SQ Encoder?

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Q8

1K Club - QQ Shooting Star
Joined
Aug 30, 2012
Messages
1,751
Location
Ohio
Heya. With that Kenwood SQ encoder floating around on eBay. It is reminding me how much I would like to have an SQ encoder. Personally, can't really justify the $350 that is the asking price right now I would rather have an SQE-2000. Those are usually waaay more expensive when they come up.

So my question is. Is it possible to home build an SQ encoder for a reasonable price? If so, does anyone have any details?

Is there any way to create a software encoder that encodes in real time?

thanks!
 
I did it years ago, see my article http://www.4channelsound.com/encode.htm . I also tried software encoding and it works great, you get a more accurate phase shift using Phase Bug in Adobe Audion, than with using hardware. I thought that I could write a script to run the encode process but it seems that you can only run scripts in the edit mode, and you have to use the multitrack view. I've also tried QS encoding via software and it works well. Today with people making DTS encoded and DVDA discs, even SACDRs, there is not really any great need to encode anything. I'm sure it's possible to software encode in real time but I don't see why you would want to do it, better to keep it discrete, then encode the discrete tracks to SQ,QS or any other matrix format, that you require.
 
here at QQ some time ago was offered encoder/decoder in real time through PC and as i remember,
it works well. only downside with it, that it writen in java, which put limitation on the audio stream at 44.1KHz
 
I'm sure it's possible to software encode in real time but I don't see why you would want to do it, better to keep it discrete, then encode the discrete tracks to SQ,QS or any other matrix format, that you require.

Normally I would agree with you but I have my reasons ;)

Thanks for the link. I will read it again later cause looking at it at the moment made me :mad:@:
 
here at QQ some time ago was offered encoder/decoder in real time through PC and as i remember,
it works well. only downside with it, that it writen in java, which put limitation on the audio stream at 44.1KHz

That encoder/decoder had serious issues because it didn't implement the correct matrix.

I'm working on an encoder, just for the fun of it, and it willbe real time, but it's on on/off project at the moment
 
It's nice to see that someone is using my design. The encode is correct using the original SQ encode coefficients, it could be modified for forward or backwards oriented encoding as well. The major shortcoming is the accuracy of the phase shift network +- 10 degrees 100 Hz to 10 KHz, which can be seen from the Audio Analyzer display. A better phase shift network design could improve on this. On the bright side if a similar circuit is used to decode the errors should cancel.
 
I'm not sure if that actually does a correct encode, there's no circuit for me to be sure.

I didn't realize that the schematic was not shown in the link, it was in the original article. As I recall all we do is feed the back channels through the (basic SQ) decoder (Lb to Rt (input) and Rb to Lt (input), take the Rb decoder output and the Lb decoder output and mix these with the L and R front signals. An additional (basic SQ) decoder can be used to help maintain an equal phase relationship between front and back by passing the fronts through it first.
I'll dig up the original article and post the schematics. I never provided the schematic of the SQ decoders, they are available elsewhere, any basic SQ decoder will work (but don't use one with 10-40 blend). You can quickly try this trick by connecting a basic SQ decoder's rear outputs to your Tate or other logic type SQ decoder's input, signals fed into the inputs of the basic SQ decoder will emanate through the rear of your Tate.
 
Referring back to the article, Lb output would combine with Rf to create Rt and Rb combines with Lf to create Lt.

Afraid your missing my point. There's the inversion of the Rear Left channel on the Right. If that's not done, it isn't SQ
 
I recall in the past adding an inversion to Lb to create a forward oriented encoder. Lb comes out of the decoder inverted already (according to the published encode and decode phasor diagrams) see the explanation in the article. There are many different versions of SQ used to encode, all alter phase differently. We need + and - 90 degree phase shifts to encode back channels in SQ, anything that does that is a form of SQ encoder. The encoder described here uses the original SQ coefficients, I understand that the forward oriented encoder and position encoder were more often used latter on. With inputs swapped Lb decoder output = -0.7Lb + j0.7Rb and Rb decoder output = 0.7Rb - j0.7Lb (.7 is really 0.707 or -3dB) just as they should be.
 
Maybe I should wait till I can afford/find an SQE-2000 :mad:@:
 
A Google search brought up this link http://midimagic.sgc-hosting.com/encodsqd.htm . This illustrates exactly what I was doing. What I referred to as basic SQ encoding he's calling 4-corners. Internal encoding (where the front phase reference is bypassed) is what you get if you use only one decoder to encode the back channels. That version randomizes the phase front front to back, which helps prevent cancellations in the encode process. It also is said to be useful in creating acroperiphony or height effects, I understand that that was it's main purpose. The many possible variations of SQ encodes are enough to make your head swim. Running these different encodes back through our decoder produces correct output signals but sometimes phase shifted in odd ways. It's no wonder OD was having such trouble getting the phase just right on his decodes! RM is described as - Regular Matrix, including Dolby Surround, Sansui QS, Dynaquad, and Electro Voice. That might not be technically correct as coefficients differ and 90 degree phase shifting is used in QS and Dolby Surround (his encoder omits this) but I'm still comfortable with that definition as sometimes RM is considered to be QS without the phase shifting (rather than just another name for QS).
 
It was a year ago this month that i tired of the problems i was having with SQ decoding, and decided to dig deep. Brings on a headache just thinking about it.

EV & DY aren't considered "Phase Matrix's" in the real sense, and aren't to be confused with QS or SQ which are. I found a document sometime last year (but can't remember where i saved it to) that shows that RM was the basis of QS, Sansui just tweaked it very mildly, so it's fair to say the two are the same.

As i've said many times in the past, Dolby Surround was (stolen) based on SQ, just tweaked enough to make it incompatible.

In the SQ encoder, there was no connection between the fronts and the rears. The fronts are put through "all-pass" phase filters while the rears are put through their respective 90 degree phase shifters.

At no time is there any "connection" between them until they are mixed to create the stereo stream.

So i am a little confused about what you have written.

Later encoders (London Box, etc) were far more complex, allowing (i believe) 16 seperate channels to be encoded internally.
 
OD I agree with most of what you just said but don't know what you are getting at about a connection between front and rear??????

Phase matrix is the term Sansui uses to distinguish SQ from QS/RM.

QS uses phase shifts as does Dolby Surround to prevent cancellation in the encoder, but not as part of the actual encode process. They should not be considered to be phase matrix's. If you ignore phase of the decoded outputs phase shifting for those systems is unnecessary. The original Dolby surround is in fact the same as QS with only Cb encoded.

Dolby Pro Logic II added two surround channels and looks more like SQ and would be considered a phase matrix.
 
In your posting, you say:

"Internal encoding (where the front phase reference is bypassed) is what you get if you use only one decoder to encode the back channels. That version randomizes the phase front front to back, which helps prevent cancellations in the encode process."

There is no such thing as a Front Phase Reference in the SQ encoder. The Rears are phase shifted and mixed with no reference to the fronts at all.

Also, i'm afraid your wrong when you say:

"QS uses phase shifts as does Dolby Surround to prevent cancellation in the encoder, but not as part of the actual encode process. They should not be considered to be phase matrix's. If you ignore phase of the decoded outputs phase shifting for those systems is unnecessary. The original Dolby surround is in fact the same as QS with only Cb encoded."

The phase shifts in both SQ & QS are part of the encoding, and NOTHING to do with preventing cancellation in the encoder. I suggest you do some investigation on the way the phase matrix's actually work, and the first thing to do is ignore the large amount of incorrect information on that website.

If you do a search elsewhere here, i recently had to correct someone else's belief that Dolby Surround was based on QS. It was stolen from SQ. Find that posting, and look at the equations i supplied. There's also enough information on the internet regarding this.

By the way, in Dolby Surround, Cb wasn't encoded at all,. it was created in the decoder:

Cb = 0.707 Rt + 0.707 Lt
 
I'm sorry OD that I can't agree with you.

I guess I'll have to pull out the big guns, diagrams from US patent 3,890,466. 3,890,466 SQ Backward.jpg3,890,466 SQ Basic.jpg3,890,466 SQ Forward.jpg3,890,466 SQ Internal.jpg

All except for the internal encoder show the fronts passing through the phase reference network, how else would you keep them in phase?
 
Regarding Dolby Surround, OD I now see your point that it is like SQ in that L, R and Cf are exactly the same (same as stereo as well). Cb is out of phase from left to right also the same as with SQ. The point that myself and others are trying to make is that in the encoding of Cb the signal is phase shifted 90 degrees in the left and -90 in the right, the same as the rears in QS. And yes there is no physical Cb channel (but there could be) it's a phantom channel. Many disagreements arise from people viewing the same information from a different viewpoint!
 
Back
Top