Ponderings that keep me awake...

[tejanoboy]

OK, for the technical ones out there, 2 questions that have bugged me a bit...
First, Sansui corrected the 90º shifts in the QS ENcoder, with equal/opposite shifts in their DEcoders and receivers, BUT, Dolby MP and PL-II ENcoders ALSO have those shifts, (same direction, too, j+ on left and j- on right)...do some/most/all DPL-II rigs un-do the shift like Sansui did? I've looked inside a couple A/V rigs, didn't see anything that looks like a phase shift network. (not an analog R/C one, anyway)
Second, When "QS Synthesizer" is selected, are those phase shifters bypassed? If the Synth mode works by injecting a bit of out-of-phase into each channel from the opposite channel (as I'm told it does), then the rear channels should end up out of phase a total of 180º (j- in Lb and j+ in Rb) , like EV-4 and DY if shifted the same as in "QS" mode.
In my current location, an odd-shaped room, ears alone can't be empirical, but it SOUNDS like the rear hemisphere is more diffuse when in "QS Synth" mode.
Sorry if these are dumb or silly questions, but I can't afford to go back to the bottle!
tB

 

[Lynn Olson]

It is not essential to have a set of phase-shifters on the playback end, unless you want to have a Center-Back localization. Unlike SQ, phase-shifters are not an essential part of QS, Dolby MP, or DPL II or other "regular matrix" systems. All they do is allow a full circular pan, or a front-to-back sweep. Without the phase shifter array, they become more like EV4/Dynaquad, which works fine for 270 degrees of localization, and do not need phase shifters for encoders or decoders.

If QS, Dolby MP, or DPL II encoded recordings are played back through a static or dynamic matrix without phase shifters, nothing is lost except for the Center Back localization, since the rear speakers are out of phase with each other - while they are in-phase with the fronts. All the -90 and +90 degree phase shift for the rear channels accomplishes is "spreading out" the phase difference, resulting in a 90-degree phase spread between front and rear instead, instead of 180 degrees at Center Back.

The same applies on the encoding end. If a Center Back localization is not needed, the phase shifters are not needed either. It is a little surprising that all this heat and light about phase shifters is just for Center Back localizations, and is not needed for other localizations. For some producers, though, Center Back is important, so commercial encoders cater to their needs.

Assignment of phase shifting in an SQ decoder was a little tricky too; you had a choice between in-phase Center Back (this was the standard setup common to all commercial decoders) with one diagonal split in-phase and the other out-of-phase (and this was clearly audible on some recordings), or inverting the phase of the Left Back channel, which resulted in an out-of-phase Center Back and both diagonal splits in-phase. I preferred the latter - it sounded more symmetric to me - but I was in a minority.

 

[oxforddickie]

I'm sorry to disagree with Lynn Olson, but as the creator of the QS360 & SQ360 scripts i have to say your comment "It is not essential to have a set of phase-shifters on the playback end, unless you want to have a Center-Back localization. Unlike SQ, phase-shifters are not an essential part of QS, Dolby MP, or DPL II or other "regular matrix" systems" is totally incorrect.

There needs to be the correct phase correction to allow for accurate image placement etc.

Again, you don't seem to grasp the correct way of looking at the phase issues created by using Phase Encoding. Your statement "Assignment of phase shifting in an SQ decoder was a little tricky too; you had a choice between in-phase Center Back (this was the standard setup common to all commercial decoders) with one diagonal split in-phase and the other out-of-phase (and this was clearly audible on some recordings), or inverting the phase of the Left Back channel, which resulted in an out-of-phase Center Back and both diagonal splits in-phase. I preferred the latter - it sounded more symmetric to me - but I was in a minority" is iagain incorrect.

During the devolopment of the SQ360 scripts, i looked deeply into the phase issues that many people complained of, but were unable to pin-point. Because of the proccess there will always be a 180 degree phase difference between the Front and Rear, but that is NOT an issue because of the way the phase relationship between sounds based from the Front to Center and Rear to Centre relate to each other.

I performed many listening tests, and the improved scripts were always prefered over the original scripts.

QS360 needed a lot more work due to additional phase shifts caused by the decoding process itself.

If you've not heard my releases (or the two QS albums released by Romanotrax) please do listen to them. They are probably as good as we can get from a system that was not designed to give the results we can now obtain - Near Discrete reproduction with as stable as possible Phase Correction as possible

Oxforddickie

 

[Disclord]

Dolby MP was never meant to create stable 'center' imaging or the kind of 360 soundfield that SQ and QS were designed to achieve. The phase shift used in Dolby MP is to prevent the cancellation of the surround signal with Right Front when sounds are panned front to back or halfway between front and surround - which is the most widely used surround effect - that way, the surround doesn't cancel in mono playback. For a good overview of why/how the phase shifters are used in Dolby MP (PL-II encoding is the same but with a 5db level imbalance) get the Shure AES paper called "A High Performance Surround Sound Process For Home Video" by Stephen Julstrom, JAES July/August 1987.

 

[oxforddickie]

But surely Dolby PL-II has full bandwidth Lb/Rb seperation, so can't be the same as basic Dolby which is bandwidth limited mono. No matter what system, some form of Phase Correction is required to enable some form of image stabiluty or all you'll end up with is a vaig phasey mess, especially when sounds are moved.

 

[Disclord]

The encoding is the same as regular Dolby Surround, just 180 out of phase with an amplitude bias of 5db to encode left or right surround. There is no bandwidth limiting applied during encoding or decoding, but that's it. One of the surrounds, Left Back, is out of phase with Right Back in PL-II Movie mode. PL-II cannot make side-phantoms - it's a 5-point matrix, Left Front, Center Front, Right Front, Right Back, Left Back. All 'front' sounds are in-phase... back sounds are out of phase. This doesn't matter because it's not meant to make stable phantoms between front-back or between the back channels. It's meant to create speaker-feeds, not phantom imaging. The time-delay required for Dolby Surround and PL-II Movie mode also preclude phantom imaging since it imparts great phase shifts between front and back. Look up the Dolby papers and the Dolby AES paper on PL-II and you'll see this is the case. PL-II Music mode is a different animal, meant to synthesize surround from non-encoded sources and thus, the rears are in-phase to create side-phantoms out of pure left or pure right signals.

The modern surround systems are very different from the quad formats - 360 imaging isn't part of their performance objectives. The encoders/decoders simply don't provide that ability. There are some good AES papers from the early 70's by John Eargle and others that explain both amplitude and phase matrixing - Dolby Surround and PL-II are really just amplitude/polarity matrixes, with no phase encoding - the phase shifts used in the encoder are only to prevent cancellation in playback, not to encode stable imaging between speakers.

 

[oxforddickie]

I have no interest at all in Dolby, the reason i added my comments were because of the comments made regarding SQ & QS. I'll get on with DY360....

 

[Disclord]

I have no interest at all in Dolby, the reason i added my comments were because of the comments made regarding SQ & QS. I'll get on with DY360....

I have a copy of an excellent AES paper on SQ called 'Synthesis of 4-2-4 Matrix Systems" that might interest you. PM me your email address and I'll sent it to you.

 

[tejanoboy]

Wow! A nice lively discussion, and I am learning new things from it...thanks to all!
I like ALL quad, and can be thrilled by a good creative use of it, but my BIG love is total REALISM. The goal I will never reach. but never stop reaching for, is a sonic version of the Star Trek Holo-deck, with accurate distance, direction, and character of ALL sounds the ear would hear at the recording site. For my own dumb reasons, I must stick with 2-channel media for now, so a matrix is a must, and from the comments I have read here, and what my own ears tell me, QS WITH the phase corrections is the best way to go...for THAT purpose. I like DPL-II with movies, but it seems to fall a bit short in other areas.
Disclord and Lynn, you seem to be experts with SQ, and I must confess, that system has mystified me forever! Even if I never use it, I am dying to figure it out a bit better. I can grasp the "vector in front/helix in back" idea, but what about OTHER positions, like left-center, or even dead-center? (is that even possible with SQ?)
Sansui had those cool QS stylus-motion charts in their sales brochures back in the day, I would love to see one of those for SQ.
Those AES papers...is there a site where they can be read without joining AES or paying a lot to download them? (I'm still only back to work part-time :>(
I really need to enroll in "SQ-101". I'm not bad with math, but trying to grasp the whole concept of a "spherical matrix" might put me in the funny farm!
OK, I'll back out of here now, and watch for more posts and learn some more cool stuff...thanks all!
TB

 

[Disclord]

You really need to get the primary AES papers from Ben Bauer to find out all about the SQ matrix. The papers show the stylus motions, which don't matter much, plus the phasors of encoding/decoding, which better show whats happening - the SQ code on the Energy Sphere is shown in other papers and Lynn Olson's Shadow Vector patent, which you can download for free online - in fact, the SQ patents are excellent resources too - just look up Ben Bauer and Daniel Gravereaux as Inventors. An easy way to think of SQ is this way; SQ encodes front sounds via amplitude (volume) like ordinary stereo, with full L-to-R separation. The back channels are encoded by using the phase space between L/R instead of the amplitude space like the front channels. When decoded, with a non-blending SQ decoder, the back channels have full L-to-R separation too - there is basically 0db of F/B separation. SQ sacrifices front-to-back separation for full L-to-R separation, since we are more sensitive to L/R than F/B. SQ can encode any position around or 'inside' the quadraphonic array. While others may disagree for various reasons, the fact is, SQ has better stereo performance and 'won' all the FCC tests for stereo fold-down, mono performance and came closest to the discrete original. QS has serious problems in stereo FM broadcasting, as does QS synthesis for non-encoded signals. In the 70's, many FM stations dropped QS broadcasting due to its performance limitations, and they adopted SQ. Sadly, CBS didn't promote it all that hard for FM, and by 1978 or so, had stopped promoting it completely.

The USQ system (Universal SQ) is like the UHJ system - it has full discrete performance or matrix performance depending on the users equipment. It was designed for quad broadcasting or discrete LP (a combination of CD-4 and SQ) but never implemented (like UHJ). The Ghent Microphone for SQ is a kernel encoding system, like the Ambisonic Soundfield microphone, and could accurately encode the space around it.

 

[tejanoboy]

Thanks for the suggestion! I will spend some time at the U.S. patent office site this week-end for sure! I recall hearing about QS's FM problems, and believe that is why the BBC came up with Matrix H...QS with Lt shifted 60º from Rt.
Since I don't broadcast (not that I'll admit to, anyway!), and have no issues with mono or even 2-channel playback, these problems won't be an issue with me.
I think for what I am doing (live recordings mainly), QS is still the way to go. I have to build my own encoder, and if the phase shifters are off a little, it will still work...just be a little closer to RM, that's all. And I don't have a Tate or anything close, but I do have a healthy QRX-5001...afraid QS wins this one.
The stylus motions don't matter much to YOU, but they are a good way for me to visualize what goes on...I've got strange ways of dealing with math, I admit, but they work well for me.
Thanks for the leads!
Now it's off to the USPTO site...
TB

 

[oxforddickie]

"and believe that is why the BBC came up with Matrix H...QS with Lt shifted 60º from Rt"


I'm afraid this is incorrect. The 60 degree thing came about as an easy option to modify QS decoders to "VERY VERY POORLY" decode Matrix H transmitions that the BBC where doing at the time. Unfortunatley, it is knowhere near the way to decode it, Matrix H is far more complex than that.

Some months ago i started a project to accuratley decode Matrix H, basically so i could add David Bedford's "Instructions For Angels" to my releases, and with the help of some well known people from here as a listening panel, i started work. Things were going well, and out of interest the 60 degree shift on the right channel was thrown out straight away by all as awful, with feedback being very positive, but i realised i needed to get a grip on the problems with properly decoding QS (and to a lesser point SQ) before i could continue. Hense the QS360 scripts.

The Matrix H project is still to be completed.......


OD

 

[tejanoboy]

OK, so how is it more complex? Is it a phase matrix? A great-circle ("regular") matrix, or some ghastly combination of both?
Tell me more!

 

[tejanoboy]

BTW, just remembered another reason the phase shifts in QS ENcoding are needed. One of the key components of a Variomatrix decoder are phase discriminators! The logic may NEED those phase differences to determine how to adjust the matrix coefficients moment-by-moment. (Going by a very old description of how the VM operates)
TB

 

[oxforddickie]

It's probably best to read the paper released by the BBC:

http://downloads.bbc.co.uk/rd/pubs/reports/1977-02.pdf

 

[tejanoboy]

It's probably best to read the paper released by the BBC:

http://downloads.bbc.co.uk/rd/pubs/reports/1977-02.pdf

More complex is RIGHT! Weird phase angles, but sounds like it should have been a serious contender in the "matrix wars" of the Quad Age.
Thanks, very interesting read!
TB

 

[Disclord]

More complex is RIGHT! Weird phase angles, but sounds like it should have been a serious contender in the "matrix wars" of the Quad Age.
Thanks, very interesting read!
TB

Matrix H was just too late on the scene - plus, it was always undergoing 'refinement', finally being merged with Michael Gerzon's System 45J, to create System HJ, the 2 channel variant of UHJ. This was around 1981 or so, and all the promotion was botched beyond belief, forever dooming Ambisonics to relative oblivion. Heck, even SQ got used on some LaserDisc's in the early 80's - UHJ never did. And most listeners didn't feel it lived up to its claims - the 2 channel variant, that is. While others disagree, 2 channel UHJ (BHJ) to me, sounds like non-logic matrix, phasey and indistinct, with no localization. While a logic decoder for UHJ was talked about in some of the technical papers, none was ever commercially produced - and for Matrix H, only modified Sansui Vario-Matrix decoders were ever made.

 

[oxforddickie]

"and for Matrix H, only modified Sansui Vario-Matrix decoders were ever made"

And, as proved, they sounded so bad. As i've previously mentioned, the way to decode Matrix H is NOT to add 60 degree to the right channel as they did back then. it was just a quick fix, which failed.

By the way, you mention SQ being used on some laserdisc's, you happen to know which one's?

Richard

 

[Disclord]

"and for Matrix H, only modified Sansui Vario-Matrix decoders were ever made"

And, as proved, they sounded so bad. As i've previously mentioned, the way to decode Matrix H is NOT to add 60 degree to the right channel as they did back then. it was just a quick fix, which failed.

By the way, you mention SQ being used on some laserdisc's, you happen to know which one's?

Richard

David Bowie Serious Moonlight, Dolly Parton Live in London, Kenny Loggins Alive (coked up!), all from Pioneer Artists. Gary Reber did all the mixes. The porno LaserDisc "Insatiable II" is also SQ encoded (Gary Reber did the mix too!). The Pioneer discs don't carry the SQ logo on the jackets, although the on-screen credits note it, but the Beta/VHS Hi-Fi versions had the Tate System/SQ logo along with a blurb about the encoding/decoding.

All were mixed from multi-track to 2-channel SQ using the CBS SQ Position Encoder and the Fosgate Tate 101A decoder for monitoring. David Bowie on DVD has an terrible 5.1 mix that was done using one of those awful "unwrap" algorithms. But, they included the 2 channel SQ mix too, so its still available. Both David Bowie and Dolly Parton were later re-issued on LD by Pioneer with digital sound - sadly, Dolly hasn't been released on DVD. There is one other title on Pioneer Artists, but I can't remember what it is. Ruggles/Reber (Tate Audio) were trying to get SQ with Tate decoding to be the standard for home surround mixes of both music and movies. They were planning a more advanced Tate decoder that had a Center Front output too, but for various reasons, Jim Fosgate dropped all use of the Tate chips and they never got another company to use them in a decoder - from then on, till around 1987 or so, only Dolby used them in the theater decoders. Oh, I also forgot one other title "Olivia Newton-John: Twist Of Fate" - it's more Dolby-like in the encoding because they used Center Back only so it's front and mono surround. But they used the SQ encoder to mix it. Sadly, the SQ mix isn't used on the DVD.

 

[oxforddickie]

Many thanks, it's a shame it didn't catch on

R