Adding a Subwoofer to a 2-channel audio system in a large complex space

[levb]

We have a pretty good HiFi system purchased in 2004 that I haven’t done anything to besides getting some decent interconnects & speaker wires. It’s a purely audio system with Unison Research’s Unico Secondo integrated hybrid amplifier (newly acquired after the original Unico SE’s right channel got fried accidentally), Unison Research’s Unico CD and a pair of Triangle Celius ES flood standing 3-way speakers.

A couple of years ago, we moved into a house with a much larger living room and a semi-open layout. The salesperson at the audio shop where I got my new amplifier said that my space would really benefit from a subwoofer; even a pair of subwoofers if I could pull it off. Based on my room description, he recommended a JL E-112 or E110, or a REL S/510 or T/9x. Out of those, even the T/9X is more than I would like to spend ($700 - $1,000).

My other limitation is that the entire wall where my audio, wifi & tv are located runs on a single 20-amp circuit. That said, when listening to the hifi, the only other components that are always on are a couple of Panamax power conditioners, the wifi router, and the wifi extender. So when we’re listening to music, we have to stay under 20-amps with the amplifier, CD player, subwoofers & the always on components are running simulatenously.

The living room is approximately 18-ft wide by 25-ft long by 19-ft tall, but is bisected lengthwise by a mezzanine with a solid guardrail height wall, so the double height space is ~ 11-ft wide by 25-ft long, while another 8-ft of the entire length is 9-ft tall. Above the mezzanine is our open office that’s 11-ft wide by the full 25-ft length, and the continuous ceiling is 8-ft high over the mezzanine. The ceilings at both the lower portion & double height space are wood slats (2x2’s) with an equal solid / void ratio, so sound can penetrate the 12” high joist area above. The two short walls along the 11-ft double height portion are mostly glass, the long wall along the double height space is ¾” thick wood planks on furring strips while the long wall under the mezzanine is sheetrock with a couple of large steel-clad sliding doors. The guardrail height wall of the mezzanine which faces the double height wall is also clad with wood planks on furring strips. One end of the lower space is open to the dining room, and the kitchen beyond, so it’s definitely not a simple enclosed rectangle.

The left half of the living room is our relaxing area (couch, coffee table, audio & tv) while the right half is for my wife’s grand piano. The audio system is located along the wood clad double height wall and the couch is right under the edge of the mezzanine, some 10 – 11 feet away. The speakers are spaced 9 – 10 feet apart. This puts the left speaker about 3-ft from the glass wall, with a larger cactus between it and the glass, and the right speaker very close to the center of the room. We either listen to the music from the couch or turn the system up some when working in the kitchen … and the sound from afar (~ 25ft) is surprisingly good. I’m guessing that the best location for a subwoofer would be next to, or just behind, the right speaker; away from the glass wall and centered on the room, albeit not centered on the optimum listening position.

My question to all the subwoofer experts is can I get ONE subwoofer that isn’t any bigger than say 15” – 16” square that will give us a good enough improvement to what we already have, and do it for $1000, or less? The SVS SB-2000 Pro, SB-3000Pro or their 3000 Micro? Would the JL Dominion d110 bring anything extra to the equation to be worth going over my budget?

 

[skid00]

This gentleman has created a very useful google docs chart: https://www.audiosciencereview.com/forum/index.php?threads/subwoofer-comparison.20494/

My speakers naturally drop off in output below 30 Hz. I have EQ'd them to extend to 22 Hz before any drop off occurs. It's a very huge difference, especially with music that you don't expect to have deep bass.

I strongly recommend you try a sub with high output, at least to 30 Hz, if not lower.

 

[perceval]

No doubt a room opened with angles will be hard to deal with.

And the salesperson was right. It's much better having two, three or more 8" subs than a single 15" in a corner.
Standing waves using a single sub will create nodes, peaks and dips throughout your room. You'll have booming in one spot, walk a couple of steps, and it will be gone with little bass at all.

And I can probably be sure that what you think the best place for sub is, will be a bad spot!
You will have to move it (them) around to find an optimal location.

This way, having at least two 8" subs, you will be able to place them somewhere else than your main "audio" wall, saving yourself from tripping the 20A breaker!

Lots of places have return policies if the item doesn't fit your room or needs. Get a couple of 8" subs and experiment with them over a weekend. If it doesn't work, send them back, and try to plan for a dedicated audio room in your house at some point.

 

[deercreekaudio]

Your suggestion of the SVS SB-3000 Pro should be adequate, preferably located in between the main speakers. Whatever subwoofer you chose we believe proper subwoofer integration and calibration are the keys to success. Following are some highlights a whit paper we are working on.

Crossover Setting
The crossover between the main speakers and subwoofer(s) is a skill that goes beyond applying the symmetrical textbook high and low pass filter. We have found the asymmetrical filter slopes and shapes can produce significantly enhanced results. See image below for an example of a crossover we use with JBL 4349s and a high power 12-inch subwoofer. What's notable is the soft hi-pass slope for the mains and sharper low-pass slope for the subwoofer.

Delay Setting
Next, delay needs to be adjusted so that the gross arrival time between the subs and the mains is aligned to achieve the tightest non-interfering response.

Time Domain Correction (not delay)
The Dirac room correction process attempts to improve the impulse response of the combined main speaker and subwoofer system. When properly performed, a Dirac project results in the seamless integration between the subwoofer and the main speakers, allowing the system to achieve the bottom octave of music without the undesirable effects of an untuned subwoofer installation. Here's a graphic example of impulse response optimization.

It's an amazing high fidelity musical experience when you seamlessly add the last octave. Feel free to contact us at Deer Creek Audio if you'd like to discuss these concepts further.

 

[levb]

@perceval Unfortunately, I don't have the time or means to assemble a dedicated listening room / area. Since I am completely new to the topic of subwoofers, I'm trying to deduce is if I can make my current system sound fuller (even though we're happy with the current sound) by adding a sub (or two). Speaking with the SVS tech, he said that with my floor standing speakers, two smaller subs (dual SB-1000) just won't cut it compared to a single SB-3000. As much as I would like to optimize my system to our large living room, I'm also not willing to go down the rabbit hole of adding component after component (more subs, dsp's, $$$ cables, wireless options, etc) racking up thousands of dollars. My big max is $1,200 and that's pretty much it.

I looked at the Rythmik L22 & PSA S2410, as someone recommended, but their return policy doesn't include return shipping and since the L22 comes in over 100-lbs, I'm not willing to take that risk. At least with SVS, I can try it for 45-days and if it doesn't work out, I can send it back without additional shipping fees.

 

[levb]

@deercreekaudio Sorry, had my reply ready and had to walk away for a while, so didn't notice your info before clicking send. I'll have a closer looks at your info later today. Much appreciated.

 

[perceval]

The SB1000 is a 12" sub.
The SB3000 is a 13" sub.

Wondering how much more punch you will get moving to a driver only one inch bigger! Sure, it has 2.5 time the amount of watts from the amp, but then again, you are looking into adding meat to the low end, not trying to watch a superhero movie. There might be other things happening between those two models, but just looking at cone size, 1 inch will not make it much better... it just can't move that much more air.

A dual setup will always trump a single sub setup. Always.

You also didn't mention how you will split the signal from your 2 channel setup.

 

[levb]

@perceval I'm totally new to the world of subs, so I have no clue how much difference a 1" increase in driver does. I don't even know how best to integrate / split the signal with my amp / speakers. My integrated (hybrid) amp has just one unbalanced RCA SUB OUT, which I assumed was enough. Since then, I've read discussions about DPS's being critical while others rely on apps (like with the SVS subs). If I have to move the sub freely around the space, do I go with a wireless adapter, how about lag? I really don't know anything about all these fine tuning methods.

Even with our current system, I haven't tinkered with it since I got the three major components (except for upgrading to good quality speaker cables & interconnects). I'm just trying to decide if spending $1,200 will increase the quality of our music listening enough to warrant the expenditure (not something my wife is crazy about doing since we just forked over for a replacement amp). I'm a true greenhorn when it comes to the finer details of high end audio.

 

[tesseract]

I have two new SB-3000s and will be adding a third, for room smoothing purposes. My application is different than yours and my room is much smaller.

If I were to use a single sub in your space, I'd step to a ported enclosure. I do understand that the size constraint is outside of your want.

 

[Wayne A. Pflughaupt]

@levb I’ve never had any problem getting a single capable sub to sound good in a large, irregular room like yours, after parametric equalization. They typically don’t have all the problems with room modes that small dedicated rooms tend to have.

I expect the folks at SVS are giving you advice based on home theater applications, which are more demanding than most dedicated music systems. Unless you like to listen at rock-concert levels, take their advice accordingly.

An advantage of the SB3000 is that it has lower extension. That might be critical in your room given it size, in that your in-room extension may not reach as low as the sub is spec’d.

Since SVS is so generous with their returns, you might just order both the SB3000 and a pair of SB1000 and see which one you like best. Personally I’d go with the 3000, again based on its better extension.

Regards,
Wayne

 

[perceval]

Or... since you seem a bit weary about the whole process, I suggest you take a step back.
Take your time. Read some more about sub integration. Talk to your wife about stuff, not just audio (someone told me once, "a happy wife is a happy life"
Don't rush into something you might not be ready for.
Music is not going to go away tomorrow. You still enjoy your setup, and read some more. Maybe call a friend who's got a sub to come over with it. Try it, have a drink. Hear how it might work with your setup. When you feel you can make the right decision, between your components and sub(s), then make the move. Don't do it halfway, hoping it will be ok. That never comes out right.

 

[Head_Unit]

The living room is approximately 18-ft wide by 25-ft long by 19-ft tall

That's pretty big, so big subwooferage is called for, depending how far away you're sitting (I can't open the pdf on this machine).
- Your current speakers: if you crank things way way loud, are they fine or do they distort? If they distort, you need a highpass filter to cut out the low bass. Unfortunately your amp appears incompatible. and before someone says "speaker level connection" that can yes add more bass out of the sub, but will do zero to reduce the overload on your towers.
- You also do not have any room correction (like Dirac or Audyssey). While not cure-all panacea, the alternative is experimenting a LOT with placement which it seems you can't. An alternative is something like the SVS SB-2000 Pro which we run with Focal 936/CC900/SR900, or similar, with built-in parametric EQ. This is not as easy or powerful as Dirac/Audyssey, but far better than zero. It still doesn't solve the highpass problem however. Here you can see the effect of JL's built-in room EQ
https://www.stereophile.com/content/jl-audio-fathom-f110v2-powered-subwoofer-measurements
- YES, a subwoofer can add to your towers, heck it adds to the massive Focal 936. Most tower speakers really don't go as low as you might think; the 936 is ported to 40 Hz so anything below that is useless. With twin 6.5" I doubt your Triangle go lower either, so a sub can sure fill in those rare lower musical instruments and those rumbling movie effects.
- SVS is not the only good subwoofer company out there, I also admire Hsu and JL among others. I can say they package intelligently (READ the package outsides!!) and appear to offer a lot of support (which we didn't need due to clear instructions). The app is really handy as well.
- Our SB-2000 Pro came with a metal grille, which I much prefer. Do you have dogs/cats/whatever? Think about this. I read another SVS review commenting how glad he was they got rid of the metal grill, so you might want to ask SVS about this.
- Two subs *can* be better BUT it is NOT automatic. You need to be able to spend the time to integrate them. And for output, cone area is king, I would NOT get two 8" instead of a 15." Now cone area isn't everything, linear stroke and a high-flowing port are black magic, however in a big room the smallest I would want is a 12" with 400+ watts.
- Maybe an SVS PC-2000 Pro cylinder would fit in your space.

 

[witwald]

Your suggestion of the SVS SB-3000 Pro should be adequate, preferably located in between the main speakers.

That's probably a good spot, as it will be offset a bit from the centre of the wall where a low-frequency null will appear.

Whatever subwoofer you chose we believe proper subwoofer integration and calibration are the keys to success.

That's clearly going to come into play when trying to maximize the benefits of the subwoofer in a system.

Crossover Setting
The crossover between the main speakers and subwoofer(s) is a skill that goes beyond applying the symmetrical textbook high and low pass filter. We have found the asymmetrical filter slopes and shapes can produce significantly enhanced results. See image below for an example of a crossover we use with JBL 4349s and a high power 12-inch subwoofer. What's notable is the soft hi-pass slope for the mains and sharper low-pass slope for the subwoofer.

View attachment 47892

Assuming that the above plots show the nominal filtered responses, they don't show the resulting summed response. The choice of a low-order Bessel high-pass filter is concerning for a number of reasons:

1. There is a significant region of overlap in the high-pass and low-pass (subwoofer) responses. This can lead to comb-filtering like effects due to the phase mismatch between the two signals.
2. The Bessel filter high-pass response function is not something that is easily achievable when using a vented box or a sealed box loudspeaker system. These produce 4th-order and 2nd-order high-pass response functions. It is possible for a vented loudspeaker with its vent blocked to have a much shallower roll-off that is second-order and has a soft knee, but at best it is still only an approximation to a Bessel response function.
3. Expecting the main speakers to be –6dB at about 22Hz is being quite optimistic, as most main loudspeaker systems cannot achieve that level of performance (hence the need for the subwoofer).
4. There is a strong requirement for choosing filter functions that have complementary high-pass and low-pass filter characteristics. This is very well supported by Linkwitz-Riley filters. Higher-order Butterworth filters can also work reasonably well, as their overlap region is minimized so any deleterious interaction is highly localised and can usually be ameliorated by a small shift in crossover frequency.
5. The diagram above shows the filter settings, but we have no indication of what the filtered acoustic responses are like.

As an example, as I was curious to see the summed response, I have used the VituixCAD loudspeaker system modelling software to simulate the summed response of the filters shown above. Using the above transfer function plots as given, for the subwoofer (low-pass) acoustic response I used a 6th-order Butterworth high-pass filter with –3dB point at 10Hz, coupled with a 4th-order Butterworth low-pass filter with a –3dB point at 52Hz. For the main speaker I used a 3rd-order Bessel high-pass filter with a –3dB point of 25 Hz, coupled with a 4th-order Butterworth low-pass filter with a –3dB point of 16500Hz. Note that the polarity of the subwoofer was inverted to avoid a deep null at crossover.

The resulting summed response, together with the high-pass and low-pass filtered response functions, is shown below:

It is quite evident that the result is far from flat, and the output of the main speakers causes a lot of phase cancellation of the subwoofer's output below 20Hz, while there is strong in-phase summation occurring over a broad region centered on 55Hz.

Using the same 52Hz cut-off frequency, and changing the low-pass filter on the subwoofer to a 4th-order Linkwitz-Riley topology and the high-pass filter on the main speakers to a complementary 4th-order Linkwitz-Riley high-pass filter, produces the result shown below:

The above plot demonstrates the clear advantages of using complementary Linkwitz-Riley filters when integrating a subwoofer with the main speakers. To try and use anything else would seem to be overly complicated and less likely to produce good results.

I recognise that the above examples are simplified. However, I would like to suggest that they do serve to highlight how things work and provide useful information related to subwoofer integration in the absence of a more complete exposition of the topic.

 

[deercreekaudio]

Thanks witwald for your thoughtful response.
Attached are plots including the composite system response, with subwoofer and main speaker traces. Also attached is the Dirac project which shows the measured response before correction and the predicted response.
Soon I will address your excellent theoretical analysis and make some comments regarding the realities of phase and group delay responses in real non linear audio systems.
Looking forward to a lively discussion.

 

[Viabcroce]

I did not see this addressed, apologise if it was.
Extending the frequency response of an audio system is only one of the reasons why one or more subwoofers can improve your overall tone. When the main speakers are properly high-passed so to cut their lowest 1/1.5 octave (say for example @ 120 Hz), tests show that the distortion the (mid-)woofers produce is strongly reduced up to 500-600 Hz, thus dramatically improving the overall system response.

IME, adding sub(s) to a system makes most sense if you can linearise the room's response and high-pass the main speakers; and it's true, 2 properly-placed subs will create less issues than one only.
If you can't/don't want to get some all-in-one solution such as the Anthem STR for example, and don't know how to use an app such as REW to analyse your room's response, there's a prctical trick you can use: place the sub where your ears would usually be (height matters as well), and crawl around where you think you'd definitively place it. The best location(s) will be the one(s) where you hear a lot of energy but the least boominess/the shortest sound tails (you may want to use a test track for subs you can d/l from one of the several web sites on the matter, or even some generic white noise track). Better, IMO, if the main speakers are off at that stage.

If you might choose one of the above only, I'd advise for an active crossover that sets the filtering for both the sub and the main speakers.

Happy music

 

[witwald]

Extending the frequency response of an audio system is only one of the reasons why one or more subwoofers can improve your overall tone.

Not to mention having the facility for much more linear output SPLs at low frequencies.

When the main speakers are properly high-passed so to cut their lowest 1/1.5 octave (say for example @ 120 Hz), tests show that the distortion the (mid-)woofers produce is strongly reduced up to 500-600 Hz, thus dramatically improving the overall system response.

I'm not sure that a high-pass of the main speakers at 120 Hz is desirable. Wouldn't that result in localization of the subwoofers to some degree? I'd suggest that the highest high-pass frequency should not exceed 80 Hz.

The benefits of reduced cone displacement of the main speaker's woofers at low frequencies will still be quite substantial if they are high-passed at 80 Hz. Loudspeaker systems with one or more subwoofers reap the benefits of lowered low-frequency and upper-bass distortion all of the time. It's a much more high fidelity experience than without them.

 

[Viabcroce]

I'm not sure that a high-pass of the main speakers at 120 Hz is desirable. Wouldn't that result in localization of the subwoofers to some degree? I'd suggest that the highest high-pass frequency should not exceed 80 Hz.

Hey witwald, mine was just an example, and did not want to be a technical indication I just wanted to show that by not cutting the main audio speakers we miss one of the most important musical benefits of a sub. Generally speaking, I can hardly think of a universal rule.
Most recordings are basically mono up to around 100 Hz, so there's no spacial information. It's mainly the harmonics that glue to the fundamental and give us such information (hence the need for a proper phase/IR at and around the crossover freq).
Anyway, if we focus on the mid-woofer distortion, the best freq for the filter would depend on the "efforts" the loudspeaker is making in order to reproduce a general program: the smaller its diameter and the louder the wanted SPL, the more distortion at a given (low) frequency.
The tests show that the least (intermodulation) distortion is got when you cut about the first 1.5 "real" octave (hope this makes sense, I am having a hard time trying to render the Italian "useful" ).

 

[witwald]

Attached are plots including the composite system response, with subwoofer and main speaker traces. Also attached is the Dirac project which shows the measured response before correction and the predicted response.

Hello deercreekaudio. Having that very broad non-complementary overlap region between 20 Hz and 60 Hz seems to be quite counterproductive when it comes to high-quality sound reproduction. It's not as good a use of the main speakers and subwoofers as would be possible if they were more appropriately filtered and their respective responses then blended together. Those main speakers will be having very large excursions at low frequencies, which lead to nonlinear, distorted output.

When the woofer is called upon to reproduce multiple frequencies, and it's operating outside the limits of its nominally linear excursion, won't the higher frequencies that ride on that large-displacement cone motion also now be working in the distortion region?

Soon I will address your excellent theoretical analysis and make some comments regarding the realities of phase and group delay responses in real non linear audio systems.

I'm not sure of what you mean by nonlinear audio systems. The subwoofer and main speakers are linear devices. They have a magnitude response and a phase response that is a function of frequency. When driven within their linear operating range, they are linear devices, apart from any inherent distortion characteristics of their drivers, of course. Of course, when a woofer and a tweeter are combined using a crossover, their summed response generally becomes non-minimum phase in nature. However, their summed output still nominally belongs to a linear system, albeit a complex one, with its own frequency dependent directivity pattern.

Summing the responses of the main speakers and the subwoofer(s) is a linear operation. I understand that there are reflections occurring, and they do add time-delayed responses at the listening position. These could be regarded as being nonlinear additions to the overall response. At very low frequencies, those responses can be modelled as time-delayed linear additions to the main outputs from the speakers and the subwoofers.

Although the above plot looks enticingly smoothed, I'm not sure that processing the loudspeaker + room response at high frequencies is such a good idea. As there are no measurements supplied of the individual natural responses of the main speakers under essentially anechoic conditions, it is difficult to know what has been equalized by the processing.

Although I have no doubt that smoothing the acquired response above 200–300Hz at the listening position will sound different than the unprocessed natural response, there is nothing to say that the processed result is actually of higher fidelity. In fact, playing with the room acoustics in that region in that manner is probably likely to lead to an unnatural-flavoured sound quality. Just because it can be done, doesn't mean it's a good idea to actually do so. That's one of problems with audio reproduction in general, and loudspeaker–room interactions specifically. As many listeners have found, although something can sound different, it's not necessarily better when looked at more objectively, and I think that processing of "room sound" above 200–300Hz falls right into that category.

Do you have any anechoic measurements of the left and right loudspeakers at the listening position that you can share? They would be very informative to see, as we can then gauge the potential neutrality of those speakers.

In fact, any peaks and dips in the responses of the main speakers are prime candidates for removal by appropriately applied equalization in order to ensure the neutrality of their reproduction. Was any equalization applied to those speakers in order to ensure that their on-axis response was reasonably flat at typical listening distances? If not, then that's potentially quite a big error of omission.

In terms of the processed response and its measurement shown above, I assume that it is the one that is created with the microphone located at the listening position. Do you have any measurements of response taken at locations about 0.15 metres to the left and right of that listening location, and as well 0.15 metres in front of and behind the listening location, but taken after the so-called "corrected" response has been set? It would be very informative to see those as well.