Adventures in Waterfalling Part Deux: Multiple Subs, Free Bass Traps, and More

Wayne A. Pflughaupt

Moderator
Staff member
Thread Starter
Joined
May 21, 2017
Messages
466
Location
Corpus Christi, TX
My AV System  
Preamp, Processor or Receiver
Yamaha RX-Z9 A/V Receiver
Universal / Blu-ray / CD Player
Pioneer BDP-23FD Blu Ray Player
Front Speakers
Canton Karat 920
Center Channel Speaker
Canton Plus D
Surround Speakers
Canton Plus D
Surround Back Speakers
Realistic Minimus 7 (front EFX speakers)
Front Height Speakers
Yamaha YDP2006 Digital Parametric EQ (front mains)
Rear Height Speakers
Yamaha YDP2006 Digital Parametric EQ (subs)
Subwoofers
Adcom ACE-515 (for power management)
Video Display Device
Yamaha DT-2 (digital clock display)
Screen
Audio Control R130 Real Time Analyzer
Remote Control
Stock Yamaha Remote
Streaming Subscriptions
Pioneer PDP-6010FD 60" Plasma TV
Adventures in Waterfalling Part Deux
Multiple Subs, Free Bass Traps, and More

NOTE: Readers are encouraged to wade through my previous article on waterfalls and signal levels before delving into this one, to fully understand the terms and concepts presented.


Do Positive-Gain EQ Filters Cause an Increase in Ringing?
It’s known that EQ filters cause ringing, especially positive-gain (boosted) filters. That’s because equalizers use phase shift (which is time-domain) to accomplish frequency response changes. It’s part and parcel to the beast. In addition to whatever electronic ringing the filter may contribute, increasing the signal level at a specific frequency via a boosted filter means the signal is going to take longer to fade away.

This isn’t necessarily a bad thing: If your frequency response curve has a depressed area, its decay time is less than what you have at adjacent frequencies. Raising the signal level of the depression (with equalization) will naturally increase* its decay time – the two can’t be separated. There’s no downside to that, unless you happen to think your sub sounds better with certain frequencies attenuated below the rest (hat tip: it doesn’t).

But does a boosted filter increase the depressed area’s rate of decay along with its decay time? In other words, after EQ, will a time-domain measurement show that area now takes longer to fade away compared to adjacent areas, as you would see with a room mode?

A participant on that epic Home Theater Shack waterfalls thread I mentioned in the Waterfalls / Signals piece offered the following graph that shows ringing from a 1/6-octave, +15 dB filter. Further supporting the ringing contention, you can plainly see that spacing between the slices where the boosted filter is employed is much tighter than what is seen at frequencies above and below the filter. Even worse, we can see that the decay rate has burgeoned from less than 150 ms to well over – well, who knows. This indicates a dramatic and detrimentally longer decay rate.


1-rew-filter-waterfall-a-jpg.jpg

Graph 1: Electronic Waterfall


The problem with the theory is this: The filter, and waterfall graph, was generated in the REW program via a loopback connection. In other words, the graph is an electronic representation, not an acoustic representation. Although it’s no doubt accurate from an electrical standpoint, what really matters is what happens with an in-room measurement.

Naturally, it would require like-for-like filtering to definitively explore this idea in the “real world.” Unfortunately, or perhaps thankfully, 1/6-octave +15 dB filters should never be necessary when equalizing anything. But perhaps we can still make a point with equalization that’s more “real world” than that.

Here’s a waterfall graph of the subwoofers I had in a former residence after equalization, where I applied a combination of boosted and cut filters. The graph shows where the filters were set in the frequency spectrum, along with the amounts of boost or cut. The 22 Hz filter was a fairly wide one employed to enhance extension. (NOTE: I used a linear graph for this example as it makes the upper frequencies easier to study.)


2-cabinet-subs-waterfall-w-eq-graphics-linear-jpg.jpg

Graph 2: Waterfall with Positive- and Negative-Gain Parametric Filters
Window = 350 ms, Floor = 45 dB


You can see by studying the spacing between the slices that the boosted areas do not show an appreciable difference in decay rate compared to other areas of the graph. Indeed, the boosted filter at 85 Hz shows the same spacing as the negative-gain filter at 42 Hz.

How can this be, if positive-gain filters increase ringing, or more precisely, the rate of decay? Obviously, a waterfall graph generated in a sterile electronic environment bears no resemblance to one generated in a messy acoustical world. Any electronic inflation in decay rate from boosted filters is obviously swamped by the characteristics the room naturally exhibits. It would require a room heavily dampened with bass traps to show that boosted EQ filters get a real-world increase in the low frequency decay rate. For the rest of us, it’s a non-issue. There may be good reasons not to use boosted EQ filters (most of which are arguable as well), but an increase in ringing isn’t one of them.
* Increase (transitive or intransitive verb): To make greater in quantity, extent, value, or amount, etc.; to add to; to extend; to lengthen; to enhance; – opposed to diminish. Collaborative International Dictionary of English

Do Multiple Subwoofers Reduce Ringing?
Most proponents of using multiple subwoofers in a room claim they offer benefits such as reduced ringing and elimination of modal peaks.

It should be obvious that adding more subwoofers has the potential to raise signal levels, and as such make a time-domain graph look worse. However, if adding subs can prevent the formation of room modes, then it could legitimately be claimed that multiple subs reduce ringing.

I haven’t seen anyone present baseline “before and after” time-domain graphs showing the effects of additional subwoofers. By that I mean, a measurement of a single sub that shows a room mode (i.e. no EQ), followed by another measurement after additional subwoofers were added (again, with no EQ). And of course, the second graph would have to be level-matched to the first’s room mode.

The closest thing I’ve seen are the graphs below from a popular on-line magazine and forum, but unfortunately the “Multiple Subs” graph also includes the effects of equalization. The graphs aren’t precisely level-matched, but at least are closer than most.


3-single-sub-before-jpg.jpg

Graph 3: Primary Seat, One Sub, no EQ
Window = 576 ms, Floor = 50 dB

4-multiple-subs-after-jpg.jpg

Graph 4: Primary Seat, Multiple Subs, with EQ
Window = 576 ms, Floor = 50 dB


The results are mixed, as we shall see. Part of the problem with this example is that the display window is long – 600 ms instead of a more standard 3-400 ms. That makes the room mode at 58 Hz look rather insignificant, even though it is actually taking over 100 ms longer to decay to the graph floor than the areas around it (~345 ms vs. ~230 ms).

You can see by comparing Graphs 3 and 4 is that the slice spacing at 58 Hz is the same in both graphs. In addition, even though its signal level in Graph 4 is reduced by a few dB, 58 Hz is still extending all the way out to 345 ms. So, adding more subs did not make an improvement in the 58 Hz mode’s decay rate.

Turning at the 38 Hz peak, its decay rate did indeed improve between Graphs 3 and 4, even though its signal level actually increased a few dB. In fact, the spacing of the slices between the two is significant, showing very wide spacing in Graph 4 compared to Graph 3. This means the room mode at 38 Hz has been effectively eliminated. So that’s a positive outcome. However, since the graph includes equalization, we can’t tell if it was the additional subs or the equalizing that eliminated the mode – we’d need a before-EQ graph to determine that.

As I said, the results of this example are mixed. Of course, every room is different. You may get an improvement with room modes with additional subs, or you might not. But now you know how to tell.

Either way, I doubt many will dispute the notion that there are definite benefits to adding more subwoofers. For instance, other graphs in the same article show relatively consistent bass at multiple seats, which is virtually impossible to achieve with a single subwoofer.

But even if extra subs do prevent the formation of room modes (which is certainly a good thing), they aren’t going to reduce the overall decay rate beyond what the room would naturally exhibit. That requires absorption. Anyone who believes that additional subwoofers will take your room from this...


5-realtrapsbefore-l-jpg.jpg

Graph 5: Empty Room


... to this...


6-realtrapsafter-l-jpg.jpg

Graph 6: Room with Bass Traps
Courtesy of Real Traps


...also believes you can reduce flutter echo in a cinderblock room by adding more speakers.


How Much Improvement in Decay Rate Can EQ Give?
It’s a legitimate question: Just how much improvement in the decay rate of a room mode can be accomplished with parametric equalization?

Below is a graph showing a room mode at 42 Hz. We can see that it takes 450 ms to drop to the 60 dB mark (indicated by the horizontal line across the graph face).


7-cabinet-subs-baseline-a-jpg.jpg

Graph 7: Baseline Waterfall Graph at 450 ms
Window = 450 ms, Floor = 50 dB



Now let’s look at the same room and subwoofer after applying an EQ filter to the room mode. This graph is level-matched to the previous at 42 Hz.


8-cabinet-subs-w-eq-jpg.jpg

Graph 8: Equalized Waterfall Graph at 350 ms
Window = 350 ms, Floor = 50 dB



As we can see, the former room mode has now dropped to the 60 dB mark in 350 ms, and at that point is exhibiting modal behavior similar to surrounding frequencies. In other words at 350 ms, 42 Hz has dropped as far as it’s going to. Thus we can see that EQ brought an improvement of 100 ms in the room mode’s decay rate.

I don’t consider this to be hard proof as to the amount of decay-rate improvement you can get from equalizing a room mode. It’s a big world, and I’m confident someone, somewhere has had a worse room mode. However, even accepting that, it seems to me that a level-matched improvement of 3-400 ms after equalization, as I’ve seen some claim, is highly improbable. That would mean their mode was originally taking a full 800-900 ms to fade down to the noise floor, assuming a measurement peaking in the 90-95 dB range. If that’s the case, the room has more issues than an equalizer can address. No, more than likely any “improvement” on the order of 300-400 ms is merely the result of a huge loss of signal level due to severe equalization (see “Manufactured Waterfall Malfeasance” in Waterfalls / Signal Levels article).


Want a Better Waterfall? Get a Smaller Room
Something I was recently made aware of is that larger rooms exhibit more ringing than smaller rooms (although on further reflection it should have been a no-brainer).

Here is a waterfall from our current master bedroom, which measures 12’ x 18’ for 770 cu. ft.


9-lawnview-bedroom-no-eq-jpg.jpg

Graph 9: 770 Cu. Ft. Room
Window = 300 ms, Floor = 45 dB



Here is a graph from the living room of our former home, which had a soaring vaulted ceiling and was open to the entry, dining room, kitchen, breakfast room, and upstairs landing. The total volume was a fairly cavernous 9000 cu. ft. or more.


10-crossfield-cabinet-subs-jpg.jpg

Graph 10: 9000 Cu. Ft. Room
Window = 300 ms, Floor = 45 dB



The relevant data here is between 20-60 Hz. You can easily see that the large room, even with most of that range at a lower SPL than the small room, nevertheless shows a much longer decay rate than the small room.

If that weren’t sufficient to drive home the theory, let’s “go extreme.” The following graph was generated with my full-range bass guitar rig, as measured in a smallish church sanctuary. The system consists of a 15” EAW subwoofer with a Genzler top. The sanctuary is 3800 sq. ft. with total cu. ft. coming in at 46,000 – much larger than any home theater room. With this, you don’t even have to level-match the SPL or graph floor to make the point.


11-love-grace-church-jpg.jpg

Graph 11: Small Church Sanctuary
Window = 300 ms, Floor = 50 dB



Despite the sad looking waterfall, this is actually a great-sounding room. The floors are a combination of hardwood and carpet, and the side-wall construction offers tremendous dispersion. Well-padded chairs also help with absorption. Music from a three- or four-piece band and singers is completely coherent, even with house speakers that are merely average.


Free Bass Traps?
I’ve heard it said that openings in a room can function similar to bass traps and reduce ringing. I suppose the idea is that providing the bass an “escape” will make a difference.

So I dragged my new Hsu ULS-15 MK2 sub to our master bedroom and fired up REW to see what I could find. The bedroom is 215 sq. ft / 770 cu. ft. and has three doors – the main one that opens to the living room, and two others opening to a closet and a small bathroom.

The following graphs have no equalization as we want to see any potential room modes in all their umm, glory. Here is a baseline with all doors closed:


12-free-traps-baseline-jpg.jpg

Graph 12: Master Bedroom All Doors Closed
Window = 300 ms, Floor = 50 dB



Here is the graph with only the main door open. As you can see, ringing below 30 Hz does appear to be improved:


13-free-traps-main-door-open-jpg.jpg

Graph 13: Master Bedroom Main Door Open
Window = 300 ms, Floor = 50 dB



Opening the other two doors got this:


14-free-traps-all-doors-open-jpg.jpg

Graph 13: Master Bedroom All Doors Open
Window = 300 ms, Floor = 50 dB



Either the closet or bathroom, or both, appear to be acting as a kind of “resonator.” The peak at ~23 Hz has increased by 2 dB, and the decay rate below 30 Hz is definitely longer.

All these results were repeatable, by the way. So it appears that the theory holds, at least to some extent, even though I honestly doubt doors opened or closed is going to make a difference that anyone can actually hear. But experiment for yourself to find out if the “open door policy” makes a difference in your room.


Conclusion
As we can see from the above examples, there is a lot that can be determined from waterfall graphs, once you know how to properly read them.
 
Last edited:

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
469
Wayne- "That would mean their mode was originally taking a full 800-900 ms to fade down to the noise floor, assuming a measurement peaking in the 90-95 dB range"

DD- Sweep measurements and subsequent Waterfalls can see way down into the broadband noise floor. A typical domestic or prosumer room may have a Z or C weighted Leq or SPL of say 40dB. Waterfall graphs from 75dB sweeps can show decay tails meeting the noise floor at their own frequency way down at 20dB. I used to recommend 90-95dB Sweep levels until JohnPM kindly set me straight on this. 75dB Sweeps are fine.
 
Last edited:

Nordo

Member
Joined
Oct 7, 2019
Messages
73
My AV System  
Preamp, Processor or Receiver
Denon AVR X1500H
Additional Amp
Sub amp - Behringer EP2500
Universal / Blu-ray / CD Player
Panasoic BD35
Front Speakers
DIY
Center Channel Speaker
DIY
Surround Speakers
DIY
Surround Back Speakers
DIY
Front Height Speakers
-
Rear Height Speakers
-
Subwoofers
Four 15" infinite baffle drivers
Other Speakers or Equipment
HTPC (windows 10 with Kodi)
Video Display Device
Epson EMP-TW2000 projector
Screen
Matt white "blockout" curtain liner
@Wayne A. Pflughaupt
Maybe I'm reading it incorrectly but if you increase the rate of decay, the decay occurs quicker i.e. you reach -60dB in a shorter time.
If you decrease the rate of decay, the decay takes longer to drop to a certain level.
You seem to be saying the other way around.
Or have I got it wrong again. :hide:
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
469
Maybe I'm reading it incorrectly but if you increase the rate of decay, the decay occurs quicker i.e. you reach -60dB in a shorter time.
If you decrease the rate of decay, the decay takes longer to drop to a certain level.
You seem to be saying the other way around.

LOL, indeed, and what if we IMPROVE the Decay. And isn't Speed a Rate?
 

Nordo

Member
Joined
Oct 7, 2019
Messages
73
My AV System  
Preamp, Processor or Receiver
Denon AVR X1500H
Additional Amp
Sub amp - Behringer EP2500
Universal / Blu-ray / CD Player
Panasoic BD35
Front Speakers
DIY
Center Channel Speaker
DIY
Surround Speakers
DIY
Surround Back Speakers
DIY
Front Height Speakers
-
Rear Height Speakers
-
Subwoofers
Four 15" infinite baffle drivers
Other Speakers or Equipment
HTPC (windows 10 with Kodi)
Video Display Device
Epson EMP-TW2000 projector
Screen
Matt white "blockout" curtain liner
"Improve" is not a good word to use when discussing anything really.
If it's too much for you, you improve it by lowering it.
If it's not enough for you, you improve it by raising it.

In part I of Wayne's discussion on Waterfalls, etc. his terminology makes sense.
In this part II, it seems to be back-to-front. Or maybe it's just poor choice of words.
 
Last edited:

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
469
That would mean their mode was originally taking a full 800-900 ms to fade down to the noise floor.
Que? That seems OK to me. In fact modes longer than this are common and normal in untreated small rooms. In my CR the lowest mode lasted 1.4S It was reduced to about 700mS by three large SSCs. Then reduced a further 25% by Dirac Live. Still a bit wild, but concrete walls....... I kinda like the subby presence and mix close to the central null.
 

Nordo

Member
Joined
Oct 7, 2019
Messages
73
My AV System  
Preamp, Processor or Receiver
Denon AVR X1500H
Additional Amp
Sub amp - Behringer EP2500
Universal / Blu-ray / CD Player
Panasoic BD35
Front Speakers
DIY
Center Channel Speaker
DIY
Surround Speakers
DIY
Surround Back Speakers
DIY
Front Height Speakers
-
Rear Height Speakers
-
Subwoofers
Four 15" infinite baffle drivers
Other Speakers or Equipment
HTPC (windows 10 with Kodi)
Video Display Device
Epson EMP-TW2000 projector
Screen
Matt white "blockout" curtain liner
@DanDan
Wayne said :-
"But does a boosted filter increase the depressed area’s rate of decay along with its decay time? In other words, after EQ, will a time-domain measurement show that area now takes longer to fade away compared to adjacent areas, as you would see with a room mode?" (my emphasis)

Increasing the rate of decay, would shorten the decay time.

There are a few other areas where Wayne seems to make the same mistake.
 

Draki

Registered
Joined
May 23, 2017
Messages
2
Wayne

Can you please advise the exact "settings" for the various parameters, or post a clearer screenshot? So that we could use the same (visual) reference.. Thanks. Sorry if that has been already asked and posted, but I couldn't find it.
 

Wayne A. Pflughaupt

Moderator
Staff member
Thread Starter
Joined
May 21, 2017
Messages
466
Location
Corpus Christi, TX
My AV System  
Preamp, Processor or Receiver
Yamaha RX-Z9 A/V Receiver
Universal / Blu-ray / CD Player
Pioneer BDP-23FD Blu Ray Player
Front Speakers
Canton Karat 920
Center Channel Speaker
Canton Plus D
Surround Speakers
Canton Plus D
Surround Back Speakers
Realistic Minimus 7 (front EFX speakers)
Front Height Speakers
Yamaha YDP2006 Digital Parametric EQ (front mains)
Rear Height Speakers
Yamaha YDP2006 Digital Parametric EQ (subs)
Subwoofers
Adcom ACE-515 (for power management)
Video Display Device
Yamaha DT-2 (digital clock display)
Screen
Audio Control R130 Real Time Analyzer
Remote Control
Stock Yamaha Remote
Streaming Subscriptions
Pioneer PDP-6010FD 60" Plasma TV
Hey Draki,

My apologies for the graphs. With many of the original .jpg files, the font size is small but still readable. However, a LOT of resolution is lost with the uploading. I would have liked to post larger pictures, but the Forum likes to limit the size to 600, I assume for the benefit of mobile devices. I have now added text below the graphs as needed showing the window length and graph floor, so that should help. SPL levels (left side) are easy to calculate as the markers are in 5 dB increments.

As far as graph parameters go, most people seem like a window in the 300-450 ms range. Much like the signal levels discussed in Waterfalls Part 1, it can be noted that the long window in Graphs 3 and 4 (nearly 600 ms) tend to “improve” the waterfall, as the signal falls to the floor long before reaching the end of the window.

Since the purpose of a waterfall is to see how long it takes the signal to completely fade away, the graph floor should typically be something in the 35-40 dB range, as that seems to be a common room noise floor (people living in remote areas or heavily soundproofed rooms typically have lower noise floors). You’ll note that I used a higher floor than that for many of my graphs; that’s because they were merely to show a point of discussion rather than a true room analysis.

As far parameters other than window and floor, I just use the REW defaults. I’m sure others have their own opinions and preferences on displaying waterfall graphs.

Regards,
Wayne A. Pflughaupt
 

Sonnie

Senior Admin
Staff member
Joined
Apr 2, 2017
Messages
3,991
Location
Alabama
My AV System  
Preamp, Processor or Receiver
Monolith HTP-1 Processor (Movies and Surround)
Main Amp
Benchmark Media AHB2 Monoblocks
Additional Amp
Emotiva XPA-Eleven
Computer Audio
Intel NUC w/ Roon ROCK
DAC
miniDSP SHD (Two-Channel Music Only)
Universal / Blu-ray / CD Player
Panasonic UB9000 4K UHD Player (for movies only)
Front Speakers
MartinLogan Renaissance ESL 15A
Center Channel Speaker
MartinLogan Focus C-18
Surround Speakers
MartinLogan EFX Surrounds
Surround Back Speakers
MartinLogan ElectroMotion ESL
Front Height Speakers
MartinLogan EM-IC
Rear Height Speakers
MartinLogan EM-IC
Subwoofers
SVS SB16-Ultra x4 (music) + PB16-Ultra x2 (movies)
Video Display Device
JVC DLA-NX9
Screen
Elite 128" Screen
Remote Control
Universal MX-890
Streaming Equipment
Roku Ultra
Streaming Subscriptions
Lifetime Roon Subscription
Tidal
qobuz
Netflix
Amazon Prime
Satellite System
Dish Joey 4K
Other Equipment
Kaleidescape Strato S 12TB
My apologies for the graphs. With many of the original .jpg files, the font size is small but still readable. However, a LOT of resolution is lost with the uploading. I would have liked to post larger pictures, but the Forum likes to limit the size to 600, I assume for the benefit of mobile devices. I have now added text below the graphs as needed showing the window length and graph floor, so that should help. SPL levels (left side) are easy to calculate as the markers are in 5 dB increments.
Hey Wayne... you can post a pretty large image in the thread. You can literally copy and paste the image into the thread. You can then choose Thumbnail or Full Image at the bottom.

37413
 

Attachments

Top Bottom