Block diagrams of Quad (QS/SQ) systems with logic

[juanverdaguer]

I'll be compiling a first beta test version soon that will do a basic decode (no logic yet) of both QS and SQ. For now this will be available for the Mac (Windows will come later on). Any interested testers around?

 

[sukothai]

I'll be compiling a first beta test version soon that will do a basic decode (no logic yet) of both QS and SQ. For now this will be available for the Mac (Windows will come later on). Any interested testers around?

I'd be happy to give it a go. I can compare QS and SQ to a QSD-1 and Tate II respectively.

- Ben

 

[oxforddickie]

I'll be compiling a first beta test version soon that will do a basic decode (no logic yet) of both QS and SQ. For now this will be available for the Mac (Windows will come later on). Any interested testers around?

Once you've done the compiled version, would it be possible to post a decode of the test tones?

OD

 

[juanverdaguer]

Once you've done the compiled version, would it be possible to post a decode of the test tones?

OD

Definitely, I'll let you all know when it's ready.

 

[quattro64]

I'm running snowleopard, Would love to test this when its done.

 

[national-kid]

I've just been looking at the circuit diagram of his QS/SQ decoder, and it's seriously flawed, basically it fails to decode either correctly.


OD

I have you scrutinize and appreciate.
Please teach me the place of a mistake .
I correct immediately.

Now, again, I make a SQ decoder (ultimate?) in the same idea.

 

[ancientquad]

Did you ever get a satisfactory answer about the MIX rectangles in the block diagram? These rectangles refer to variable amounts of mixing of the signal in the two portions of the circuit they connect - the essence of the variable matrix concept.

The rectangles show (0~0.5) amounts of mixing, meaning that the amount of mixing varies from zero (no mixing) to 50%. The net result of this mixing is to cancel out portions of the mixed signal that were 180 degrees out of phase.

Have you seen the U.S. patent on the original QS Vario-Matrix? It contains some very good block diagrams on the QS Vario-Matrix Type A and Type B decoders, and even includes some schematics of the actual decoder circuitry, with descriptions of the logarithmic amplifiers and detector circuits (the LOG and DET rectangles in the block diagrams). I have a copy of the patent lying around some where, so I can get the patent number and you can download it on line.

I have actually copied and built the circuit shown in the patent, and it works quite well. I don't recommend that anyone try to copy this circuit unless you have time on your hands - it took me over nine months to build it, so a software implementation might be easier.

I also have a copy of the SQ "Shadow Vector" patent which contains a block diagram of that system, and you can download that also. SQ is a bit more difficult, however, as it requires accurate phase shifting for portions of the signal within the matrix itself, whereas QS only requires phase shifting for the final four outputs of the decoding system. You can actually leave out these final phase shifters for QS.

I have a question which may not quite fit this forum. I am an experienced C++ programmer, and I've done some work with Java. How do you implement phase shifting with software?

 

[oxforddickie]

SQ is a bit more difficult, however, as it requires accurate phase shifting for portions of the signal within the matrix itself, whereas QS only requires phase shifting for the final four outputs of the decoding system. You can actually leave out these final phase shifters for QS.

Actualy, your wrong. The phase shifter networks are required. They wouldn't be there otherwise


OD

 

[ancientquad]

Phase shifting is not needed for QS Vario-Matrix decoding. It is needed to bring all four output channels to the same phase relationship, something that may not even be detectable by the human ear. For a listener at a defined point between a quad loudspeaker setup, moving one or more of the speakers will change the relative phase relationship between the speakers at certain frequencies

In the 1970's, a company called Photolume made a QS Vario-Matrix kit that that used the Sansui Vario-Matrix IC's. The circuit in this kit was identical to the circuit used in the Sansui QSD-2 commercial decoder, with the exception of the phase shifters at the output. I have two of these decoders also, and they work very well.

The phase shifters at the output are there to correct the phase shifts that were added in encoding. Sansui added these phase shifts in an attempt to enhance the apparent separation in QS recordings when they were played on regular two channel stereo, but phase shifting was not needed to make the decoder work. The Photolume decoder is proof of that it works quite well without them...

Remember, we're talking about proper decoder action here -whether the decoder can produce 4 apparently distinct output channels. Relative phase relationships between outputs is a matter of exacting taste and may not be audible.

 

[juanverdaguer]

Hi there,


I've been analyzing the diagrams posted by Odaka and they were quite helpful at least in understanding the concept of the logic circuitry. My first version of the software will not include any logic though but I'm working on it in future versions. Having a look at the original patent of the QS Vario-Matrix will be extremely helpful! I've found papers about it but never the actual thing (the block diagrams for SQ Shadow Vector will be very welcome as well).

I'm programming this in Max 6 (aka MaxMSP), it's a visual programing environment developed by www.cycling74.com - I wouldn't know how to do phase shifting in Java or C++ but if you're experienced with textual coding you might want to try supercollider at http://supercollider.sourceforge.net/
It's arguably the best way to do real-time audio programming with a C-like flavor.

I'll be distributing it through the Apple App Store and later from a webstore (also compiling it for Windows). I hope I'll have a reviewed copy of the app soon and then I'll reserve a number of promotional codes for testing purposes. If you're interested please let me know so I can add you to the list.


J

 

[oxforddickie]

I know QS does not need the phase shifting for decoding, but to ensure the correct phase response, and to ensure that image creation (as good as it can be with a matrix system) is correct the phase shift networks are needed. Without them, the imaging is defused, with poorer location, stability, and (to a lesser extent) bass response when such instruments are placed away 'from the wall'


OD

 

[quadsearcher]

With 180 degrees out of phase being extremely noticeable, I would expect any lesser phase difference to be proportionately obnoxious.

 

[national-kid]

Did you ever get a satisfactory answer about the MIX rectangles in the block diagram? These rectangles refer to variable amounts of mixing of the signal in the two portions of the circuit they connect - the essence of the variable matrix concept.

The rectangles show (0~0.5) amounts of mixing, meaning that the amount of mixing varies from zero (no mixing) to 50%. The net result of this mixing is to cancel out portions of the mixed signal that were 180 degrees out of phase.

Have you seen the U.S. patent on the original QS Vario-Matrix? It contains some very good block diagrams on the QS Vario-Matrix Type A and Type B decoders, and even includes some schematics of the actual decoder circuitry, with descriptions of the logarithmic amplifiers and detector circuits (the LOG and DET rectangles in the block diagrams). I have a copy of the patent lying around some where, so I can get the patent number and you can download it on line.

I have actually copied and built the circuit shown in the patent, and it works quite well. I don't recommend that anyone try to copy this circuit unless you have time on your hands - it took me over nine months to build it, so a software implementation might be easier.

I also have a copy of the SQ "Shadow Vector" patent which contains a block diagram of that system, and you can download that also. SQ is a bit more difficult, however, as it requires accurate phase shifting for portions of the signal within the matrix itself, whereas QS only requires phase shifting for the final four outputs of the decoding system. You can actually leave out these final phase shifters for QS.

I have a question which may not quite fit this forum. I am an experienced C++ programmer, and I've done some work with Java. How do you implement phase shifting with software?

Thank you for variety. About the patent, did not know at all.
Reference in the case of digital(IIR). (Sampling is 32kHz)
http://www.hi-ho.ne.jp/odaka/quad/phase.pdf

 

[national-kid]

SQ Logic decoder ( ultimate type ) of 3 band has been completed.
Compared to the normal type of one band, the change was not dramatic.

 

[flavio81]

SQ Logic decoder ( ultimate type ) of 3 band has been completed.
Compared to the normal type of one band, the change was not dramatic.

ODAKA,

You RULE!!

... which is a way to say "you are great!"

Congratulations!!

 

[quadsearcher]

That does look good. Does it need a lengthy alignment? A lot of potentiometers on there!

 

[national-kid]

flavio81 san
>... which is a way to say "you are great!"
-> You are appreciated, the amateurish work of me and I will very pleased.

quadsearcher san
>That does look good. Does it need a lengthy alignment? A lot of potentiometers on there!
->There are 113 individual semi-fixed resistor on the schematic.
Semi-fixed resistor is required at the following.
page 1/3: (Such as VR27) 16 of phase-shift (Because accuracy)
page 2/3: (VR82) for the noise level (system environment)
(Such as VR48) 9 for the balance noise (variation in element)
(Such as VR77) 27 for the amplifier offset LOG (variation in element)
(Such as VR81) 9 for the amplifier gain LOG (variation in element)
(Such as VR50) 3 for the display sensitivity (At adjustment, sensitivity UP)
page 3/3: (VR1, VR2) 2 for the offset (variation in element)
(Such as VR25, VR26) 12 for the blend (variation in element)
total of 79.

Seems to be a good place in the fixed resistor is below,
page 2/3: (VR84) Offset resistor is 200Ω.
(Such as VR47) 9 resistor for blending is 10KΩ.
(Such as VR49) 4 resistor for adjusting the high frequency gain is 7.5kΩ, 43kΩ.
(Such as VR62) 4 resistor for adjusting the middle frequency gain is 16kΩ, 33kΩ.
(Such as VR72) 4 resistor for adjusting the low frequency gain is 13kΩ, 36kΩ.
page 3/3: (Such as VR13) 12 resistor for controlling the phase shifter is 20KΩ .

Also a non-linear circuit, the best point of logic operation and the best sound quality (the best point of the noise, the best point of distortion), so do not match, fine adjustment is absolutely necessary.

http://www.hi-ho.ne.jp/odaka/quad/3-band-figure2-e.pdf
http://www.hi-ho.ne.jp/odaka/quad/SQ3B.pdf

 

[national-kid]

In addition, Ｉ made a 4ch decoder.
This time, by creating a printed circuit board, it was produced.

http://www.hi-ho.ne.jp/odaka/quad/SQ1B.pdf

 

[Jim the Oldbie]

Wow! Very nice! Congratulations on a fine project.

-- Jim

 

[kronos77]

nice layout on pcb can you provide the pcb file sir