How audible is a notch?

JohnR · « **on:** 12 Jul 2011, 10:14 am »

Hi all

I'm doing some experiments with subwoofers. I was wondering how audible a "notch" is, such as the one at 80 Hz in the graph below. This is the third-order mode across the width of my room (and the listening seat is centered hence in the null).

I'm weighing up how much effort to put into flattening it. Any thoughts appreciated!

jtwrace · « **Reply #1 on:** 12 Jul 2011, 11:37 am »

It's worth trying to see if you can help it a little but I wouldn't go crazy if everything else looks good. I have found that the bottom end affects the mid quite a bit of what you hear if that makes any sense. I found that the "cleaner" I got the low end the cleaner the mids got. I don't think that little -10dB notch is worth going crazy over.

genjamon · « **Reply #2 on:** 12 Jul 2011, 02:55 pm »

I generally agree, but my only hesitation is that 80hz is right at the primary frequency for kick drums. If you're a jazz person, this could be a significant issue. For non-closed mic'd drum kits, I suspect you won't really be hearing kick drum as you should. You'll get a lot of deeper bass for kick drums that are close-mic'd, so shouldn't be a huge issue for rock. Of course, I'm no acoustics expert, and might be off the mark on this.

bpape · « **Reply #3 on:** 12 Jul 2011, 04:04 pm »

I assume you're running sans subwoofer John?

Are you 100% sure it's the 3rd harmonic of the width? Moving the mic 1 foot left or right kills the dip or at least moves it?

In the lower frequencies, 'notes' are pretty close together in frequency relative to higher pitches. It's going to impact a decent part of the bottom end (drums, stand up acoustic bass, etc.)

Depending on how much flexibility you have in terms of speaker placement, you could potentially deliberately introduce a peak due to boundary interaction to at least partially counteract it.

Bryan

JohnR · « **Reply #4 on:** 12 Jul 2011, 10:38 pm »

Hi Bryan, this is one main panel and sub. (Similar curve with the other panel and sub, or both panels and sub). I think it's that mode, it does change as you move across the room, although it takes more than a foot to make a significant difference... will post more graphs later. What else could it be due to?

JoshK · « **Reply #5 on:** 12 Jul 2011, 11:39 pm »

My $.02....

Is there a way to take that graph with less smoothing? That would give a better indication of the width of the null, which if narrow is very likely to be unaudible.

All else equal, I'd rather not have a null in that spot as the other poster mentioned. But who knows if you'd have the same null in stereo...

Josh

JohnR · « **Reply #6 on:** 12 Jul 2011, 11:47 pm »

Hi, the plot is unsmoothed. It's pretty much the same in stereo.

JoshK · « **Reply #7 on:** 13 Jul 2011, 12:14 am »

hmmm.. it looked smoothed, but anyway, it looks to be approx 15hz wide...I'd say that needs to be treated as it will take away the visceral impact of music. That is our most accute punch freq, so I'd say its important.

JohnR · « **Reply #8 on:** 13 Jul 2011, 09:08 am »

Quote from: JoshK on 13 Jul 2011, 12:14 am

hmmm.. it looked smoothed

Not bad eh? Well, except for the notch... but it doesn't sound like there's any visceral impact missing... so I'm not sure that the harmonics don't count more.

Anyway I figured out a placement that doesn't have the notch. It's just inconvenient... but I think I can use it to set up an apples-to-apples comparison using the DSP. Will report back.

JohnR · « **Reply #9 on:** 13 Jul 2011, 12:20 pm »

Original config in red, new one in blue:

bpape · « **Reply #10 on:** 13 Jul 2011, 01:10 pm »

The reason I asked was due to the width of the null. 15-20Hz down that low is pretty wide for a pure axial mode. Could be a combination of multiple things like tangential and oblique modes, boundary interactions, slight phase error between speakers and the sub, etc.

JohnR · « **Reply #11 on:** 13 Jul 2011, 01:31 pm »

Ah..... yes, there is a cluster of modes there that are null(ish) at the listening position:

I don't think it's boundary related, as it shows up in almost any driver position, except some of the subs against the wall. With regard to the idea of deliberately introducing a peak, the driver would need to be a half-wavelength from a wall?

bpape · « **Reply #12 on:** 13 Jul 2011, 02:26 pm »

Multiple subs???? Do you have infinitely adjustable phase or just a 0 and 180? I also assume you're cutting off the panels at 80 and crossing to the sub - or are you running the panels full range and overlapping the sub?

Bryan

JohnR · « **Reply #13 on:** 13 Jul 2011, 03:09 pm »

I've been measuring subs at a number of different positions. Part of the process of choosing which ones and where. I've decided to highpass the mains, and the subs run overlapped. Well, they always do but the highpass is not very steep:

I'm using a DSP can can delay any channel up to 9 ms. I needed to do that to get the blue plot in the previous post, which is with a second sub added to the mix.

This is actually a "pre-treatment" series of measurements, the baseline per Paul Spencer's Bass Integration Guide. But even so I'm interested to know what's going to audibly apparent.

PS. You can see in the plot that the "notch" occurs for both the subs and the main panel - which are a long way apart.

bpape · « **Reply #14 on:** 13 Jul 2011, 03:41 pm »

Which means that it is definitely related to where you're sitting rather than transducer location.

I do think that a notch that wide that low in frequency will be noticeable.

Have you tried center front wall and center rear wall sub locations at the same time?

JohnR · « **Reply #15 on:** 13 Jul 2011, 03:45 pm »

Yes, using sealed subs, but the decay times were not as good as this case with the one dipole sub. I'll just see if I can find a graph...

JohnR · « **Reply #16 on:** 13 Jul 2011, 03:48 pm »

OK, this is a pair of sealed subs at the center front and center rear walls:

This is the decay plot (150 ms shown):

bpape · « **Reply #17 on:** 13 Jul 2011, 07:15 pm »

It's a tradeoff but definitely shows that it is having a positive impact on reducing the null at 80hz which would indeed indicate that your original theory about it being a width mode primarily was correct and the 1/2 placement deals with this pretty effectively.

Bryan

JohnR · « **Reply #18 on:** 15 Jul 2011, 11:55 pm »

Thanks Bryan. I may also try measuring a pair of subs on the side walls at some point. However, I decided to do a constellation of listening measurements, and discovered that moving the listening seat 30cm forward makes a big difference. It's now a bit closer than I'm used to. Below is the original position in red, new position in green. I'm very happy with the sound at this point.

Acoustics are complicated! I guess you're used to seeing this kind of thing.

bpape · « **Reply #19 on:** 16 Jul 2011, 02:24 pm »

There's always a definite tradeoff. With the subs centered front and back and with the new seating position, we can hopefully still address the higher dip and mitigate the resulting lower dip.

Pretty tough to get things perfect but small changes can make a big difference. Many times, it's just an iterative process. An RTA helps a TON in making this a quicker process.

Bryan