Reality Teck Involve Decoder - proper QS decoding? - test results

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Hey there OxfordDickie and everyone else. Some passionate chatting in this thread. Possibly the defamatory title is part of the reason. :)

The first thing we did was listen to the files you've submitted. I'm assuming you're using the same source as the DVD-Audio torrent file of the Involve Decode of that album that has appeared around the various music-sharing sites, given that they are multi-channel MLP files. I don't know who put it out there but I'm all for extra publicity at the moment :)

As it stands, I'm not sure why you think the modified file is more distinct or accurate. We are using electrostatic speakers to do our listening tests - as you know they are much more unforgiving, and therefore suited to this type of listening test. The original file from our decode is (to our ears) much sharper and clearer at the rear. After a bit of checking, I can see that you've inverted the phase on the rear right channel. I can't understand why you would want to do this. One of our older pre-production prototype versions did have an issue with the rear channels being out of phase, however we fixed that problem quite early on, and well before the final version went into production.
The only other difference I can see visually is that your file has an introduced DC offset on the rear right channel as well as the phase change. We don't have a DC offset on our outputs.

The front channels in your modified file seem unchanged

As it stands, we're still taking this issue seriously. This Friday we will have access to the original QS vinyl recording of "Photographs and Memories" by Jim Croce (which is where this decode came from) and we will also have access to a Sansui QSD-1. We (Well, that's the royal "we" - knowing Charlie it'll be me) will record all the channels, analyse the separate channels and side by side for any destructive differences (such as phase or directionality) between the QSD-1 and the SM, and also do a vectorscope comparison, which is the easiest way to visually illustrate phase coherency.

In the meantime, here is a comparison of Surround Master vs QSD-1 on the Vector 4 Quad demonstration LP that we did last week. If we had a rear phase issue, rear signals would appear to go to rear left and right, instead of straight back, as both decoders do properly.

http://www.youtube.com/watch?v=n5axJdfVrCY
(It's not in HD yet, give it another few hours to finish processing, but the 360p res is still more than adequate really)

Cheers
David
Involve Audio
 
Last edited:
Folks, I cleaned up this thread and changed the title to be more fair to Involve audio. Please continue the discussion and stay friendly and cordial. Disagree all you like, but do NOT get personal and call out individual members knowledge or validity, regardless of their position.

Thanks
 
http://www.youtube.com/watch?v=n5axJdfVrCY[/URL]
(It's not in HD yet, give it another few hours to finish processing, but the 360p res is still more than adequate really)

Cheers
David
Involve Audio

Firstly, thanks.

Yes, exactly the same file, but you are incorrect in regard to the phase reversal of the rear right channel.


For now, i'll leave it at that
 
Firstly, thanks.

Yes, exactly the same file, but you are incorrect in regard to the phase reversal of the rear right channel.


For now, i'll leave it at that

Fair enough. Well, regardless of the intention of your modification / filter, it has definitely resulted in a near 180-degree phase shift in the RR channel, easily heard and quite visible on the wave form.

~David
 
Actually, you are incorrect, and i feel this now needs to be taken off this list as it's going to get very technical.

I will send Chucky my direct email address so this can be continued further
 
Actually, you are incorrect, and i feel this now needs to be taken off this list as it's going to get very technical.

I will send Chucky my direct email address so this can be continued further

Why can't technical be discussed here?

At any rate, feel free to include me in the email correspondence. I did after all have to study the theory behind it and implement the algorithms, and will be incorporating any changes, so if you have issues with the technical design I'm going to need to be included regardless.

Cheers
~David.
 
This is not the place for such discussions.

In no way am i questioning the basic algorithms you may have used, although hthere may be other tweeks needed to ensure a more smoother decode, but that is way beyond where we are at now.
 
Hello All

Finally we have finished a fresh round of tests including comparisons to the Sansui QSD -1 with an aim of examining whether we decode QS "by the book" or not.

I wish to thank Richard "Oxforddickie" for pressurizing us into doing this and for all his good technical advice and assistance over the past week or two. In addition many thanks to Ron "Rustyandi" for the loan of his spare QSD -1 for test comparisons.

Please note that the right hand side output of the QSD - 1 was not performing to specification (analogue logic was frozen) but we were able to do some useful tests on the left hand outputs that were performing at or above quoted specification. I claim this is like doing tests on a stereo amplifier with one side faulty, the results on the good channel are still valid. And beggars cannot be choosers.

For those that do not want to be bored by the lengthy report my summary is:

CHANNEL SEPARATION: QSD -1 achieved an average of 23.8 dB separation, Surround Master achieved an average separation of 36.3 dB. Please note both obtained great results but frankly anything more that 12 dB is just a numbers chase.

PHASE ACCURACY: Based on the only valid comparison of the side wall result the best I can say is that both machines are probably similar, QSD -1 had a phase error of 33.8 degrees on one side wall, the Surround Master had figures of 21.8/ 70.9 degrees. All front and rear wall phase measurements for the Surround Master were fully in phase with zero measurable error.

IMAGE LINEARITY: Please refer to the graphs in the report. I is fair to say both machines performed well but we could only test the left hand side wall image linearity. If anything the Surround Master had the edge slightly.

In summary the INVOLVE Surround Master is in my opinion decoding "by the book" as written by the great Tachashi of Sansui in the early 1970". We have however written a new chapter that is not tested in these results in how we hide the control action from detection by the listener. On listening tests I heard occasional evidence of pumping on the QSD -1 but not on the Surround Master. Having said that the QSD - 1 is an stunning device and far better than SQ, Tate, Dolby pl2, neural etc.

We will be conducting a few more tests on decoding of simultaneous 4 channel tone and dynamic response over the next week or two - time permitting.

Regards

chucky
 

Attachments

  • QS tests Feb 2013 R3.pdf
    368.6 KB · Views: 202
Last edited:
Hi again

As promised we have documented some more comparative test on the Surround Master and QSD-1. This test is perhaps the most difficult one for any matrix based decoder system - Continuous sine wave tones into all 4 encode / decode channels simultaneously. In essence the "analogue logic" control is given no basis for separating the signals.

The biggest "trick" of analogue steering logic is to pick directional dominance on a transitory basis and adjusting the matrix parameters to suit (or gate if you are SQ, Tate or Dolby). This is very valid as human directional perception takes a significant time to pick the direction of multiple separate continuous signal sources. In music there is little need to be able to separate multiple time synchronized tones as the listener cannot separate those events in the short term. Most analogue steering logic cannot cope with this event and at best defaults to the basic non logic separation of a theoretical 6 dB in the case of QS.

In summary the Surround Master performed well again with an overall average separation of 10 dB. The Sansui QSD-1 suffered slightly with an average separation of 3.7 dB, this result was worse than a non logic decoder mainly because it enhanced the wrong channel in one instance (sent the signal to the wrong channel).

Please note the Surround Masters result is close to the 12 dB threshold where we claim listeners cannot distinguish from discrete (you really do not need 100dB!).

I have not seen this type of test published previously - maybe some one could prove me wrong! Anyway it is only useful in showing the limits of performance and perhaps is some evidence that we really have a different logic steering enhancement compared to the QSD -1.

Happy reading

Chucky
 

Attachments

  • Surround Master Involve Proper.pdf
    612 KB · Views: 165
Back
Top