Two Subwoofers... Comb Effect

Shadorne, I disagree with the basic premise of the author of that article. His third sentence describes his foundational belief:

"I soon learned that realistic loudspeaker reproduction requires reproduction of the input waveform, which seems somewhat obvious."

Not all that seems obvious is necessarily correct. The ear does not hear waveforms. Instead, to quote Dr. Earl Geddes:

"You are making a huge assumption here that the ear hears "waveforms". It doesn't. It deconstructs the waveform into a pattern of excitations in the ear (along the Cochlea) which are detected in complex ways."

How then to explain away the consistently-noted subjective superiority of low-Q sealed boxes? Doesn't that prove that good group-delay behavior is the key to natural-sounding bass?

Well, let's take a look into what we know to be audibly significant. Is low frequency group delay audible? The body of published research is inconclusive on the subject. Apparently it is on test tones, but not necessarily on music. But we'll concede that it might make a slight audible difference.

Is there anything else that we know makes a significant audible difference?

Yes - frequency response! We know that the large peaks and dips in the bass region are audible if they are far enough apart to not be smoothed out by the ear's averaging characteristic (which averages and smoothes the much more closely-spaced in-room peak-and-dip patterns at midrange and treble frequencies). We also know that broad, gentle trends in the frequency response are even more likely to be audible than are narrow-band peaks and dips, even though the latter look much worse on paper.

Given that frequency response is known to be highly audible, could the subjective superiority of low-Q sealed boxes be related to frequency response issues? At first glance, that doesn't seem to be the case: A vented box that measures "flat" to 30 Hz sounds boomy, while a low-Q sealed box that is -6 dB at 30 Hz sounds tight and natural. But wait - that's not the whole picture! We are leaving out a very important piece of the puzzle!

And what is that? The room, of course! Typical room gain is +3 dB per octave below 100 Hz. So, our vented box is probably up 5 dB at 30 Hz - no wonder it sounds boomy. On the other hand our low-Q sealed box is -1 dB at 30 Hz, which is much better.

So once we examine bass reproduction taking into account the room's effect on frequency response, the subjective superiority of a slow-rolloff bass system (low-Q sealed box) makes a lot of sense.

But wait - wouldn't the ideal be -3 dB per octave rolloff? How about an ultra-ultra-low Q sealed box? It turns out that the shallowest rolloff you can get from an unequalized sealed box is about 4.5 dB per octave. With equalization, it would be possible to achieve a -3 dB per octave rolloff starting with a more conventional sealed box alignment (assuming adequate excursion capability). But ironically, a unequalized vented box can be designed which comes very close to the theoretical ideal -3 dB per octave rolloff, at least down to system tuning.

Here's a link to a subjective evaluation of a room-complementary-tuned multisub system:

http://www.audiocircle.com/circles/index.php?topic=60103.20 (scroll down to post by ro7939, near the bottom of page 2)

Now I would agree that doing the same thing with large-magnet, large-displacement woofers in room-complementary-equalized sealed boxes would be even better. But, it would cost many times more. When comparing bass systems, there has to be some apples-to-apples basis, or else the biggest and/or most expensive always wins. We must compare approximately equal dollar solutions, or equal size solutions, or equal output-level solutions, or something like that.

So to recap, yes low-Q sealed boxes sound better than most vented boxes, but I believe the reason is that they produce a far more desirable in-room frequency response, rather than their superior group-delay performance making an audibly significant difference. And there are more cost-effective techniques for acheiving a desirable in-room frequency response.

Duke

Interesting post Duke. What do you think about the concept expresed in the "subwoofer conundrum" under the subheading 'Timing is everything' at www.sound.westhost.com/subcon.htm It's not the greatest techno reference but it's quick and expresses what I was trying to get at with variable distances.It mirrors what I have experienced in my home and what the folks at jl audio preach. Thanks - Jim

Aldavis, in my opinion that author makes a mistake in assuming an anechoic environment for the sake of simplicity. I believe the room's influence to be the dominant factor in determining the perceived low-frequency characteristics, assuming competent speaker system design.

Apparently the author is aware that by the time we even hear a low frequency tone, we're well past any sort of "first-arrival" time window and into the time region where the frequency response is totally dominated by the room response. I quote him here:

"A 40Hz bass fundamental cannot be fast or slow - it is simply a 40Hz (transient) tone, and our hearing is depressingly bad at even hearing such frequencies until they have been present for several cycles."

Why he then proceeds to focus on the first less-than-a-cycle, which he pretty much just told us is virtually inaudible (or audible only insofar as it makes up part of the first several cycles), I do not understand.

Also, I've never seen a setup like his "Figure 1 - Typical Listening Room Setup". That looks somewhat contrived to me, probably to better make his point; but I certainly wouldn't call it "Typical".

I'm sure it's possible to come up with a positioning scheme and crossover frequency and slope that gives some sort of worst-case scenario, but consider this: The lowpass driver will be phase-lagging the highpass driver by 180 degrees for a second-order crossover, and 360 degrees for a fourth-order crossover. So the argument might well be made that the subs should be placed either 1/2 wavelength closer to the listener than the mains at the crossover frequency, or 1 wavelength closer to the listener, depending on the crossover type. I wouldn't make that argument personally, but my point is that placing the subs the exact same distance away as the mains doesn't time-align them anyway, because there is still the issue of a frequency-dependent time delay (a phase lag).

I think that if you can get the in-room frequency response in the bass region correct (and not just for one microphone location), you've already fixed all the major problems. At that point timing might become an audible issue, if the initial setup was something like Figure 1.

Duke

Thanks for the thoughtful answer Duke. Your posts are always well resoned. Reguarding your 1/2 wavelength remark I agree that you need to compensate for second order crossovers by inverting phase( 180 degrees available on most subs). Once you have done that does the distance argument not hold ? You can attempt to compensate for various distance discrepancies by utilizing continuous phase adjustment on the sub but you have to pick a frequency to correlate this with.( adjusting to be in phase at one frequency will take you further away in another. Figure 1 I think is an exagerated 'sub in one corner placement'. Your dead on with the major deal being in room frequency response at all likely listening positions which far outways this one argument. I just think this is a simple practical place to start. -
jim

Duke,

Thanks - that we hear frequencies or cycles and NOT transients in the bass makes sense - you have convinced me that John Watkinson may be barking up the wrong tree. Your contention that integration is better with a sub close to the mains when using only 12 db/octave cut-off makes sense (because of higher frequencies above sub cut off).

However, there must be a limit to what extreme phase shifts can be allowed and not mess things up. Perhaps a full cycle is too much? This would be 30 feet of distance at 40 HZ - and so many small ported speakers have a 3 db point at 55 Hz and a port at 40 Hz and will therefore achieve this!

While I can't say for such large 360 phase shifts (are they really audible at all) However, my impression is that an EXTRA cycle is certainly audible/perceptible. In fact, I suspect the overdamped response of a low Q system is so pleasant and "tight" in the bass precisely BECAUSE you get absolutely no extra cycles as you would from an underdamped ported resonant design (the better in room roll off response may certainly be part of it too). So John may be right but for the WRONG reasons?