Manually sweep the tone up from 50hz or so, up through 100hz and on to 200hz or so. Go slowly and see what you hear. If it seems to correspond to the graph, your measurements are subjectively confirmed. If it's obviously way different, you might need to measure differently.

 

[krabapple]

As per Toole 'Sound Reproduction" (Fourth Edition), p 438, you can deploy subwoofers to attenuate (make nulls less deep, and highs less high) room modes two ways. Assuming a rectangular room, and the main width modes as an example 1) Putting a sub at the width midpoint, e.g. where there is typically a first-order mode null, reduces the efficiacy of energy transfer to that null , makign it less energized than if put the sub elsewhere along the width. (see (a) ) 2) Attenuation can also be achieved using two subs (same polarity and phase) by destructive acoustical interference, i.e., putting one sub in the + lobe of the mode, and the other in the - lobe. The maximum effect of this is when each sub is placed at lobe maximum, typically the front corners. (see (b)) Also the effect can be modulated by adusting subwoofer levels (b.) This also attenuates the third order axial width mode (for the same reason, subs in lobes of opposite polarity).(c) The effect still works (to different degree) as long as the subs remain symmetrically somewhere within opposite lobes of a mode. Thus, if the two are placed at the two width nulls of second-order mode (d), you get attenuation of that mode (via mechanism 1) , as well as of the other two (mechanism 2). (d)

but NB if the modal frequency is ~100 Hz (which may be above the sub's output limit), repositioning loudspeakers could be needed instead. And this doesn't account for length and height modes.

 

[Sunsanvil]

It seems that there is no problem in your case, because, as you say «it actually sounds pretty good»?

Indeed, thanks I’m sure to Audyssey applying as much boost as it dares, which still falls short of its presumed in-room target. Given just how little wiggle room there is I think that the only thing to try is upping the absorption game to something which will have at least some effect down around the 100hz, but I wont loose sleep over it.

 

[Sunsanvil]

Thus, if the two are placed at the two width nulls of second-order mode (d), you get attenuation of that mode (via mechanism 1) , as well as of the
other two (mechanism 2). (d)

but NB if the modal frequency is ~100 Hz (which may be above the sub's output limit), repositionng loudspeakers would be needed isntead.

If I understand correctly, focusing on just one axis (in this case the room length) a first order mode can be cancelled by a single sub/speaker at the null, but second order modes by definition require two subs to achieve any sort of cancelation. Stands to reason then that, where I am pretty much sitting in one of the nulls, it doesnt matter that the speaker in question is sitting in the other: the mode is very much still manifest.

Thank you for the refresher (its been years)!

 

[krabapple]

If I understand correctly, focusing on just one axis (in this case the room length) a first order mode can be cancelled by a single sub/speaker at the null, but second order modes by definition require two subs to achieve any sort of cancelation.

Two subs lets you cancel (more accurately, 'attenuate') first, second and third order modes along one axis*. That's what Dr. Toole's illustration shows (and I've put up a screen cap of the whole page now, so you can read all his texts and keys to the illustration)

And of course this was a 'simplified, exaggerated' illustration, as Dr. Toole says. You have to know with some accuracy/experimentation what the modal behavior is in your actual room. Mode predictors represent idealized rooms. But if you were armed with e.g. REW and a mic, you could start from Toole's model and shift the subs until you get it 'right'. Note that attenuation can be fine tuned by adjusting the level of one of the subs.

*NB it's along one axis -- conceptually you'd have to do this along the length axis too! (Most people don't bother with height...)

Stands to reason then that, where I am pretty much sitting in one of the nulls, it doesnt matter that the speaker in question is sitting in the other: the mode is very much still manifest.

Correct, each seat has a different relationship to the room's modal nulls and peaks. It's common for one listener to sit in the midline of the width, in the vicinity of a first order width mode null.

I myself use the Toole 'trick' here to position my subs along my front wall, and I wish I could show you the measured results, but I've still been too lazy/distracted to do proper room measurments...