Rene, If there is no centre speaker, then the centre channel content is split evenly between the front left & right channels. This is done automatically as a function of setting the speaker configuration, so there's no need to manually control centre channel steering.

BTW, what do you mean by digital and analogue matrix modes? All the matrix decoding and surround processing in the 990 is done digitally. PLII, PLIIx, Neo:6, etc, have always been implimented in the digitial domain (not analogue) from the very begining when they were introduced.

If there is no centre speaker, then the centre channel content is split evenly between the front left & right channels. This is done automatically as a function of setting the speaker configuration, so there's no need to manually control centre channel steering.

Except I can't seam to tell the 990 that there is no centre speaker!

BTW, what do you mean by digital and analogue matrix modes?

Dolby PLII and PLIIx are matrixed into two channels, no? Historically Dolby Pro Logic decoding was done in the analog domain, IIRC. So, I tend to think of matrixed multi-chanel audio as "analog", because it can be carried over a conventional stereo analog communication channel.

Dolby Digital and Dolby Digital EX (and other) formats are discrete 5.1 and 6.1 channel formats, often found on digital media. (Naturally, one could also have PLII and PLIIx on two digital media audio channels).

Digital processing is possible for all these formats. It's just that it was not done that way originally for plain old PL.

Rene, But that's a bug in the speaker configuration function. The solution is to fix the specific problem (i.e., allow the centre to be set as 'none').

End user control of centre channel steering is a useful feature when the centre is derived using steering, but it is not the solution to a speaker configuration problem. PLIIx can be applied to 2-channel and 5.1/6.1-channel material. It was when Pro Logic first came out, but since 1988 that decoding has been done in the digital domain. That's 17 years ago. Other matrix decoders (PLII, PLIIx, LOGIC7, Neo:6, Circle Surround II, etc) have all been digital since introduction.

But that's a bug in the speaker configuration function. The solution is to fix the specific problem (i.e., allow the centre to be set as 'none').

True, but if steering is available at all, what earthly reason should there be to not permit it in all modes? While likely necessary for PLII and PLIIx decoding gone awry (i.e. from ordinary stereo sources), I can see it useful for discrete sources for two reasons: (a) to accomodate for badly encoded material, with too much folded into the center (a PLII to DD transfer, perhaps, if unlikely?); (b) (which is more likely), a customer using a ribbon centre channel speaker, like the BG Radia 220i. Such speakers have a cylindrical dispersion pattern, and one would think that a centre speaker would have the ribbon driver mounted at 90 degrees to offer a dispersion pattern rotated with respect to its orientation (i.e. horizontal in the room). In fact, this is not done: a vertical (non-rotated with regard to the speaker) dispersion pattern is used for the benefit of listners straight ahead of centre. This avoids listeners to the side placing the center speaker off to the opposite side. But, the signal drops off rapidly to the sides (so instead of misplaced center, you get almost no center). Blending some of the center signal to left and right avoids this problem.

Some go so far as to argue that, with ribbon L+R mains, one should not have a centre channel speaker with high frequency response, or any at all. I can vouch for this: I'm currently running a 2.1 setup with Radia 520i mains and a sub, with a stereo downmix sent to the mains, and centre is dead centre even up to 30 degrees off-axis. One would think that centre would move closer to the nearer speaker in this case, and it does, of course, but the effect is attenuated by the fact that ribbon drivers' output falls off at 3 dB as one doubles the distance, not 6 dB, if one would have a spherical dispersion pattern (i.e. with a conventional speaker).

In lieu of a fix for the bug, it is frustrating to not be able to hack a fix by using centre channel steering in all modes.

PLIIx can be applied to 2-channel and 5.1/6.1-channel material.

Yes, but it always struck me as silly to use PLIIx to go from 5.1 to 7.1.

It was when Pro Logic first came out, but since 1988 that decoding has been done in the digital domain.

Granted.

I just always thought of matrixed (as opposed to discrete) surround modes as "analog" because they could be recovered/processed from two-channel analog matrixed mixes (i.e. the stereo audio connections from a DVD player). That was never done for discrete multi-channel audio carried over analog lines, because it was already decoded. I suppose the closest would be external bass management/processing for signals from 5.1 analog outputs from a DVD player, but even here, the decoding has already been done.

That's 17 years ago.

You haven't been around very long, have you? At my age, 17 years ago feels like "yesterday".

Rene, In this specific case, what you're asking for is the ability to control how front channel content is distributed amongst the three front speakers. With matrix decoding of 2-channel material, implimenting this feature is not difficult nor unreasonable. I have no problem with more options.

The centre channel is going to be extracted using matrix decoding technology, which is not 100% perfect. At some point, the processing has to choose how much gets extracted. It's a balancing act: too much extraction and you collapse the front soundstage to the centre; too little extraction and you're stuck with the very phantom imaging you're trying to avoid.

Since this balance can vary by source and, more importantly, personal taste, it makes sense to let users control the extraction level. With movies you can set it higher for clean dialogue; for music you can set it lower for a more continuous front soundstage.

But in the case of discrete multi-channel, what you're asking for is the ability to literally create a personal of remix the front three channels. After all, the three channels are normally routed to their respective speakers with no use (or need) of processing. So what you want to do is redistribute the contents of those channels.

The solution for you might be to actually buy an inexpensive mixer, where three channels come in and three channels go out, but the content is remixed (with the mixing being controlled by you). Remixing the front soundstage is not a solution to problematic speaker dispersion, any more than it is for a speaker configuration bug. Those problems should be dealt with, not covered up. Huh? Since there is no discrete 7.1 content, how else can you play back 5.1-channel material on a 7.1-speaker set-up? What specifically do you find "silly" about using PLIIx on 5.1 sources. That's interesting. I guess I never associated "digital" or "analogue" as having any relation to something as arbitrary as number of channels. I've always viewed them as being orthogonal and really don't see what one has to do with the other. But that's just me.

But in the case of discrete multi-channel, what you're asking for is the ability to literally create a personal of remix the front three channels. After all, the three channels are normally routed to their respective speakers with no use (or need) of processing. So what you want to do is redistribute the contents of those channels.

I agree that remixing the front three channels is useful when it comes to matrixed formats and is not "necessary" with discrete formats. However, I presume that, given the ability to remix for one (in the digital domain), it is trivial to remix for all. If that can improve support for 2.1, or 4,1 speaker configurations, why not offer it? I suppose one can argue that 2.1 or 4.1 systems are rare and so are not worthy of support, but again, if it's trivial to do, why not?

The bottom line is that I'd like the remixing capabilities to be orthogonal to the source formats even if they make more sense for some formats than others.

Remixing the front soundstage is not a solution to problematic speaker dispersion, any more than it is for a speaker configuration bug. Those problems should be dealt with, not covered up.

I would not consider cylindrical dispersion patterms problematic. In fact, they neatly address the problems of ceiling and floor bounce (for the mains) when compared to sperical dispersion patterns. A rotated dispersion pattern for the centre speaker is indicated as a matter of course. Remixing the front three channels would be a way to tweak the effect this has, but is not essential in typical rooms. I mention it as a possiblity, not a requirement.

The solution for you might be to actually buy an inexpensive mixer

No, the solution is a center channel speaker. (I think we'd agree here.) The BG 220i runs around $1800, however, so it will be a while before I purchase one. I would consider that a decent mixer would cost as much, unless I built it myself.

What specifically do you find "silly" about using PLIIx on 5.1 sources.

I'm of the opinion that the PLIIx processing would do more harm than good. I can understand PLIIx to go from matrixed two channel to five speakers and a sub. I can't understand the benefits of using it to go from five discrete channels to synthesize two more. I'm of the opinion that the resulting mangling of the surround channels as provided would do more harm than good. PLIIx on 5.1 to 7.1 strikes me as an excuse to justify two rear speakers instead of one.

I never associated "digital" or "analogue" as having any relation to something as arbitrary as number of channels. I've always viewed them as being orthogonal and really don't see what one has to do with the other. But that's just me.

History has a way of shaping one's perceptions. It's not so much the number of source channels, as it's the fact that matrixed channels were first distributed and decoded in the analog domain. So, when I read "matrixed", I think "analog". IIRC, the matrixing transform is not reversable, which means that encoding for matrixed distribution is a fussy, unreliable business, that also reinforces the "analog" adjective in my mind.

Who are you agreeing with? I said that if centre content is going to be extracted, then it makes sense to allow user control of the extraction level. The extraction process is ocurring anyway.

That's very different from remixing three discrete front channels by bleeding centre content into the left & right speakers. Who says a user adjustable remix capability is trivial? Besides, support for 2.1 and 4.1 speaker configurations already exists (and the solution is not a variable bleeding of the centre channel content, as you're proposing). Buy a mixer. After everything companies like Dolby and DTS have done to introduce a discrete centre channel to consumers, they're not about to create technology that lets you spread discrete centre content across the front soundstage. That goes against all they've worked for since the Pro Logic era. I would, especially based on your description: "the signal drops off rapidly to the sides (so instead of misplaced center, you get almost no center)".

The very point of using a centre speaker in home theatre is for the benefit of off-axis listeners. If you are using a centre speaker whose output basically disappears for off-axis listeners, then you most definitely have a speaker dispersion problem. Remixing the front three channels is not a solution to a center speaker with problematic dispersion. By bleeding the centre content to the other two speakers, you will be reproducing the dialogue in triple-mono.

This will not only introduce comb filtering and problems with dialogue inteligibility, but will cause the location of the dialogue to drift away from the display and towards the speaker that is nearest to you. After all, you no longer have the dialogue locked in the centre speaker only.

Again, this goes against everything that Dolby, DTS, THX and others have tried to do for dialogue reproduction at home. So don't expect any of them to come out with processing that allows you to remix the front three channels. A quick clarification on nomenclature: PLII creates up to 5 output chanels; PLIIx creates 6 or 7 output channels. The sub output is not created by the processing but by the bass management system in the receiver. There are several reasons for steering 2 surround channels over 4 surround speakers:

Localization: A car leaves the left side of the screen, the sound disappears to your left. A plane flier overhead, the sound disappears behind you. Even the two best surround speakers in the world can't be in two locations at once (at your sides and behind you).

Stability: No matter where you're sitting in the listening area, sounds that are intended to be heard from behind you are always heard from that direction. No magic involved, just a pair of speakers physically placed behind you (makes it hard for those sounds to come from any other direction).

Envelopment: Four surround speaker can literally 'surround' you better than two speakers can. What "mangling" are you talking about? If you were sitting in the sweet spot, right in between the two surround speakers, any correlated mono surround information would end up phantom imaging behind you. If you're out of the sweet spot, those sounds collapse to the nearest speaker.

PLIIx extracts those sounds and sends them to speakers behind you. Now even if you're out of the sweet spot, those same sounds will always image behind you.

Same exact directionality, just greater imaging stability. You really consider this "mangling"? That makes no sense. Dolby and DTS and THX were already recommending the use of two rear speakers for EX/ES playback. Dolby extended PLII to PLIIx because they noticed that 7.1 set-up were catching on with consumers. The fact that people were using two rear speakers was the justification for PLIIx, not the other way around.

Finally, despite your explanation, I still don't understand why you would deliberately refer to digital surround processing as "analogue". Again, it's probably just me.

Sanjay took my quote about remixing the front three channels out of context. I said I agreed it was useful for matrixed sound formats. I also believe it is useful for other formats when there is no centre speaker. So, if you can do it at all, why not be able to do it all the time for a 2.1, or 4.1 setup?

Contrary to what you say, the 990 does not support this in non-PLII modes because it does not allow specification of no centre channel. One way to do this is to allow a remix in all modes and, perhaps, set it to 100% center by default to left and right (50% each) for 2.1 and 4.1 configurations.

The need to remix stems from the speakers available, and not any defect in the discrete multichanel formats.

But, if you allow 0-100% remix in PLII why not in all modes, when there is a centre and 100% when there is none? Personally, when watching a movie alone in the "sweet spot" I prefer no centre channel dialog speaker.

The cylindrical dispersion patterns of planar magnetic speakers are not a "problem": they allow for greater sound levels at greater distances with less power because the SPL drops off by 3 dB for a doubling of distance, instead of 6. In their operational range of 150Hz to 40Khz, their distortion figures are amazing. Their dispersion pattern solves the problems of floor and ceiling bounce. Their one flaw is that they generally don't operate well below 250 Hz. However, they offer the big advantage of not requiring a crossover in the sensitive midrange area, common in small two-way and three-way speakers. They are usually combined with woofer arrays (like the Carver ALS, c. 1987 which combined three dipole 12" woofers with a 48" ribbon driver in each speaker), and tended to suffer from inefficiency problems (the Carver ALS were rated 86 dB/W/m) until modern magnets were developed (today 88-90 dB/W/m is common). Still, they like power, and clean power: cheap amps driven into clipping will destroy a ribbon.

The Radia 520i is designed to be crossed to a sub at 80 Hz, for example, and is rated 50-250 W.

However, the center channel dispersion pattern, if unaltered, will place dialog to the left of centre for right off-axis listners and vice-versa. Planar magnetic centre channel speakers contain difusing panels to offset this (and tend to be quite long, often more than a 32" set, to ensure off-axis response). This could also be achieved with a L-C-R blend, if available for other reasons. Either deal with the problem mechanically, or electronically.

The benefit of all of this is amazing sweet spot (which tends to be quite large, actually) imaging. In fact, I've heard PLII material over two stereo planar magnetic speakers with phantom sounds from the sides and rear, if the material and room is just right.

What particular "context"? Apparently you think this features is useful "all the time" (see quote above).

I'll ask again: who is it you're agreeing with? Who else in this thread has said that remixing the front three channels is "useful for matrixed sound formats"?

2.1 and 4.1 set-ups are already covered by initial speaker configuration. The centre channel content is reduced by 3dB and split equally to the front L/R channels. Nothing further needs to be done.

You seem to be under the impression that "remixing" is happening during PLII matrix decoding, when that's simply not true. Only when a centre channel is being extracted can the extraction level be chosen by the user. This is very different from a remix function that would bleed discrete centre channel content to other speakers on a variable basis. That's a bug, which Outlaw Audio will hopefuly deal with. The solution is to fix that bug, not re-mix the source material. There is no "need to remix stems". If the problem is the speakers, then that's what should be addressed. The solution is a centre speaker that actually does the job, not some re-mix function that spreads the discrete centre content across three speakers. Because there is no re-mixing (i.e., re-directing discrete centre content) going on in PLII. This is a feature you've invented in your imagination and are now lamenting for not applied to other modes. It doesn't exist.

And for situations where you have a discrete centre channel but no centre speaker, then 100% of the centre channel content is sent to the L/R speakers. Hey, I'm just going by what you described earlier. Any centre speaker that's essentially useless for off-axis listeners is a real "problem" for any home theatre.

I'll ask again: who is it you're agreeing with? Who else in this thread has said that remixing the front three channels is "useful for matrixed sound formats"?

Why you did, Sanjay: "When processing 2-channel material, the centre channel is derived by steering certain information out of the front L/R channels. It makes sense then that Dolby allows the end user to control the steering (i.e., how much is sent to the centre output)."

I consider that remixing.

You go on to say: "You seem to be under the impression that "remixing" is happening during PLII matrix decoding, when that's simply not true."

Well, yes, I am under that impression. Taking a variable amount of L and R and no C (because there is none), and deriving new L and R and C strikes me as remixing. Perhaps we are stuck on terminology.

Only when a centre channel is being extracted can the extraction level be chosen by the user. This is very different from a remix function that would bleed discrete centre channel content to other speakers on a variable basis.

I don't consider it any different. Consider a 100% extraction of available centre channel signal from a PLII matrixed source. 0 to 100% of that could be mixed back to L and R with anything not mixed back sent to C. Now I agree it likely is not done that way (100% extraction and remix), but it certainly could be, and that would allow for a remix from three discrete sources as well, and therefore a 100% mix of C into L and R for 2.1 and 4.1 setups.

Basically, I'm expecting that the PLII centre channel extraction algorithm, if coded correctly, could do double duty as a downmix from LCR to LR. It's simply a case of software refactoring. The only way this would not work would be if the PLII centre channel extraction algorithm were lossy, but even that could be dealt with by keeping the parts extracted from the left and right channels separate until finally combined to form a derived centre.

That's a bug, which Outlaw Audio will hopefuly deal with. The solution is to fix that bug, not re-mix the source material.

And how else can the bug be fixed, except by remixing the source material? To wit, 50% of C added to L and 50% of C added to R, for 2.1 and 4.1 setups.

Because there is no re-mixing (i.e., re-directing discrete centre content) going on in PLII.

But whatever is going on is idempotent to such a remix and so may be considered as one. Consider L'=L-eLc, R'=R-eRc, C=eLc+eRc, where e is the "extraction level". One can certainly say L"=L-Lc, R"=R-Rc, C=eLc+eRc, L'=L"+(1-e)Lc=L-Lc+Lc-eLC=L-eLC, R=R"+(1-e)Rc=R-Rc+(1-e)Rc=R-Rc+Rc-eRc=R=eRc. That is, derive full centre channel "halves" from a PLII matrixed source, and then remix part of them back to their source channels. For a discrete C, Lc=Rc=C/2.

2.1 and 4.1 configurations then just fix the value of e at 0 (or, for complete remix control, permit it to remain variable).

I am not suggesting the 990 does it this way. I am suggesting that it could, and it would kill several birds with one stone.

And for situations where you have a discrete centre channel but no centre speaker, then 100% of the centre channel content is sent to the L/R speakers.

Exactly! (At least that's what's supposed to happen). Sounds like remixing to me.

Hey, I'm just going by what you described earlier. Any centre speaker that's essentially useless for off-axis listeners is a real "problem" for any home theatre.

The BG 220i is hardly useless for HT. The ribbon driver alone is problematic unless oriented properly within the speaker (which it is, but can be special ordered "the wrong way", IIRC, for vertical mounting of the 220i). I was merely speculating how a remix of the front three channels could also be used to combat a driver oriented "the wrong way" (for example, if one had three identical R32i speakers in the front wall without reorienting the centre ribbon), as a hack, in the same way that it makes 2.1 and 4.1 setups a snap.

This is not as useless as one might think, as it permits accomodation of the degree to which one has viewers off-axis: imaging is shot, to some degree, with a center channel speaker, for the lone listner in the sweet spot. Rarely is dialog spot on centre, and anchoring it there collapses the soundstage for the sweet spot listner to one degree or another.