Why is good, deep bass so difficult? - Myths and their Busters

Designing and building a woofer system that is theoretically flat to a very low frequency is not that difficult, but it is of academic interest only once that woofer system is placed in a room: At low frequencies, the room’s effects are totally dominant.

There is usually a LOT of room for improvement at low frequencies. Equal-loudness curves predict that the ear is especially sensitive to differences in SPL (peaks and dips) at low frequencies. A 5 dB change at 40 Hz sounds like a doubling of loudness, the same as a 10 dB change at 1 kHz. (This also explains why it takes so long to fine-tune the level control on a subwoofer system - a small change in SPL makes a disproportionate change in perceived loudness.)  Therefore, smoothing the in-room bass makes a greater subjective improvement than we would have expected from eyeballing the before-and-after curves. 

We all want "fast" bass, but what is often not appreciated is that "smooth bass" IS "fast bass". Literally. Because speaker + room = a linear phase system at low frequencies, the time-domain response and frequency response track one another. Fix one, and you have fixed the other. Because room effects are dominant at low frequencies, the most direct path to "fast" bass includes addressing those room effects.

The precise details of the room’s effects differ from room to room, but the basic issue of room-induced, large, highly audible peaks-and-dips is pretty much universal, and there are similarly universal solutions. Remember this is an acoustic problem, so it is most efficiently addressed with an acoustic solution.

One way to get smooth in-room bass is the distributed multi-sub system. Inevitably, each of the subs generates a unique in-room peak-and-dip pattern (and this is true for any listening location within the room). BUT the SUM of these dissimilar peak-and-dip patterns is significantly smoother than any one of them on its own.

An example of a non-acoustic solution would be equalization. When we fix the response of a single subwoofer at a single listening position with EQ, we are (almost inevitably) making the response worse somewhere else. And as we widen the area where we want to make an improvement with EQ, we reduce the amount of improvement that can be made. Distributed multisubs + EQ can work REALLY well, because the multiple subs significantly reduce the spatial variation in frequency response (in addition to making the frequency response significantly smoother), such that if we still need EQ, chances are it will be addressing a global (room-wide) problem, rather than a local one, so it will not be making the response worse elsewhere in the room.

In my experience - which admittedly includes a disproportionate amount of work with distributed multisub systems - a good distributed multisub setup is more effective from a sound quality standpoint than EQ or room treatment alone... though of course the use of one does not preclude the use of the others. Deepest loudest bass for the dollar comes from using a single equalized ubersub, but quantity without quality becomes fatiguing over time.

There are quite a few different ways to implement a distributed multisub system. The main points are, use enough subs (small ones are fine) and get ’em spread out.

Imo, ime, ymmv, etc.

Duke (yeah I got a dog in the fight... four small ones, actually...)

geoffkait wrote: "I hate to judge before all the facts are in but it appears a big advantage of headphones is you can get very good bass performance without all the angst, effort and cost oft required to obtain very good bass performance for speaker systems. And that’s if you’re lucky and don’t actually make matters worse."

Headphones are a lot of fun, but they are not good for accurate bass reproduction. This is at least in part because a) we perceive bass, and in particular deep bass and impact, with our whole bodies - not just our ears; and 2) there is no room reverberation tail on the notes, and room reverberation done right improves our ability to perceive pitch accurately, plus room reverberation is a component of perceived loudness (sounds that last a little bit longer are perceived as being louder).

If headphones were good for bass, they would dominate in the recording industry. Mixing and mastering would be done on headphones. Instead mixing is usually done on small nearfield monitors, and mastering on big main monitors, and if headphones are used at all, they are never relied on to tell the truth in the bass region. Some beginners hope to rely on headphones for mixing in their "budget" home studios, and you can find and follow their painful learning curves on prosound forums... you know, where those other Acme graduates end up...

Duke

"This is one reason I have a *pair* of Golden Ear Triton Reference. Each one on its own eliminates the need for a sub, let alone the two of them combined." - gdhal

Yup, when it comes to using multiple bass sources in the pursuit of in-room smoothness, the more the merrier.

In general, with intelligently distributed multiple subs (and there are different distribution strategies), the in-room bass roughness is approximately cut in half for every doubling of the number of subs. So two subs are potentially twice as smooth as one, and four subs are potentially twice as smooth as two... and eight subs are grounds for divorce in most states.

"audiokinesis, thanks for the Acme Audio Engineering School lecture." - geoffkait.

I flunked the Acme Laws of Physics class... which teaches that gravity doesn’t start working until you look down...

http://commentphotos.com/gallery/CommentPhotos.com_1407347817.jpg

Duke

Skanda wrote: "Do sealed speakers integrate better into these rooms?"

It’s ALWAYS a matter of speaker + room, and if the room is contributing a lot of boundary reinforcement, sealed boxes generally result in better synergy. If not, then vented boxes generally result in better synergy. I can go into more detail if you’d like, BUT the specifics ALWAYS matter more than these sort of generalities.

* * * *

A lot of different suggestions have come up in this thread, things that have obviously worked for the people who tried them. Because of the ear’s exaggerated sensitivity to small changes in SPL at low frequencies (shown by the bunching up of equal-loudness curves south of 100 Hz), combined with its poor time-domain resolution (which is why we can’t localize the source of very low frequency sine waves), we can infer that the improvements we hear are primarily due to changes in SPL, even if they "sound like" changes in "speed".

At the risk of over-simplifying, and being open to correction on any of these points, some things simply make a bigger difference than others. To set the stage, it’s not uncommon for in-room response across the bass region to have peak-to-dip swings of ballpark 12 dB, or +/- 6 dB. Changes to speaker damping (mechanical or electrical or acoustics) seldom result in more than a 1 dB difference, but in the low bass region that's as audible as a 2 dB change in the mids. Improvements to room acoustic damping (bass trapping) can result in 2 dB or maybe 3 dB reduction of the maximum peak-to-dip swings. EQ is generally good at chopping off the peaks but not so good at filling deep dips, still +/- 3 dB is often feasible, and perhaps better if optimized for a small sweet spot. A distributed multisub system results in smaller and more numerous peaks and dips (which has psychoacoustic benefit), with +/- 3 dB over a wide listening area being reported by many users. Remember, smooth bass = "fast" bass, perceptually.

The good news is, these different approaches are not mutually exclusive. You can start with one and then add another as your piggy bank recovers.

One final implication of the bunching up of the equal-loudness curves south of 100 Hz is, there is subjectively a LOT of room for improvement over the typically poor low frequency response of most speaker/room combinations.

Imo, ime, ymmv, etc.

Duke

gkr7007 wrote: "...I gravitate toward planars and other very fast response speakers."

Planars are subjectively "fast" because they have smoother in-room response than monopoles (even though the actual low-frequency transient response of their diaphragms is often quite poor, certainly not "fast" at all). This smoother in-room response of dipoles arises from the 180 degree phase difference between the backwave and frontwave, which effectively launch in opposite directions. When the frontwave and backwave meet up again, after several bounces off of room surfaces, their phase response is significantly more randomized than would be the case for a monopole speaker’s room bounces. And the sum of highly random-phase bass energy is much smoother than the sum of largely in-phase bass energy. "Decorrelation" is the proper word... decorrelation = smoothness, and is highly desirable in the bass region, and is something big rooms do better than small rooms. Decorrelation is also the advantage that a distributed multisub system offers over a single big sub... same basic mechanism as planars, but set in motion by different means.

As the wavelengths get very long relative to the room dimensions, planars tend towards cancellation because half of their in-room energy is out of phase with the other half, so planars don’t make very good subwoofers unless they are very big and can move a lot of air, and are in a big room.

In general, two intelligently-positioned monopole subs approximate the in-room bass smoothness of a single dipole main speaker. So it takes four intelligently-positioned monopole subs to approximate the in-room smoothness of two dipole mains speakers.

Duke

Soundrealaudio wrote:  "[Woofers] must use very stiff suspensions to bring them back to neutral position after each excursion."

The suspension system is not what brings the woofer's cone back to the neutral position.  IT IS THE AMP!!  As long as the amp is sending signal to the woofer, ITS MOTION IS ALWAYS BEING POWERED BY THE AMP!!  It never has to rely on the suspension system alone.  In more technical terms, Qts is almost always dominated by Qes; Qms typically makes only a minor contribution to Qts. 

Duke

Willemj wrote: "I have tried to imagine what is meant by fast or slow bass. It cannot be the speakers themselves for reasons that have been explained. My hypothesis is that what people are referring to is really the delay that is visible in waterfall graphs of bass response in real rooms: the reflected sound of room modes lingers on."

Yes!!

What the ear interprets as "slow" is all happening on the trailing end of the notes. It is ALSO showing up as a frequency response peak, as you are about to see (and, this is the key to the in-room bass puzzle):

Speakers + room = a "minimum phase" system at low frequencies, and what this means is, the time-domain response tracks the frequency response, and vice-versa! In other words, where you see a slow-decaying ridge of energy in a waterfall plot is also where the system has a frequency response peak!

It gets even better: If we fix the one, we have SIMULTANEOUSLY fixed the other! So if we improve the decay time via bass trapping, we have simultaneously improved the in-room frequency response. And if we improve the in-room frequency response via EQ or distributed multisubs or whatever, we have simultaneously improved the decay time!

So any talk about the bass "speed" of a small woofer vs a large woofer is coming from an incorrect paradigm. The correct paradigm is, the in-room frequency response is marching in lock-step with the in-room decay times. THAT is the only "speed" that matters in the bass region, and we can fix it by fixing the frequency response!

* * * *

In these internet forum discussions it can be hard for observers to discern which posts contain accurate information. I don’t open up my trophy case very often, but given the myth-busting theme Eric envisioned for this thread, I think it may be relevant: A subwoofer system I designed using the principles I’ve posted about in this thread received a "Product of the Year" award from a major magazine (The Absolute Sound, 2015). This doesn’t definitively prove that the principles I’ve described are correct, but it does raise the possibility.   If so, then credit to my teacher, Earl Geddes. 

Duke

dealer/manufacturer

Thank you, willemj. 

The amp I supply with the Swarm has a single band of parametric EQ, but most people don't use it.  The only time I've used it has been to extend the very bottom end a bit when all of the modules were in sealed-box mode,  in a situation where we really should have left some of the ports unplugged but the customer insisted they all be plugged.  We used a real-time analyzer but didn't save any of the curves.  I haven't heard from anyone who has made measurements with the parametric EQ in play.  I have heard from multiple customers reporting +/- 3 dB in-room across the bass region, with the -3 dB point clocking in a bit south of 20 Hz, without EQ.  

One of my customers had been using a Meridian processor that had been professionally calibrated for his previous sub.   He reset all the filters to flat and called in the technician after he had set up the Swarm.  After making his measurements, the only thing the technician did was adjust the level a bit.  No further equalization was needed.  Apparently the technician said he'd never seen anything like it, and he'd been doing this for many years.   

But it's not either/or!   I am sure the Swarm would work very well with EQ, especially something well thought-out like the Antimode, in part because the spatial variation (change in frequency response at different locations) is greatly reduced by the distributed multisub configuration. 

Duke