2020 SM v2 Testing

[Sonik Wiz]

Early on Jon did some great testing of the SM v2 mainly by wave form comparisons on his DAW. And in the early days the SM v1 operation was documented by quadscope demonstrations. Except for Involve I don't think anyone on the forum has run a numbers game on the SM v2, that is separation crosstalk.

I finally was able to pull my SM v2 out of my system for a bit (that was hard. I listen to it every day) to do some testing. I just wanted to satisfy my own curiosity & see if I could learn anything about how better set it up or optimize performance. As you see in the attached it was mainly about checking levels & separation for 8 different directions, QS, SQ, 4.0 & 5.1.

On my PC I made eight test signals corresponding to corner & center positions. I used a 1kHz sine wave that fitted in pretty nicely to the 300Hz>3kHz of the SM. I didn’t do any other frequencies as I had confidence what I tested here would be mirrored. The things that would make a difference would be attack/decay times & over all flatness through the audio spectrum because of the filters, & I couldn’t measure that.

My test signals were at 1V input to the SM. It makes it easier to figure voltage change to dB ratios. The input level clip light comes on about .548 V with the level set to max. Setting the input level to 12:00 with 1 V input put it just a bit below the max LED clip level. With the output level controls set straight up, it is unity gain. It is very close probably with in the tolerance of the individual pots. Now my Oppo has a max analog output level in the unbalanced mode of 2V. So if you plug a disc player like this into the SM you will definetly need to back off on the input level. For myself the Oppo plugs into a Chase RLC1 control unit before going to the SM so it's very easy to control the input level with out having to adjust the Oppo.

I started by sending corner signals to the unit & & adjusted output gain to read 1.0 V on my meter. After doing that I used a center front signal & in the 5.1 mode I adjusted center front out put to 1 V.

I went back to the 4.0 mode & repeated the QS corner signals & this time I measured the cross talk on the other chs. I repeated with center left, right, front, back. In the attachment I normalized everything relative to 1.0 V & just calculated the dB attenuation. For example a left front only (or right front only) might output 1.0V but measuring them with center front each ch drops down to .70 V. As it should be! I don’t think any vintage decoder could do this. I also noticed it on center left, right, & back.

https://www.dropbox.com/s/77psu5spg7mi7h2/INVOLVE QS.pdf?dl=0
I learned that the SM is very accurate & predictable & if I use test noise & an SPL meter to fine tune, I’m adjusting to account for speaker & room acoustics not the decoder.

About the only oddity I found in both the QS & SQ modes is that 5.1 center front does not have as hard a center front output as I expected. The front L/R outputs are only about 6dB down. I sent these test results to Chucky as I wanted validation that my testing method wasn't full of gaping holes. He rplied in part:

We spent a hell of a lot of time trying to get the best center channel blend into the frontal image. We really have a powerful center channel separator algorithm but as I have frequently said we are not in a numbers pissing contest......we really have adjusted things to where THEY SOUND BETTER! Early versions of the center separator we "let it fly" with 40 db separation, and it sounded crappy and mechanical. In addition the frontal image was not totally stable. We found after a hell of a lot of LISTENING that 6 db was around the right number for a smooth stable frontal image with the center channel bandaid.

So the measurement was correct. Interestingly when I was using DPL II a lot I came to the same conclusion. Adjusting the width between hard center front & phantom center I chose in between where the center front was dominate but front L/R down by about 6dB.

I did some SQ test tones & checked that out but not with as many positions. Mainly I was curious about all the stuff regarding center front to back leakage in SQ. The front to back separation in SQ is not as high as QS but at about -16dB that’s still plenty good. I guess the only other way to reduce the leakage is to blend the rear output a bit & that reduces the rear ch width. Who wants to do that?

Over all I am just so delighted that the SM decodes that precise even after over a year of frequent use. I’d expect all analog to need to trim pot adjusting by now. And really with a clean digital source instead of LP & phono cartridge, with the Surround Master decoder and the encoder at the front end, this is better than we old quaddies could have ever dreamed of from 2 chs.

 

[J. PUPSTER]

Nice SW!

I’m wondering what will go on with the upcoming v3, when we feed the calmed Blue Eyed Monster with enlightened guts, Hi-Res. files and output input with Toslink?
*although for most uses I’ll be staying analog.

 

[J. PUPSTER]

Nice SW!

I’m wondering what will go on with the upcoming v3, when we feed the calmed Blue Eyed Monster with enlightened guts, Hi-Res. files and output with Toslink?
*although for most uses I’ll be staying analog.

Oops, I guess that should be “input” Toslink?

 

[furui_suterioo]

Awesome test, much better than what I've been doing lol, but my measuring tool is only the VUs on my receiver while test tone is playing(8-pos 1khz). I just increase the volume until the opposite VU of the test tone begin to rise, the difference between the opposing VUs is my rough seperation measurement. Even though its rudimentary, does provide some results, when I get my QSD-2 back(covid delay) I will still post the photos(possibly video) of the VU meters between Involve, Audionics, Fosgate and the Sansui(and maybe the Kenwood too just for fun).

 

[Sonik Wiz]

Awesome test, much better than what I've been doing lol, but my measuring tool is only the VUs on my receiver while test tone is playing(8-pos 1khz). I just increase the volume until the opposite VU of the test tone begin to rise, the difference between the opposing VUs is my rough seperation measurement. Even though its rudimentary, does provide some results, when I get my QSD-2 back(covid delay) I will still post the photos(possibly video) of the VU meters between Involve, Audionics, Fosgate and the Sansui(and maybe the Kenwood too just for fun).

Thanks!
Ya I've been the same route squinting at the little meters on my 9940 trying to discern decoding seperation. I had much bigger meters om my old Sony 4ch R2R but still doesn't do much good for anything under -20dB. I did this the 'ol fashioned way; a DMM with shielded leads & checking one ch at a time. To do it that way for multiple decoders like you have would be cruel & inhuman punishment.

I expected the Involve/QS decoding to be quite good but I was surprised & very pleased to see how good SQ decoding was. I don't know how Involve implemented it but SQ decoding in a Sansui QS Variomatrix environment is pretty wonky. The Sansui method used the equivalent of basic SQ decoding with push/pull front/back blend on the output chs. In official CBS SQ decoding Vari-Blend works only as an almost on/off blend on the rear chs by way of FET. The Sansui approch worked smoother. Still the sep specs I see on Involve SQ are much better than Sansui SQ. And no audible artifacts.

 

[Sonik Wiz]

Oops, I guess that should be “input” Toslink?

Yeah it will be interesting to see what finally evolves at Involve. There's been several posts about wish list's for the new product. Everyone has a different point of view. In addition to the features on the SMv2 I think an input level balance & phase balance or something to replicate the Sansui Synthesizer mode. All analog id fine by me as I have excellent main source in my Oppo 105 and pure analog through put on my Anthem pre-pro. For others it's an HDMI world & it's a well known thorn of contention with Chucky.
And so it goes.

 

[par4ken]

Thanks!
Ya I've been the same route squinting at the little meters on my 9940 trying to discern decoding seperation. I had much bigger meters om my old Sony 4ch R2R but still doesn't do much good for anything under -20dB. I did this the 'ol fashioned way; a DMM with shielded leads & checking one ch at a time. To do it that way for multiple decoders like you have would be cruel & inhuman punishment.

I expected the Involve/QS decoding to be quite good but I was surprised & very pleased to see how good SQ decoding was. I don't know how Involve implemented it but SQ decoding in a Sansui QS Variomatrix environment is pretty wonky. The Sansui method used the equivalent of basic SQ decoding with push/pull front?back blend on the output chs. In official CBS SQ decoding Vari-Blend works only as an almost on/off blend on the rear chs by way of FET. The Sansui approch worked smoother. Still the sep specs I see on Involve SQ are much better than Sansui SQ. And no audible artifacts.

By far the best way to check separation is with an oscilloscope. Involve does test very good. Testing via a scope revealed that the Involve uses the (proper} QS phase shift on the outputs. It tests better than the QSD-1, separation wise and phase wise. IMHO all vintage QSD-1's should be checked and adjusted, for optimum performance. At least the QS chips are not yet unobtainium! The midband decoder on mine wasn't operating correctly and that did cause a blurring of the soundfield, causing me to question the use of multibanding altogether. After adjustment things are much better. Obviously the Involve doesn't have that problem!
I installed the SQ evaluation module inside my Sony SQD-2010, switching between the two decoders the fixed blend across the rear of the 2010 is noticeably audible.

 

[chucky3042]

Thanks!
Ya I've been the same route squinting at the little meters on my 9940 trying to discern decoding seperation. I had much bigger meters om my old Sony 4ch R2R but still doesn't do much good for anything under -20dB. I did this the 'ol fashioned way; a DMM with shielded leads & checking one ch at a time. To do it that way for multiple decoders like you have would be cruel & inhuman punishment.

I expected the Involve/QS decoding to be quite good but I was surprised & very pleased to see how good SQ decoding was. I don't know how Involve implemented it but SQ decoding in a Sansui QS Variomatrix environment is pretty wonky. The Sansui method used the equivalent of basic SQ decoding with push/pull front/back blend on the output chs. In official CBS SQ decoding Vari-Blend works only as an almost on/off blend on the rear chs by way of FET. The Sansui approch worked smoother. Still the sep specs I see on Involve SQ are much better than Sansui SQ. And no audible artifacts.

It sent Dave the Bitch quite mad!

 

[chucky3042]

Yeah it will be interesting to see what finally evolves at Involve. There's been several posts about wish list's for the new product. Everyone has a different point of view. In addition to the features on the SMv2 I think an input level balance & phase balance or something to replicate the Sansui Synthesizer mode. All analog id fine by me as I have excellent main source in my Oppo 105 and pure analog through put on my Anthem pre-pro. For others it's an HDMI world & it's a well known thorn of contention with Chucky.
And so it goes.

Wash your mouth out Sonik.....HDMI.......Grrrrr

 

[chucky3042]

By far the best way to check separation is with an oscilloscope. Involve does test very good. Testing via a scope revealed that the Involve uses the (proper} QS phase shift on the outputs. It tests better than the QSD-1, separation wise and phase wise. IMHO all vintage QSD-1's should be checked and adjusted, for optimum performance. At least the QS chips are not yet unobtainium! The midband decoder on mine wasn't operating correctly and that did cause a blurring of the soundfield, causing me to question the use of multibanding altogether. After adjustment things are much better. Obviously the Involve doesn't have that problem!
I installed the SQ evaluation module inside my Sony SQD-2010, switching between the two decoders the fixed blend across the rear of the 2010 is noticeably audible.

CRO you say, I attach some test Little Max did a few years ago- with a real analogue CRO not the digital crap

 

[Electric Moo]

this is really interesting, thanks for doing all that work Sonic Wiz!

also intriguing that you can't just go by numbers to evaluate how good something sounds. It echoes some other real life audio results I have heard ( of ). For example someone I know who repairs a lot of *really* high end gear says some of it sounds like crap. But I bet the numbers look good, on paper.

 

[Sonik Wiz]

this is really interesting, thanks for doing all that work Sonic Wiz!

also intriguing that you can't just go by numbers to evaluate how good something sounds. It echoes some other real life audio results I have heard ( of ). For example someone I know who repairs a lot of *really* high end gear says some of it sounds like crap. But I bet the numbers look good, on paper.

Yes, Sir. I believe in the old audio axiom that if something sounds bad, but measures good, you're measuring the wrong thing.