Isn't the Tate II 30db of separation? It sounds like you may be able to blow that away. I'm excited. I'm glad it's down to some minor corrections and refinements. Thanks for the updates! I'm guessing you guys will have more scope comparison once there is a demo quality unit.
INVOLVE SQ - IS COMING
- Thread starter chucky3042
- Start date
Help Support QuadraphonicQuad:
- Joined
- Apr 9, 2012
- Messages
- 2,707
Hi Q8
Your observation as it turns out is central to our final checks at the moment.
The latest data (at rustyandi's test room ) we have is the apparent error in the placement of the center left signal (going to left rear) is the result of the test disc we were using - the left / right relative magnitudes were not correct in the disc for that left center placement (disc was out by 2 db). This error under SQ is sufficient to swing the placement to left rear. In addition some of the other signal level inconsistencies we observed at "Rustyandi's" setup are not present in our test setup - we suspect and will check with Ron if there are slight level shifts in his system (even say 0.5 db will cause errors).
With SQ you basically have +/- 3 db of depth information to encode say 20 - 40 db of separation so a tiny error in level of say 1 db can result in 10 db of level shift in the decode. With QS we have 7.7 dB of encode "width" to decode from - much more tolerant.
What separation do we really want, we can give you 60 dB but the system will be highly unstable with pumping and level shifts. The trial unit at Rustyandi's test room was set for 20 dB separation - we did not hear any surging or pumping but we think the slight level shifts were due to minor level issues in the system (vinyl disc's were used). I know its a bit controversial but our tests have shown that a trained listener cannot hear the difference between 12 dB and 100 dB separation! The CD age has caused a bit of a pointless pissing contest on this parameter. I remember some very nice phono cartridges only testing to 15 dB separation.
As David (we refer to him as "Bitch") stated we are not entering a numbers chase on separation, instead we are attempting to get the most convincing surround sound. We probably will release the unit with 30 dB separation if we think it is stable for most systems (personally I would set mine for 12 dB ).
Please note that we have been successful in implementing a 3 band SQ decoder, I am not sure but I think this will be the first SQ decoder with this. To our knowledge the Tate unit was a single band system - Please someone correct me if this is a wrong statement. I believe the benefits of the 3 band system combined with the psychoacoustic techniques used in INVOLVE will far out weigh the much overrated parameter of separation.
We will do full "scope" and other test documentation but we first need to make our own SQ encoder as the test discs we have used are inaccurate (have not used Oxfordickies yet- I believe it is accurate) in regards to magnitude and phase (80 degree shifts not 90). But as stated before we will not release an inferior sounding unit.
Many thanks to Ron (rustyandi), his incredible equipment setup, music collection, knowledge and hospitality has been of huge assistance to us. His home is just 40 minutes away by car - its like an extension to our own laboratory!
Regards
Chucky
Your observation as it turns out is central to our final checks at the moment.
The latest data (at rustyandi's test room ) we have is the apparent error in the placement of the center left signal (going to left rear) is the result of the test disc we were using - the left / right relative magnitudes were not correct in the disc for that left center placement (disc was out by 2 db). This error under SQ is sufficient to swing the placement to left rear. In addition some of the other signal level inconsistencies we observed at "Rustyandi's" setup are not present in our test setup - we suspect and will check with Ron if there are slight level shifts in his system (even say 0.5 db will cause errors).
With SQ you basically have +/- 3 db of depth information to encode say 20 - 40 db of separation so a tiny error in level of say 1 db can result in 10 db of level shift in the decode. With QS we have 7.7 dB of encode "width" to decode from - much more tolerant.
What separation do we really want, we can give you 60 dB but the system will be highly unstable with pumping and level shifts. The trial unit at Rustyandi's test room was set for 20 dB separation - we did not hear any surging or pumping but we think the slight level shifts were due to minor level issues in the system (vinyl disc's were used). I know its a bit controversial but our tests have shown that a trained listener cannot hear the difference between 12 dB and 100 dB separation! The CD age has caused a bit of a pointless pissing contest on this parameter. I remember some very nice phono cartridges only testing to 15 dB separation.
As David (we refer to him as "Bitch") stated we are not entering a numbers chase on separation, instead we are attempting to get the most convincing surround sound. We probably will release the unit with 30 dB separation if we think it is stable for most systems (personally I would set mine for 12 dB ).
Please note that we have been successful in implementing a 3 band SQ decoder, I am not sure but I think this will be the first SQ decoder with this. To our knowledge the Tate unit was a single band system - Please someone correct me if this is a wrong statement. I believe the benefits of the 3 band system combined with the psychoacoustic techniques used in INVOLVE will far out weigh the much overrated parameter of separation.
We will do full "scope" and other test documentation but we first need to make our own SQ encoder as the test discs we have used are inaccurate (have not used Oxfordickies yet- I believe it is accurate) in regards to magnitude and phase (80 degree shifts not 90). But as stated before we will not release an inferior sounding unit.
Many thanks to Ron (rustyandi), his incredible equipment setup, music collection, knowledge and hospitality has been of huge assistance to us. His home is just 40 minutes away by car - its like an extension to our own laboratory!
Regards
Chucky
Neil Palfreyman
1K Club - QQ Shooting Star
- Joined
- Apr 9, 2012
- Messages
- 2,707
Keep up that attitude and your price on the SQ unit will be +50%.
Enjoy the Rugby yesterday?
Neil Palfreyman
1K Club - QQ Shooting Star
No attitude at all; just pointing out that your arses were handed to you on a platter.
LizardKing
2K Club - QQ Super Nova
Enjoy the Rugby yesterday?
Yeah - Great win for the welsh - coached by a Kiwi - to beat the Aussies (coached by a Kiwi)....
- Joined
- Apr 9, 2012
- Messages
- 2,707
Stop picking on us poor little Aussies.
chucky3042 said:
I agree that proper channel placement is more important than extreme seperation. As long as it sounds discrete. Proper channel placement shouldn't suffer because of a numbers war. Also, keep in mind that I'm sure a lot of other people will be like me, and will use the SQ mode primarily with original SQ vinyl records. I will probably be using a linear turntable with a p-mount cart and a shibata stylus (the CD-4 setup), so I'm guessing that will give me very consistant, accurate and balanced L/R info to decode. It sounds like the slightest tracking error creates decode errors. I guess the matrix systems aren't as forgiving as they are cracked up to be once you get to real seperation levels.
- Joined
- Apr 9, 2012
- Messages
- 2,707
Hi Q8
I will keep your comments in mind. The old QS is much more forgiving than SQ. The SQ placed far too much emphasis on the importance of infinite left / right separation and so very little "weight " was placed on rear information. QS treated all major directions evenly giving more dB space to encode data on. The SQ format was very tolerant.
Regards
Chucky
I will keep your comments in mind. The old QS is much more forgiving than SQ. The SQ placed far too much emphasis on the importance of infinite left / right separation and so very little "weight " was placed on rear information. QS treated all major directions evenly giving more dB space to encode data on. The SQ format was very tolerant.
Regards
Chucky
That makes sense. By focusing on Left/Right separation, you trade off the rear channel and directional cues.
It helps explain why some used pre-processing before SQ decoding like the "axial tilt" feature in the Audionics Space & Image Composer Tate SQ decoder and more recently the pre-processing mentioned by OD in his SQ Final decoding projects.
oxforddickie
1K Club - QQ Shooting Star
The pre-processing i use is to help errors in phase. Not quite sure i agree that by concertrating on L/R seperation causes a trade off in the rear cues. A ferfect example of what SQ is capable of, when decoded correctly, is 'Rocket Man' on John Keatings -Space Experience, although to be honest the whole album is THE perfect test album when dealing with SQ decoding.
The placement and movement of instruments allows the perfect testing ground for any decoding process, i hope the 'topsy-turvy' folks are using it as part of their tests
Soundfield
1K Club - QQ Shooting Star
Hi Chucky - thanks for all the exciting news, this will surely turn out to be a landmark product in many ways and one which finally puts the venerable (but less than perfect) Tate out to long overdue retirement. I know of no other 3 band SQ decoder, and obviously not one with the state of the art psychoacoustic techniques that the INVOLVE engine brings. I admire your dedication to getting the 'sound' of this thing right (sadly something that few people bothered much with in the 70's!) and not just releasing any old "that'll do" product. I look forward to seeing the test results soon. Cheers.
oxforddickie
1K Club - QQ Shooting Star
I agree with 'Chucky' in that all of the CBS SQ test discs are inaccurate. Not good when trying to test a piece of equipment, but CBS isn't alone in this respect. All of the test tones on the Quadraphile double LP (SQ, QS, BMX/UD-4) are inaccurate too. Makes you wonder what on earth went on.
chucky3042 said:
I have a couple more questions. Firstly, do you have a link to the published results of your tests that show 12db of seperation is enough for most people to consider discrete. That number fascinates me as other well known 70s decoders have advertised 15-25 and have been pretty dissapointing in the seperation area. Maybe this is just due to the fact that they didn't decode SQ properly in the first place, or the slow slow steering that made everything a confusing blur of directional cues.
Now for the next question. What is the benefit of 3-band decoding/what is 3-band decoding?
- Joined
- Apr 9, 2012
- Messages
- 2,707
Hi Q8
Really good questions.
In regard to our 12 db separation rule I do not have any external published work. It really came down to a question we at involve asked a year or two ago. In the 70's the holy grail was 20dB and in the digital era anything less than 100 dB is clearly inadequate. So we set up a test arrangement where we could control the actual separation and got a bunch of golden eared types and simply trialed them on an instant A/ B switch asking the question could they identify the the CD 100 dB source. The results got random at 12 dB. Really interesting stuff.
If you check out the Haas precedence curve you will also note that 12 db is the "magic" number on that curve. I suspect its all tied in with evolved (or intelligent designed- for our christian brethren) human perception.
Our preference for multi band decoding comes from the typical sound envelopes at different sections of the frequency spectrum. At lower frequencies eg bass guitars you have moderate attack rates and slow decays, at treble frequencies sound is characterized by very fast attacks but decay times below say 20 ms. A multi band decoder can adjust the front/back/ left/ right steering constants to match these packets of information - thereby reducing pumping. It should be noted that if a treble time constant is used on bass frequencies the result will be low frequency distortion. If a low frequency time constant is used for treble sounds the result will be blurred imaging and pumping. Please note our attack times are substantially faster than our decay times.
In addition the use of multiband decoding means it is easier to assign "weight" to different sections of the frequency spectrum as the human ear will perceive mid frequencies in preference to bass and treble frequencies. With single band decoders you usually need to ignore steering below 800Hz as you will get a poor transient response.
Hope that helps.
Chucky
Really good questions.
In regard to our 12 db separation rule I do not have any external published work. It really came down to a question we at involve asked a year or two ago. In the 70's the holy grail was 20dB and in the digital era anything less than 100 dB is clearly inadequate. So we set up a test arrangement where we could control the actual separation and got a bunch of golden eared types and simply trialed them on an instant A/ B switch asking the question could they identify the the CD 100 dB source. The results got random at 12 dB. Really interesting stuff.
If you check out the Haas precedence curve you will also note that 12 db is the "magic" number on that curve. I suspect its all tied in with evolved (or intelligent designed- for our christian brethren) human perception.
Our preference for multi band decoding comes from the typical sound envelopes at different sections of the frequency spectrum. At lower frequencies eg bass guitars you have moderate attack rates and slow decays, at treble frequencies sound is characterized by very fast attacks but decay times below say 20 ms. A multi band decoder can adjust the front/back/ left/ right steering constants to match these packets of information - thereby reducing pumping. It should be noted that if a treble time constant is used on bass frequencies the result will be low frequency distortion. If a low frequency time constant is used for treble sounds the result will be blurred imaging and pumping. Please note our attack times are substantially faster than our decay times.
In addition the use of multiband decoding means it is easier to assign "weight" to different sections of the frequency spectrum as the human ear will perceive mid frequencies in preference to bass and treble frequencies. With single band decoders you usually need to ignore steering below 800Hz as you will get a poor transient response.
Hope that helps.
Chucky
- Joined
- Apr 9, 2012
- Messages
- 2,707
Me again
Ooops, forgot to mention it also enables the decoder to separate SIMULTANEOUS tones at different parts of the frequency spectrum. So in theory it is possible to have simultaneous instruments in each corner of the room - providing they are frequency separated. See the qq forum - Reality Teck Involve Decoder - proper QS decoding? - test results. A single band SQ encoder can only separate 3 dB front / back on SIMULTANEOUS tones a multi band will do considerable better.
Regards
Chucky
Ooops, forgot to mention it also enables the decoder to separate SIMULTANEOUS tones at different parts of the frequency spectrum. So in theory it is possible to have simultaneous instruments in each corner of the room - providing they are frequency separated. See the qq forum - Reality Teck Involve Decoder - proper QS decoding? - test results. A single band SQ encoder can only separate 3 dB front / back on SIMULTANEOUS tones a multi band will do considerable better.
Regards
Chucky
HI. All
It seems the NEW UNIT, is just around the corner. I`m more than excited I`m over the moon. :woopie But at the end of the day it seems that garbage in garbage out, revering to Chucky and Oxforddickie Conclusion that the CBS test records aren’t accurate. So this might start someone to put up a list of good to exhalent lp`s to go for.
It seems the NEW UNIT, is just around the corner. I`m more than excited I`m over the moon. :woopie But at the end of the day it seems that garbage in garbage out, revering to Chucky and Oxforddickie Conclusion that the CBS test records aren’t accurate. So this might start someone to put up a list of good to exhalent lp`s to go for.
Similar threads
