Hi all,
Thought this might be an interesting discussion.
I'd like to start with an observation: I think I can reliably hear my subwoofer's location.
Context: 2x Genelec 8030C, 1x Genelec SE7261A, driven from an RME UCX-II. I've used the built-in RoomEQ to implement a 4th order 60Hz lowpass on the AES feed to the subwoofer. The frequency response is good and flat, extending down to just below 20Hz in-room. I've also used the sub's built-in lowpass at 85Hz to absolutely minimise any content in the >100Hz range, where people tend to be able to locate sources very easily.
The subwoofer is a little off to the side from the centre. During music playback, its location is very audible to me, to a point where I'll often leave it switched off to reduce the distraction of it being "wrong".
To expand on my observation a little, test tones sound as you'd expect: they fill the room with no obvious origin, and if I walk around, the level comes and goes.
Playing time-varying signals (music playback, repeated tone bursts, etc) with only the subwoofer switched on, I can point to the source of the sound reliably. If I spin on my chair, the source seems to spin - as would be expected.
I've tried moving the sub to the other side of the TV, re-did the crossover etc etc - same result.
This has been a problem for me in the past, but I'd typically attributed the issue being harmonic distortion generated by the (lower-quality) subwoofer(s) providing me with location cues from the higher-frequency output. The SE7261A, though, is rated for low distortion, and I'm running at very moderate levels. I believe the sub is working correctly, too: no nasty noises etc, even turned up really loud. It sounds clean and flat down to VLF. I can just hear where it is.
You're welcome to tell me I'm crazy. It's a possibility I haven't yet ruled out.
If anyone else has had similar experience, or would like me to do some more testing (I could take binaural measurements to see how much of a level/phase difference there is L-R), let me know.
It has been hard to keep discussions on this topic on track lately. It seems to take some kind of opinion based directions.
A subwoofer at 3 meter listening distance, moved 1 meter to one side of the center, will give you about 8 degrees of phase difference at 100Hz. It will also give an IID of 0,2dB. However, this is given that the room is anechoic. The ITD difference of 76mm is clearly audible at higher frequencies, but one does not need to go very low before this effect stops being reliable.
I will try to elaborate:
First of all, we need to look at how hearing works. For this to work, we need to separate between tactile and aural hearing.
There are some obvious challenges when it comes to testing tactile hearing only. We could in theory isolate the head while using heavy duty ear protection. On the other hand, we have a similar problem with isolating hearing. It is not as easy as just using headphones as most signal sources are either mono measuring signals, or recordings meant to be played back on loudspeakers, where bass is sometimes panned towards one side. As the left ear will also hear the right speaker, and vice versa, we need to take into account that locating bass with headphones is not the same as locating the subwoofers location.
Then we have the "why could we potentially hear where the subwoofer is?" question. As you have pointed out, the cross over point and steepness is one factor that affects the high frequency content. Another one is the harmonic distortion of the subwoofer itself. But there are at least four mechanisms connected to our hearing that our brain uses to localize sound. We have the ITD (Interaural time difference) IID (Interaural intensity difference) and IPD (Interaural phase difference).
For ITD to work, we rely on some type of time signature. This means we need a range of frequencies, and we need a defined starting point for a given event. A sine wave has by definition no starting or ending point. There are tons of documents trying to explain that ITD with pure sine waves works this or that way, but it is by default a complete misunderstanding. If we delay one side to achieve virtual paning to the other side, we are really talking about IPD, not ITD, as long as we talk about sine waves. All studies that show dependable results within the field of ITD are impulse based, and they show extreme precision when it comes to angular resolution. However, all studies (that I have read) that tries to mix ITD in with low frequency localization fail to keep IPD out of the equation, and very often fail both to keep higher frequencies, tactile experience, distortion and other factors that might skew the result from affecting the result. Like I said in the introduction, it seems to take an opinion based direction, and quasi science is used to defend this position.
I have not yet seen a study that uses ITD correctly to explain any type of sound localization below 100Hz. I believe we will never see such a study.
IID is very reliable at all frequencies, but it is not a very sensitive function of the ear. This means we rely on more than a few tenths of a dB to be able to localize the source. This is also frequency dependent, and we are not at all great at this for low frequencies. I think it is important that if we run a test where the subject is able to tell on what side the subwoofer is located is not in itself a proof that we can reliably hear IID at low frequencies. Firstly, there are numerous other possible explanations. A study that does not take those into account is worthless and has no value for determining if something is correct or not. It is important to note that low frequency IID is not the same in room as in an anechoic chamber. We can measure SPL at 1 meter distance, and at 4 meter, but instead of the expected 12dB loss, we often just see similar results as the room gets pressurized. It is also important to note that we can measure large local differences that follows from room modes rather than from the direction of the sound from the subwoofer. Lastly, as we move the subwoofer further to the side, the difference between the ears does not reach more than around 250mm as it is limited by the width of the head. At 100Hz the distance between a peak and a null is just 85cm, meaning it is far more likely to cause significant ITD than the direction of the incoming direct sound.
I have not yet seen a study that uses IID correctly to explain any type of sound localization below 100Hz. I believe we will never see such a study.
IPD is very reliable at 180 degrees at all frequencies except very high frequencies. This is likely because as the wavelengths are so small, we most of the time have several periods of phase difference between the ears, and this information is therefore not useful to located sound. At frequencies below 100Hz, we rarely ever have this experience in real life as it requires at least a 1,7 meter difference, and head dimensions at that level is quite rare.
There are numerous studies that claim that IPD is reliable down to as low as 1 degree. However, the way the studies I have read are set up, they do not fully rule out other factors. The contribution of higher frequency distortion, the start of tones etc all point towards that these results are not as reliable as the claims might suggest. One obvious problem with tests like this is that while we do hear something going on when we change the phase, we do not reliably determine the direction of the angular difference between the two signals, especially for pure sine waves.
Then we have the issue of room modes blending in. We will most likely have a null at one ear at one frequency, while the other ear do hear some direct or indirect sound. Close to this frequency, we are likely to have a 180 degree difference between the ears even if the wavelengths are far to big for this to happen by angular direction as signals in and out of phase are mixed unequally on the two sides of the head. However, this effect is even more prominent when we leave the aural sense.
So when we move over to tactile experience, I feel more confident, as our bodies are larger than our heads, and they play a very important role when it comes to low frequencies in general. Using just well calibrated tactile drivers, it is really hard not to be tricked into believing you actually hear bass. Our bodies are also quite a bit larger than our heads, meaning we can more effectively get the notion that the bass pressure comes from one side. However, I am still tempted to blame room acoustics rather than the angle of direct sound.
So to conclude, I is not by accident that impulse response based room correction at frequencies below 100Hz is actually very efficient. Also, there are tons of examples where the bass sources are symmetrically placed, but the room is asymmetrical, leading to an asymmetrical offset of the room modes. As room modes are typically very high Q, they are also by nature very narrow band. This means that you (
@ChrisG ) may find the same effect that you experience when playing music if you step through the frequency range really slowly. You are likely to find one, or probably more, narrow points where a strange sideways effect is clearly audible with a pure sine wave. You will also find that just above and below this point, the changes may be quite radical. I think that if you start looking for a location where the 60-100Hz range is really pure, you will get less of these issues, even if it means the sub has to go somewhere you did not expect it to be any good. I would also encourage you to turn off the internal cross over and even experiment with less steep filters in order to reduce the phase shift and improve the summation between the sub and the speakers. If you run two 4. order filters in the sub, one 4. order filter in the fronts, and also the bass reflex tuning adds another 4. order filter, you have a lot of phase shift, and not very low in frequency. This is typically an acceptable compromise in compact pro audio systems, but for hifi, this should be avoided if possible.
O/k I know that "being argumentative" can have a positive and a negative connotation. Your're making fun of me an little. That's o/k. The both of us leave it at that.