- Joined
- Apr 9, 2012
- Messages
- 2,707
Hi Owen
The "original" QS and SQ decoding algorithms are just the raw decode matrix equations that result in:
For QS 3 dB separation between front / back and left/ right and theoretical infinite separation across diagonals.
For SQ Theoretical infinite separation left/ right but only 3 dB front / back
Basically just as the original non logic decoders were. I think it is universally agreed that these decoders were very inadequate and just created a blurred image. Later - particularly with SQ "gain riding logic" decoders were added to the basic SQ decode matrix to enhance separation to say 20 db all round. The massive downside of these decoders were pumping and surging sounds with shifting images. Sansui were much more successful with there post matrix "vario matrix" processing. In the case of QS because it is a much more circular balanced matrix pumping was easier to reduce.
In the recent development of our version of SQ logic decoding we trialled early version based on the known works of several prior logic decoders - without our Involve processing and the result was really bad pump/ surging problems. The problems were so bad that David at one stage was at a loss of how to fix the problem. I then listened to the decoded output and simply told David to now add the Involve processing which he did resulting in an absence of surging and pumping. In all the intense work David had done he actually forgot about how well Involve processing hides the pumping!
Involve in no way effects the decode logic of either the SQ or QS matrix logic decode techniques but it does greatly effect the manner of how the "steering constants" are applied. Involve "listens" to the music in the same way your brain perceives the music and applies the logic steering factors in a manner that is undetectable to the listener. In addition Involve picks directional clues with the same "biases" that your brain has evolved (or intelligently designed for you religious nutters) to perceive.
I know most HiFi purist insist human hearing is super sensitive and can undoubtedly detect amplifier distortion to 0.00001% but in reality is actually easily fooled - it is these cracks that Involve utilises.
Yes it is possible to give you just the "pure" SQ/ QS decoder but the result will be vastly inferior (actually not listenable in my opinion). The intent of all logic decoders is to somehow enhance the basic decode matrix - any thing outside the basic (3 dB) decode matrix is impure and subject to the "tricks" of the decode logic. Its all dependent on how well it is applied.
Please review the "
Reality Teck Involve Decoder - proper QS decoding? - test results " thread and you will see we actually out perform the legendary Sansui QSD -1.
There is my bias showing again - forgive me its Sunday night.
Regards
Chucky
The "original" QS and SQ decoding algorithms are just the raw decode matrix equations that result in:
For QS 3 dB separation between front / back and left/ right and theoretical infinite separation across diagonals.
For SQ Theoretical infinite separation left/ right but only 3 dB front / back
Basically just as the original non logic decoders were. I think it is universally agreed that these decoders were very inadequate and just created a blurred image. Later - particularly with SQ "gain riding logic" decoders were added to the basic SQ decode matrix to enhance separation to say 20 db all round. The massive downside of these decoders were pumping and surging sounds with shifting images. Sansui were much more successful with there post matrix "vario matrix" processing. In the case of QS because it is a much more circular balanced matrix pumping was easier to reduce.
In the recent development of our version of SQ logic decoding we trialled early version based on the known works of several prior logic decoders - without our Involve processing and the result was really bad pump/ surging problems. The problems were so bad that David at one stage was at a loss of how to fix the problem. I then listened to the decoded output and simply told David to now add the Involve processing which he did resulting in an absence of surging and pumping. In all the intense work David had done he actually forgot about how well Involve processing hides the pumping!
Involve in no way effects the decode logic of either the SQ or QS matrix logic decode techniques but it does greatly effect the manner of how the "steering constants" are applied. Involve "listens" to the music in the same way your brain perceives the music and applies the logic steering factors in a manner that is undetectable to the listener. In addition Involve picks directional clues with the same "biases" that your brain has evolved (or intelligently designed for you religious nutters) to perceive.
I know most HiFi purist insist human hearing is super sensitive and can undoubtedly detect amplifier distortion to 0.00001% but in reality is actually easily fooled - it is these cracks that Involve utilises.
Yes it is possible to give you just the "pure" SQ/ QS decoder but the result will be vastly inferior (actually not listenable in my opinion). The intent of all logic decoders is to somehow enhance the basic decode matrix - any thing outside the basic (3 dB) decode matrix is impure and subject to the "tricks" of the decode logic. Its all dependent on how well it is applied.
Please review the "
Reality Teck Involve Decoder - proper QS decoding? - test results " thread and you will see we actually out perform the legendary Sansui QSD -1.
There is my bias showing again - forgive me its Sunday night.
Regards
Chucky
Reading through this, the SQ and QS decoding then has Involve(tm) DSP post processing applied. I'd rather have just the original quad decoding algorithm as intended applied and nothing else. Is it possible to turn off the Involve processing?