Subwoofers and Phase Question For You Sub Experts

@rooze asked, "How does running a swarm setup, with 4 subs, affect phase/time integration with the mains? Does it create twice or half the issue or remove it altogether?"

Briefly, a distributed multi-sub setup is quite flexible in positioning and phase. I have yet to have a customer fail to achieve a good integration with his mains. The relative uniformity of the low frequency sound field with a good distributed multisub system means that you aren’t "rolling the dice" when it comes to blending with your mains. 

The minor differences in arrival time at the listening position are a much smaller fraction of a wavelength than what the ear can resolve at low frequencies. How uniformly the low frequencies decay in-room is of greater audible consequence. Speakers + room = a minimum phase system at low frequencies, which means the time-domain response tracks the frequency response. Fix one and you have fixed the other.
@kenjit wrote: "There is no such thing as swarm."

A the risk of stating the obvious, the guy who manufacturers the Swarm begs to differ.

Duke
yes, I’m that guy

Audiozenology wrote: " If you force time alignment, then you force maximum wave reinforcement in the listening position which negates the point of a distributed bass array to even out frequency response by negating room modes."

My understanding is that the in-room frequency response (and therefore our perception) will still be dominated by the steady-state response, to which those initial arrivals obviously contribute, but their synchronization would not negate the benefits of modal smoothing. So I think that as long as we get good modal smoothing there’s nothing wrong with arrival time synchronization, but personally I wouldn’t trade off anything that mattered (including aesthetics) to achieve it. 

Duke

@kenjit quoting millercarbon and then replying:
"

ideally the sound from the sub should reach the listening position at exactly the same time and at exactly the same phase angle as the woofers in the satellite speakers.

"Which is not possible with the duke le june swarm technology"

Of course it’s possible, if arrival time synchronization is your priority when you set the system up. It is not my priority because other things matter much more.

The ear does not BEGIN to have enough time domain resolution in the bass region for small differences in subwoofer arrival times to matter. The ear DOES have PLENTY of resolution in the loudness domain, so frequency response smoothness (which correlates precisely with good decay characteristics) matters a LOT.

Duke

Noble100 wrote: "This knowledge was the impetus for Duke creating his 4-sub Swarm DBA system that has won multiple product of the year awards (2015 and 2019):"

Just to clarify, the Swarm only received a single Product of the Year award, for 2015, from The Absolute Sound magazine.  It has also received three Golden Ear awards and three Editor's Choice awards, including one of each in 2019.  So apparently the concept has a decent shelf life. 

And as I have said before, my product is definitely not the only way that the distributed multisub concept can be employed.  And results can be further improved with EQ and bass trapping.  Nor is a distributed multisub system  necessarily the most practical solution for most people... if it were, I'd be sitting in my corner office smoking a cigar while my minions did all the work. 

Duke

Thank you very much, ctsooner. I am honored that Richard Vandersteen thinks the distributed multi-sub approach (which is actually Earl Geddes’ idea that I’m using with his permission) has merit. 

Duke

These three different approaches (bass trapping, EQ, distributed multisubs) each do something different. The goal is "smooth bass", because "smooth bass" = "fast bass"; it is the peaks in the in-room frequency response which decay more slowly and make the bass sound boomy or muddy or whatever, and each of these techniques contributes to "smooth bass" and therefore to "fast bass".

Bass trapping absorbs in-room bass energy which shortens the decay times and thereby reduces the magnitude of both the peaks and the dips everywhere in the room. (The in-room frequency response tracks the time domain response, and vice-versa, in the bass region.)

EQ is very good at reducing peaks but not so good at filling in dips, which are caused by modal cancellation, and driving that cancellation harder can eat up a lot of power and excursion (a 6 dB boost to fill in a dip would call for a doubling of excursion and a quadrupling of power). EQ of a single subwoofer is most effective at a single location, because the room-interaction peaks-and-dips will be at different frequencies for different locations within the room. So EQ can be great in the sweet spot, but it usually make things worse elsewhere in the room. In general the larger the listening area we try to fix with EQ, the less improvement we are able to make at any specific location within that area.

A distributed multisub system results in a significantly improved peak-and-dip pattern. Each sub contributes its own unique peak-and-dip pattern, but each sub is only contributing maybe 1/4 of the total bass energy, so the net result is many small peaks and dips instead of a few large ones. This can have a non-obvious psychoacoustic benefit: When a peak and a dip are fairly close to one another (within about 1/3 octave), the ear/brain system averages them out. So the perceptual improvement tends to be greater than what we would infer from eyeballing before-and-after curves. And this improvement is not limited to a given location, but extends throughout the room.

Obviously there are differences of opinion about how much subjective improvement each of these approaches offers, and the devils are probably in the details.

But as you can see, these three different techniques are almost asking to be used in combination, as each does something different that is beneficial. The reduced spatial variation of a distributed multisub system makes EQ much more effective over a large listening area, and bass trapping provides smoothing regardless of what else you are doing.

Duke

" Tried the "swarm" approach and moved four subs into the room but unfortunately found out that they all have about the same tops and dips wherever placed (+/- 20 db from 25 to 100 Hz). "

Did each sub have the same in-room frequency response curve regardless of where you placed it, or did each one have a different frequency response curve but with the same amount of variation? (+/- 20 dB... that’s a lot of variation.... I don't recall ever measuring worse than +/- 10.)

Did you measure with all four playing at the same time?

Can you tell us about your room and approximately where the subs were placed within it?

Thanks!

Duke

@noble100 wrote: "... I’m still having difficulty recalling exactly why it’s not important that the timing of the bass soundwaves reaching our ears matches the timing of the midrange and treble soundwaves reaching our ears.... Hopefully, Duke will chime in again and explain it again. I’ll write it down someplace this time if he does."

The short answer is, the ear doesn’t hear small differences in arrival times at low frequencies.

Here’s a much longer answer:

It makes intuitive sense to line up everything up so that the arrival time for the sub(s) is the same as for the mains, but the real world is more complicated. What we overlook is, the effect of the phase response.

Let me give a fairly simple example: Suppose we have a 4th order crossover at 80 Hz (maybe 4th order lowpass filter on the subs, maybe 2nd order acoustic rolloff + 2nd order highpass filter on the mains). With a 4th order crossover the lowpass and highpass sections are theoretically "in phase" at the crossover frequency, but the lowpass section (the subwoofer) is lagging the highpass section (the mains) by 360 degrees... one full wavelength. In order to align their arrival times, the subwoofers would need to be one wavelength at 80 Hz closer to the ears. That’s fourteen feet! (This same principle holds up for shallower slope crossovers and for asymmetrical crossovers, the fraction of a wavelength is less but more math is involved which is why I picked 4th order for this example.)

The ear’s poor time-domain resolution at low frequencies comes to our rescue. We don’t notice that the output from the subs is arriving one wavelength (fourteen feet) later than it should. I’m not saying there would be no subjective improvement from correcting that, but it’s not "what matters most"... which is a topic I’ll come back to later.

It also makes intuitive sense for the output of all of the subs to arrive at exactly the same instant. This is inherently accomplished if you only have one sub, and can be accomplished at one listening position if you have more than one sub. However if arrival time was what mattered the most in the bass region, then one sub would be what sounds the best, especially outside of the sweet spot.

The reason arrival time isn’t what matters most is, the ear has very poor time-domain resolution at low frequencies. This is why we are so poor at hearing the direction of very low frequency sine waves in a room - we cannot separate the first arrival from the reflections. But this makes the ear very forgiving of small timing errors at low frequencies.

A very illuminating study was conducted in which short-duration low frequency signals - including mere fractions of a cycle - were digitally created and played over headphones (to avoid room effects). Listeners were UNABLE to even DETECT the presence of bass energy from less than one full wavelength. Consider how long wavelengths are at low frequencies and you’ll see that, unless your room is very large, by the time you BEGIN to hear the deep lows, that energy has already reflected off of multiple room surfaces. In this context, a difference in subwoofer arrival times which amounts to a tiny fraction of a wavelength is inconsequential.

(Something which to the best of my knowledge has not been studied is what the time-domain resolution is for the tactile -"felt" with the body rather than heard with the ears - perception of bass energy. I would guess that the time window for "perceptually simultaneous" impact is related to receptor nerve and/or neuron firing rates, which I have not studied.)

Of far greater perceptual consequence is what’s happening to the trailing edge of the bass tones... how smoothly do they fade away? (The first time I encountered a designer giving precedence to the trailing edge over the leading edge was Jon Dahlquist. Jon wrote that, in the course of designing the legendary DQ-10, he had to choose between aligning the leading edges of waveforms, or aligning the trailing edges of waveforms. Listening tests led him to the counter-intuitive conclusion that the trailing edge of the notes mattered more. So this concept is applicable elsewhere in the spectrum, but it is especially applicable at low frequencies in the size listening rooms we have in our homes.)

Well it turns out that speakers + room = a minimum phase system at low frequencies (this according to multiple researchers, including Floyd Toole and Earl Geddes), which means that the time-domain response tracks the frequency response. So where there is a frequency response peak, that’s where the energy takes longer to decay into inaudibility (not that it necessarily decays slower; but because it starts out louder, it takes longer to finish decaying).

I said I’d come back to "what matters most" in the bass region, and imo it's the in-room frequency response. This is predicted by equal-loudness curves, which bunch up south of 100 Hz. They tell us that a 5 dB difference at 40 Hz or so is perceptually comparable to a 10 dB difference at 1 kHz! No wonder big in-room peaks in the bass region are so detrimental to sound quality.

So to recap, the ear has very poor time-domain resolution in the bass region, but has exaggerated sensitivity to frequency response errors in the bass region.

The good news is, the ear really appreciates any improvements we can make in the bass region, whether they be by way of EQ, bass trapping, a distributed multi-sub system, or just working with positioning... or any combination thereof. The information I’ve seen leads me to believe that a distributed multisub system usually makes a bigger improvement than these other techniques in an already competent system, but I’m hardly a disinterested observer.

Sorry this is probably a much longer answer than anyone was looking for.

Duke