Stereo-4?

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
My pleasure...who would have thought somebody would look back at a low-paying job with Rat-Shack with FOND memories? But Quad was new, and exciting, and something this boy couldn't resist exploring...for the next four decades, hah-hah!
TB
 
According to Feldman, he was contacted by Fixler, who had come up with something independent of Scheiber. Fixler's was intended for Quad headphones, but proved to work great with speakers. The two of them worked together, adjusting the parameters in real-time, the encoder going straight into the decoder while they listened and tweaked it!

They sold it to EV, but the final parameters took into account things like mono-mixdown compatibility, after being fed into a computer for the final equations EV used. What kind of computer EV supposedly had access to in 1970 I do not know lol.

Also significant to me is that Scheiber's was concieved from the beginning to incorporate logic, while EV was intended to work correctly without the expense.

I don't know why, but retro-fitting an EV decoder with logic seems extremely cool to me. I guess you'd have to see the sound distribution illustrations in Feldman's book to appreciate just how evil SQ and QS really are. Note that Feldman does NOT present them that way himself. It was a decade before "stereo image expanders" used phase relationships to affect apparent separation, and neither CBS nor Sansui had any idea what a mathematical abomination their systems really were. You could use a stereo image expander to compensate for the reduced seperation in EV or Dynaco, but in SQ or QS it would just make everything a whole lot worse. The Tate is so great because it uses logic to kill crosstalk by injecting reverse phase signals into other channels, but with EV or Dynaco or any other amplitude matrix, a stereo image expander would do the exact same thing without logic, because there are no whacky assymetrical phase angles like with the other two.

So say a Left Front signal has a 20% bleedover into the Right Front. No problem, just have a fixed setting to inject that exact amount out of phase into that channel, no logic necessary.

I can't wait to try this :)

Hello Dreadstar!

Your idea of injecting anti-phase into adjacent channels to kill the cross-talk?
I have chased that dream for 30 years, but STILL believe it is not the mathematical impossibility I've been told it is...but then, I'm no Einstein.
I played around with real and imaginary matrices, and always got rid of one cross-talk...and ended up with another.
You have inspired me to go back into my ancient notebooks and take another look! I did try your idea many moons ago, using the QS matrix, but without the phase-shifting. By injecting -3 dB. anti-phase into both adjacents, the cross-talk DOES disappear...only to show up full 0 dB., and out-of-phase, in the diagonal...ouch! And when the signal is multi-channel, say center-right, it gets even crazier. (Maybe this could be turned to something useful, like boosting the level in centers between pairs, to fill in that "hole" we are told we get with speakers spread out so far apart?)

If a rank amateur might offer a few ideas, maybe dealing with PAIRS of channels would lead somewhere interesting? Or, treating each channel at a separate stage in the game? Chasing after this so-called math-dream was fun for me for years, even if it lead nowhere. One idea that DID show promise, though I dropped it, was HALF-WAY canceling the cross-talk, then canceling some of the resultant (new) cross-talk, etc., for several stages.
Might have been hear-say, but an audiophile years ago told me that Peter Scheiber had actually built a (gulp!) 32-stage decoder, and obtained incredible separation. Wish I knew whether to buy into THAT story or not!

OK, I'm new here, so I better quit rambling, but thanks for inspiring me to take another look at these ideas!

Regards from all corners,
TB
 
Here is the engineering math to EV-4 as well as all other basic 4-2-4 matrices.
https://www.quadraphonicquad.com/forums/attachment.php?attachmentid=2150&d=1239994229
As you will notice, SQ has its' rear channels encoded into 67.5 and 112.5 left-phase and EV-4 has it's rears at 67.5 and 112.5 right.

Which is how they supposedly got 12 channels out of a supposed 4-2-4 matrix in their tests from the early 80's, laying opposite matrices upon one another and then trying to decode. And where they eventually got the ideas for 7.1 matrix. Yes it was fun to listen to.

Yes it was also Halloweenvile/Ghost City and not at all discrete. Of course they never used it commercially from too much swimming around, that and you couldn't take steering logic circuity to 12 channels yet. You probably could these days.
It is also said that one can introduce the inverse matrix (SQ into EV-4 or vice-versa for example) to kill crosstalk.
 
Last edited:
Right.

Of course, I'm still wondering the one thing that still hasn't ever been answered. How would you software decode it?

Although, at this point I would probably work on the Q8 over the vinyl.
 
Probably rewrite the SQ script and invert the rear-phase decoding utilizing right-phase values vs. SQ's left-phase values,
since SQ is looking for 67.5 and 112.5 left and EV-4 needs 67.5 and 112.5 right.

Or, you might be able to use the Extraction tool and just look for a 45-degree slice centered on 67.5 right and another
45-degree slice centered on 112.5 right like we used to have to do for SQ on the left before SQ scripts were available.
 
This thread is old - I give it a try.
I have a realistic ARS-4 Stereo-4 decoder. The on/off master volume knob does not let the line ac thru, which then goes to the resistor for the pilot lamp, and an input side of transformer. The on/off part of volume control has markings for 3Amp/125vac, and is brown in color. Can I bypass this. The similar model is Heath-kit, but heath-kit has a fuse before this, and a transistor in circuit, which seems main differences between units. I have some stereo-4 lp's, and thought I would compare this unit, against the sansui quad, and realistic quad using QS as above mentioned. Most interesting also to just make a new decoder for the EV-4 and Stereo-4 decoding. Any suggestions appreciated, and what is this on/off current handling potentiometer/switch?
 
Probably rewrite the SQ script and invert the rear-phase decoding utilizing right-phase values vs. SQ's left-phase values,
since SQ is looking for 67.5 and 112.5 left and EV-4 needs 67.5 and 112.5 right.

Or, you might be able to use the Extraction tool and just look for a 45-degree slice centered on 67.5 right and another
45-degree slice centered on 112.5 right like we used to have to do for SQ on the left before SQ scripts were available.

Did anyone ever attempt this? Just wondering to see before I start researching how it can be done in AA 3

Thanks
 
Did anyone ever attempt this? Just wondering to see before I start researching how it can be done in AA 3

Thanks
I don't profess to be any kind of math wiz, but I don't understand what those phaser diagrams are supposed to represent. EV-4 and SQ are nothing alike, swapping the rear channels won't convert one to the other. EV-4 was modified to become EV-44 (the new encode coefficients were never used), which was almost SQ, but don't think that's what is shown here. EV-4 uses only in phase and out of phase signals.
A script shouldn't be too hard to write. I would think you would use a bit of out of phase bend to get perfect separation of the left and right front signals and then run the centre channel extractor at 180 degrees to remove the cross-talk from the rear channels. Then do likewise for the rear, first in phase bled to get full left to right separation then run the centre channel extractor to remove the cross-talk from the front.
 
I just found the meaning of those vector diagrams, the same thing is shown in a paper "QS Technical Manual"; they refer to the disc cutting angles. Remember that stereo is cut in a 45 degree V groove, mono is cut laterally side to side, 0 degrees. Out of phase signals such as record warp produce up and down motion 90 degrees.
Sansui claims that their choice of 22.5 degrees (front channels) and 67.5 (45+22.5) degrees (rear channels) is the best choice for Quad. Now what does this really have to do with script decoding, absolutely nothing!
 
Back
Top