Is One Subwoofer Really Not Appropriate For 2-Chnl

IME, two subs typically yield better results, because two subs usually provide smoother in-room frequency response.

Having said that, I use a single sub for 2 channel listening, and it sounds great. But part of the reason is that I am able to (1) EQ the sub and (2) control the delay on the sub and the mains, so that the two are time aligned to within about a millisecond.

Regardless of whether you use one sub or two, I believe that it is essential to SQ that the sub(s) are time aligned with the mains, either through positioning, or through the use of delay. This is my opinion, and it is a somewhat controversial one, as you can read on another thread. If you cannot position/delay BOTH subs so that they are time aligned, but you can time align a SINGLE sub, then personally I would go with a single sub.

Good luck.

Bryon

Bryon

04-18-11: Roscoeiii
Can you point us to a good explanation of how to time align subs and mains?

04-18-11: Stanwal
There is a good set of set up instructions on the Sumiko site.

The Sumiko subwoofer setup instructions can be downloaded here. According to Sumiko...

The optimal position for a REL is in one of the corners behind the main speakers...

No doubt the folks at Sumiko are knowledgeable, but I respectfully disagree with their advice. IMO, placing sub(s) BEHIND the mains may result in good *frequency* response, but it rarely results in good *transient* response.

The reason is because, by placing the sub(s) behind the mains, as Sumiko recommends, the sound from the sub(s) will arrive at the listening position several milliseconds AFTER the sound from the mains. Hence the sub(s) and the mains will not be time aligned.

Some people say that a few milliseconds of time misalignment isn't audible at low frequencies. My experience, and some scientific research, suggest otherwise.

Adjusting the sub's phase will not solve the time misalignment created by placing the sub(s) behind the mains, since the signal that needs to be delayed is NOT the signal for the sub, but rather the signal for the mains. Adjusting sub's phase will do nothing for that.

The time misalignment created by placing the sub(s) behind the mains CAN be solved by digitally delaying the mains. But judging from the systems on A'gon, very few people have that capability. So the most practical way to time align the sub(s) with the mains is...

On the z axis, place the sub(s) roughly coplanar with the mains.

By placing the sub(s) coplanar with the mains, the sound from the sub(s) and the sound from the mains will arrive at the listening position at close to the same time. Hence the sub(s) and the mains will be time aligned, resulting in better transient response. This is audible as better coherence and PRaT.

Keep in mind that subs typically introduce latency into the low-passed signal. Therefore, if you cannot digitally delay the mains, the sub position that results in optimal time alignment will often be slightly IN FRONT of the mains. This is, of course, the *opposite* of what Sumiko recommends.

Another benefit to placing the sub(s) coplanar with the mains is that, in addition to giving better transient response, it will often result in better *frequency* response, for the following reason: Coplanar sub placement minimizes the destructive interference around the crossover frequency. An explanation for why that is so can be read here.

That brings me to Roscoeiii's question about how to time align the sub(s) with the mains. The easiest method is the following…

1. Flip the polarity of either the sub or the mains (but not both).
2. Play a test tone at the crossover frequency.
3. Use an SPL meter to measure the output level.
4. Adjust the sub position (or digital delay, if you have that capability) until you MINIMIZE the SPL at the listening position.
5. When the sub is in place, flip the polarity back so that the sub and the mains are the same polarity.

In steps 1-4, you are essentially maximizing the *destructive* interference between the sub(s) and the mains. In step 5, you are restoring the sub to the correct polarity, which now maximizes the *constructive* interfere between the sub(s) and the mains.

Following this method, the sub(s) should be time aligned with the mains to within about a millisecond or two. You can further improve the time alignment, by adjusting the position of the sub(s) in 1-3 inch increments, searching for the position that results in the best coherence and PRaT. This last step has to be done by ear, so it requires some experience and patience. But it is well worth it, IMO.

You can read more discussion/debate of sub time alignment, transient response, and the limits of human temporal resolution, on another thread.

Hope that helps.

Bryon

Bifwynne - Glad my post was helpful to you. If you can manage it, it would be worth the experiment to move the sub roughly coplanar with the mains and see if it improves the system's transient response. It will be audible as better coherence and PRaT.

The fact that you sum the channels into a mono signal for the sub doesn't affect anything in the procedure I described in my last post. If your sub has a polarity switch, use it. If it doesn't, just reverse the positive and negative leads on your speaker wire for *both* of the mains. Then the mains and the sub will have opposite polarity, assuming that your sub doesn't invert polarity.

Adjusting the sub's phase can be helpful, but not nearly as helpful, IME, as time aligning the sub. As I mentioned in my last post, if the sub is behind the mains, the signal that needs to be delayed is *the mains*. Adjusting the sub's phase will do nothing for that. If you cannot digitally delay the mains, you can time align the system by placing the sub coplanar with the mains, or possibly a little in front of them (to compensate for the potential latency - i.e. delay - introduced by the sub itself).

To sum up, if you can do it, move the sub so that it's roughly coplanar with one of the mains, then follow the procedure in my last post. That should get you pretty close to time aligned. IME, the results of sub time alignment are dramatic, especially when the sub is "fast" enough to keep up with the mains.

Bryon

04-20-11: Dbphd
I wondered why someone from a university in South Carolina would publish such an article in other than JASA. Some of the contentions about levels of neural processing of temporal information seem astonishing, so I wonder if peer review may have been a problem.

The link I provided to the paper by Milind Kunchur was published in the Journal of the European Acoustics Association, which is a peer-reviewed journal, according to this page on their website.

I'm skeptical about your arguments and your source of "scientific' support.

Whether Kunchur's science is valid, I cannot say. I am no expert on auditory perception, but even if the limit of human temporal resolution is twenty times higher than Kunchur claims, that would still place it at only 1ms, in which case placement differences of approximately a foot or more would result in potentially audible time misalignment.

What I can say is that, IME, differences in z-axis sub placement are audible down to about 4 inches. That would correspond to about 25µs.

Regarding the science, you are certainly entitled to your skepticism. Regarding my statements about subwoofer time alignment, I would humbly suggest that you experiment with it before concluding that they are incorrect.

Bryon

Correction to my last post. It should have read...

...even if the limit of human temporal resolution is *two hundred* times higher than Kunchur claims, that would still place it at only 1ms, in which case placement differences of approximately a foot or more would result in potentially audible time misalignment.

AND...

What I can say is that, IME, differences in z-axis sub placement are audible down to about 4 inches. That would correspond to about *300µs*.

I was off by an order of magnitude. D'oh! I should not do math in my head at midnight.

In any case, my error actually *underestimated* the degree to which Kunchur's estimate could be off while still making time misalignments of about a foot potentially audible, leaving my argument for the importance of subwoofer time alignment unaffected.

04-20-11: Dbphd
I wondered why someone from a university in South Carolina would publish such an article in other than JASA.

It's true that Kunchur's article wasn't published in JASA, but here is a list of articles that WERE, all on the subject of human temporal resolution:

--B. Leshowitz, “Measurement of the two-click threshold”, J. Acoust. Soc. Am. 49, 462–466 (1971).

---Ronken, D. (1970). “Monaural detection of a phase difference between clicks”, J. Acoust. Soc. Am. 70, 1091–1099.

--Henning, G. B., and Gaskell, H. (1981). “Monaural phase
sensitivity with Ronken’s paradigm”, J. Acoust. Soc. Am.
70, 1669–1673.

--Plomp, R. (1964). “Rate of decay of auditory sensation”, J. Acoust. Soc. Am. 36, 277–282.

--Penner, M. J. (1977). “Detection of temporal gaps in noise as a measure of the decay of auditory sensation”, J. Acoust. Soc. Am. 61, 552–557.

--Eddins, D. A., Hall, J. W., and Grose, J. H. (1992). “Detec- tion of temporal gaps as a function of frequency region and absolute bandwidth”, J. Acoust. Soc. Am. 91, 1069–1077.

And here is Kunchur's synopsis of the conclusions of those JASA articles...

In one experiment by Leshowitz (1971), listeners were presented with a single pulse or two narrower pulses (with the same total energy) separated by an interval Δt. The click and click-pair could be distinguished down to Δt ≈ 10 μs. In this case, the two stimuli have differences in their amplitude spectra and their discernment was explained on this basis. Isospectral variants of this experiment were carried out by Ronken (1970) and later by Henning and Gaskell (1981) where one stimulus consisted of a short pulse followed by a taller one separated by an interval Δt. The second stimulus was a similar pair with the time order reversed and hence had the same amplitude spectrum. The shortest Δt for which these stimuli could be distinguished was about 200 μs. Another type of constant-amplitude-spectrum experiment involves the detection of gaps in noise (Plomp, 1964; Penner, 1977; Eddins et al., 1992). In these the threshold for gap detection was of the order of 2 ms.

As you can see, those estimates of the limits of human temporal resolution range from 2ms to 200μs all the way down to 10 μs. If we average those estimates, we get 736μs, which corresponds to differences of less than a foot.

Hence there is reason to believe that the human limits of temporal resolution may be sensitive enough to make time misalignments of a foot or less potentially audible.

Bryon



Bryon

I make it a point not to hijack threads. But that is exactly what I've done to this one. My apologies to Ryder (the OP) and to everyone else following along.

Dbphd - Your last observation is a valid one, but I have experiences that are inconsistent with attributing the audible variations resulting from small differences in sub room placement to frequency response. You can read about my experiences on a recent thread I initiated.

I'd be happy to discuss this further, as it is a topic that interests me a lot. But, in the interest of courtesy to other folks, we should move our conversation to the thread I linked above.

Bryon

Ryder - Thanks for being understanding.

Dbphd - I have posted a response to your last post on this thread.

Bryon