SQ - Gain Riding (only) vs. Variable Matrix - Logic Decoding

[kfbkfb]

I thought of this issue a few weeks ago and some posts in another thread caused me to post now.

I used the reader service card in the 1972 H F special 4 channel magazine and received a copy of the AES paper by CBS describing the SQ matrix. The AES paper didn't mention variable matrix decoding of SQ (IIRC, CBS was fixated on claiming infinite LF to RF and LB to RB channel separation), only gain riding logic decoding.

I'm wondering if the most accurate SQ decoding of (early) SQ LPs would be to use gain riding only SQ decoders since these SQ encoded LPs were (most likely) monitored using gain riding only decoders (consumer or CBS Labs professional).

Anyone have a gain riding only SQ decoder and a Surround Master SQ decoder and some of the early SQ LPs to try a comparison decode?


Kirk Bayne

 

[furui_suterioo]

Comparing the "full logic" SQ on the Kenwood 9940 receiver to Tate, it attempts to produce the same image but has noticable laggy pumping and much less seperation, I wouldn't call that more accurate.

 

[Soundfield]

I'm wondering if the most accurate SQ decoding of (early) SQ LPs would be to use gain riding only SQ decoders since these SQ encoded LPs were (most likely) monitored using gain riding only decoders (consumer or CBS Labs professional).

That sort of assumes that there was any encode - decode monitoring! Sort of makes sense to do it that way with hindsight but Is there any evidence that this was a common practice (or indeed ever really happened much at all)? I'm just a bit doubtful, as apart from the additional time and expense, I think that SQ encoding would often have been treated as just another post production engineering processes to be applied to the 4 track mix down to the 2 track master cutting tape.

 

[fredblue]

That sort of assumes that there was any encode - decode monitoring! Sort of makes sense to do it that way with hindsight but Is there any evidence that this was a common practice (or indeed ever really happened much at all)? I'm just a bit doubtful, as apart from the additional time and expense, I think that SQ encoding would often have been treated as just another post production engineering processes to be applied to the 4 track mix down to the 2 track master cutting tape.

steelydave mentioned on QQ at some point that CBS NY had on the fly monitoring so they could hear how their Quad mixes would sound when decoded.

also interestingly CBS were loathe to use any EQ on their Quad mixes lest it create any phase issue nasties that could trip up the SQ system.

in the words of Ian Dury, "there ain't 'arf been some clever b@st@rds!"

 

[Soundfield]

steelydave mentioned on QQ at some point that CBS NY had on the fly monitoring so they could hear how their Quad mixes would sound when decoded.

also interestingly CBS were loathe to use any EQ on their Quad mixes lest it create any phase issue nasties that could trip up the SQ system.

in the words of Ian Dury, "there ain't 'arf been some clever b@st@rds!"

Yes, interesting that. I just wonder how widespread it was around the world (and what notice anyone took of it if they did it!). Given how ghastly some SQ stuff sounds it's difficult to imagine anyone was monitoring anything!

 

[fredblue]

Yes, interesting that. I just wonder how widespread it was around the world (and what notice anyone took of it if they did it!). Given how ghastly some SQ stuff sounds it's difficult to imagine anyone was monitoring anything!

true! i imagine many engineers possibly eventually just focussed on how the discrete came across and if it sounded ok in SQ was seen as a bonus!

 

[par4ken]

That sort of assumes that there was any encode - decode monitoring! Sort of makes sense to do it that way with hindsight but Is there any evidence that this was a common practice (or indeed ever really happened much at all)? I'm just a bit doubtful, as apart from the additional time and expense, I think that SQ encoding would often have been treated as just another post production engineering processes to be applied to the 4 track mix down to the 2 track master cutting tape.

My understanding is that is what CBS did, monitor their mixes via a full logic decoder. Argent "In Deep" is an example of encoding for SQ full logic, the fronts and rear channels predominate alternately in pairs. I first noticed that when playing the Q8 in my car years ago. Listening to A&M mixes though I think that they mixed for discrete and simply ran that to an encoder.

 

[fredblue]

My understanding is that is what CBS did, monitor their mixes via a full logic decoder. Argent "In Deep" is an example of encoding for SQ full logic, the fronts and rear channels predominate alternately in pairs. I first noticed that when playing the Q8 in my car years ago. Listening to A&M mixes though I think that they mixed for discrete and simply ran that to an encoder.

i honestly don't think A&M's main Quad guys (Bornstein & Chiate) gave much of a toss about how their Quads were going to sound in decoded SQ form! it certainly sounds that way!

 

[doity]

I have often pondered why seemingly the technology was good enough to encode SQ properly before records being pressed, but consumers were stuck with decoders that left much to be desired. You would had thought that whoever owned the patents and technology behind SQ would had not let the software go to market before suitable decoders were available. Maybe it was just a money grab? I mean they did something similar with THX in that they had to certify that each receiver or whatever was compliant and that manufacturers weren’t just slapping the logo on there for a quick buck.

 

[fredblue]

I have often pondered why seemingly the technology was good enough to encode SQ properly before records being pressed, but consumers were stuck with decoders that left much to be desired. You would had thought that whoever owned the patents and technology behind SQ would had not let the software go to market before suitable decoders were available. Maybe it was just a money grab? I mean they did something similar with THX in that they had to certify that each receiver or whatever was compliant and that manufacturers weren’t just slapping the logo on there for a quick buck.

its a really good question!

i don't know but i think maybe SQ essentially had such limited front to back separation it needed all sorts of help beyond what was available in the early days of Quad's rollout?

perhaps the system was inherently limited so encoding took less "processing" than decoding with the necessary logic, etc required to try and improve upon its basic separation?

 

[fizzywiggs41]

SQ encode monitoring was probably lacking for Vanguard's titles on the handful I've heard.
At least to my ears.

 

[riknik]

Every label (including CBS) could have learned from Enoch Light and Project3 records on how to encode Quad.

 

[DuncanS]

its a really good question!

i don't know but i think maybe SQ essentially had such limited front to back separation it needed all sorts of help beyond what was available in the early days of Quad's rollout?

perhaps the system was inherently limited so encoding took less "processing" than decoding with the necessary logic, etc required to try and improve upon its basic separation?

That's what I would think. Its easy to encode (signals are phase shifted and combined, its all analogue), but once you've got them encoded, so phase shifted into the matrix it relies on (or used to) all analogue circuitry to extract the channels (adding & subtracting phase shifted combinations) and SQ is also not good at getting separation with out 'tricks'. Its why the Involve SM2 impresses me.

 

[MidiMagic]

What you get depends on how much effort they put into the product.

CASE 1: They mixed the discrete 4-track and then just encoded it.

This is probably the worst case. They made the discrete tape, and then just expected the SQ mix to be nearly the same. But artifacts of the 4-corners encoder can shift images from where they were intended to be. The images placed in the front should be about the same. But images placed to the sides and back can shift around.

CASE 2: They made separate discrete and SQ mixes.

Here they probably used the SQ forward-oriented and backward-oriented encoders with a multi-bus mixer. The SQ would be monitored through an SQ decoder.

Here the choice of decoder can affect the mix:

- The originally described decoder in the literature would have been useless.
- The front-back logic decoder was tricky, often taking too much time to ride gain.
- The 10-40 blend decoder is stable, so always affects placements the same way.
- The wavematching variblend decoder was better, but was not available until 1973.

- The best kind of decoder available in the 1970s was the Sansui phase matrix through the variomatrix.

All of the original SQ gain riding systems suppressed concert-hall ambience in the effort to increase separation.

I would use the 10-40 decoder to make the mix because the sound would always be heard in the correct direction during the mixing process.

I would have several different decoders connected (if available) and play the mix tape back into each one to listen for anomalies.

 

[kfbkfb]

- The originally described decoder in the literature would have been useless.

Do you mean the gain riding only CBS professional SQ logic decoder (described in the aforementioned AES paper)?

IIRC, CBS submitted a gain riding only SQ logic decoder to the BBC for their 1974 tests of matrix systems, I wonder why they didn't offer a Sony consumer version full logic w/wavematching and variblend SQ decoder?

Is there much evidence what Quad mixers making mixes for SQ encoding used for monitoring the SQ decoded mix (if they did monitor it at all)?


Kirk Bayne

 

[kfbkfb]

- The best kind of decoder available in the 1970s was the Sansui phase matrix through the variomatrix.

I had wondered about that since picking up a Sansui Quad receiver brochure back in the mid-1970s, it appears that the Sansui phase matrix/SQ decoder was the only variable matrix (no gain riding) logic directed SQ decoder available then.

The attached document refers to an "SQ conversion circuit" prior to variomatrix decoding, what, exactly, did that do and are there any channel separation measurements of the Sansui phase matrix/SQ decoder?


Kirk Bayne

 

[MidiMagic]

Do you mean the gain riding only CBS professional SQ logic decoder (described in the aforementioned AES paper)?

IIRC, CBS submitted a gain riding only SQ logic decoder to the BBC for their 1974 tests of matrix systems, I wonder why they didn't offer a Sony consumer version full logic w/wavematching and variblend SQ decoder?

Is there much evidence what Quad mixers making mixes for SQ encoding used for monitoring the SQ decoded mix (if they did monitor it at all)?

Kirk Bayne

I mean the decoder originally mentioned in magazine articles with no blend and no gain riding. It was the converse of the 4-corners encoder.

The attached document refers to an "SQ conversion circuit" prior to variomatrix decoding, what, exactly, did that do and are there any channel separation measurements of the Sansui phase matrix/SQ decoder?

Kirk Bayne

I don't see any SQ conversion circuit mention in the article. But it would have only one direction of varying instead of two for each channel.

 

[kfbkfb]

I don't see any SQ conversion circuit mention in the article. But it would have only one direction of varying instead of two for each channel.

It's the pale blue box on the Type-A QS vario-matrix page of the attached pdf in my previous post, I don't know what the equations are that preprocess SQ for the QS vario-matrix decoder.


Kirk Bayne

 

[par4ken]

I mean the decoder originally mentioned in magazine articles with no blend and no gain riding. It was the converse of the 4-corners encoder.



I don't see any SQ conversion circuit mention in the article. But it would have only one direction of varying instead of two for each channel.

I'm sure the earliest mixes were done with a basic decoder for monitoring as the logic decoders hadn't been developed yet. Somewhere (a bit later on) I read in a magazine article that CBS monitord all their mixes through a full logic decoder.

The original Audionics decoder used no logic but had an accurate 6-pole all-pass network, and it sounded very good. I never used blend with it either, that just compromised SQ's greatest strength. I would assume that the CBS monitor decoders would of had accurate all-pass/phase networks as well. Consumer decoders were sadly not as good.

Conversion for a QS decoder to decode SQ would require the addition of all-pass networks so the the left and right inputs could be mixed in quadrature (90°).

As implemented by Sansui the SQ decoding was referred to as half-logic.

 

[MidiMagic]

OK, I see it now. For some reason, the last article in that attachment didn't load yesterday. It ended with the ad.

The QS variomatrix varied the matrix for each channel in two ways. It could move the channel toward the front or back, or it could move it toward the left or right.

The SQ variomatrix could only vary the matrix in one direction, moving the channel toward the left or the right.

The converter changed the control circuits to recognize the SQ pattern of varying.

The all-pass networks were already there for QS. The psi+0 was used for the front,. and the psi+90 for the back. They were just recombined for SQ.