Amazingly enough, non-localized frequencies when combined with the playback of their higher frequency harmonics really are localized by the brain to the speaker they were crossed from.
But my issue was not with that. It's that I'm wondering how the person programming the digital crossover maintains phase coherence when using different crossover points. People like to complain about comb filtering between individual drivers playing the same content. That's a measurement artifact which varies with mic placement. What about the comb filtering which results in a digitally summed signal comprised of out-of-phase components? There's no getting around that. The only reason wouldn't be noticed is that the bass content of the mains/center/surrounds isn't likely to be the same. But if it ever was and there is a phase mismatch in the cross-overs the resulting output to the sub would have serious artifacts.
Hey Chris, I read this article a long time ago so I don’t know how much of it is still valid but it does talk about your point:
http://www.hometheaterhifi.com/volume_9_3/feature-article-multiple-crossovers-9-2002.html
Mixing high and low frequency crossovers in a multi-channel set up
Once you wrap your head around the fact that in most products you are setting a high-pass on each main channel and a single low-pass on the sub, the use of a different setting for each speaker (or pairs of speakers) no longer sounds like such a good idea (pun most definitely intended). Lets take an extreme scenario, just to illustrate the point.
We set the high-pass on the main left and right to 35 Hz because we think its in the best interest of our massive tower speakers. We set our center channel high-pass to 100 Hz because it isn't very big. What is the subwoofer low-pass in the processor going to be?
If set at 35 Hz to complement the main speakers, the center channel signal will have a huge hole from 35 Hz - 100 Hz. Whoa! Lots of bass on that channel we don't want to miss out on. So let's try setting the subwoofer low-pass to 100 Hz. Oops! Now we have IN-ROOM 6 dB too much from 35 Hz - 100 Hz on the main channels because BOTH the main speaker and the subwoofer are voicing it. You CANNOT correct for this. If you lower the subwoofer level, you lower it for everything, and now you don't have enough bass from the center channel.
By now some of you are thinking, "Why not low-pass a copy of each main channel at the various frequencies I want and sum that with the full LFE channel?". Possible, yes, and if fact there are some SSP models which do this, but at a price: doing so inherently results in frequency response aberrations due to phase issues. Bass is often common to the front three channels and even more often common between the LFE channel and the fronts. Summing different low-passed copies of the same material would by definition result in a messy frequency response. Take the ubiquitous 4th order low pass as an example: At the crossover frequency its phase has come around to 180deg, absolutely inverted (compared to material it is being summed with). Granted the relative amplitude of that low pass at the crossover frequency is down 6dB but is still enough to create the aberration.
The THX design manual references the Dolby Digital licensing manual which mandates that the subwoofer output be arrived at the way it does for these reasons. If there was a better way to do this, without adding a lot of cost and/or making the product overly complex, I think Dolby would tell us.
One alternative found in some decoders is to take a low-pass copy from the center (in our extreme example, at 100 Hz), add that to the front left/right and still high-pass those at 35 Hz, the balance going to the subwoofer (though you still waste 35 Hz - 120 Hz off the LFE channel). This can be both good and bad, depending on the rest of the design:
- Unless proactively addressed, you can still have the phase issues described above.
- When mixing channels digitally, S/N is lost (approximately 6 dB when two channels are added for example), because after the summing, the combined level has to be attenuated to the original level. Might not sound (pardon the pun) like much but its something a designer has to consider when weighing the pros and cons of doing something.
We acknowledge that a different crossover point for each speaker is a desirable thing from the point of view of real world acoustics and dynamics. The different positions of the speakers in the room virtually dictate it, and the various members of a mismatched speaker set will each have different points of intersection for increasing dynamic range and maximizing bass performance. But without also having a selection of slopes in the SSP and some VERY expensive measuring equipment, one is likely to end up further behind than ahead
Like I said above this is and old article so I’m not sure how valid the points still are.
