Archive | 10:08

The Importance Of Accurate Decoding: SQ

26 Jul

Lt = Lf + (0.707 -j Lb + 0.707 Rb)

Rt = Rf + (-0.707 Lb + 0.707 j Rb)

It looks so simple, doesn’t it, and yet this frightenly complex little (encode) equation can cause nightmares when it comes to decoding it…, and i speak as one of those who have been affected!

SQ was but just one of Benjamin Bauer’s inventions, and it’s fair to say that it was the more popular of the two competing PAM systems during the 1970’s. But it’s not perfect by any means.

Unfortunately the one point that CBS used in early advertising against the Sansui QS system was also it’s (never mentioned) Achilles Heal, and it wasn’t until a couple of years ago it was finally dealt a blow in the name of depth, separation and imagery.

So, let’s look at what a correctly decoded SQ signal looks like:

Front Left’ = Front Left + 0.707 -j Rear Left + 0.707 Rear Right

Front Right’ = Front Right + -0.707 Rear Left + 0.707 j Rear Right

Rear Left’ = Rear Left* + j Front Left

Rear Right’ = Rear Right* + -j Front Right

* = as part of the decoding process the levels of the rear channels are automatically returned to their ‘pre-encoded’ level

And this is what you get when it’s not decoded correctly: (Assuming correct usage of +-90 degree phase shifting)

Front Left’ = Front Left + 0.707 -j Rear Left + 0.707 Rear Right

Front Right’ = Front Right + -0.707 Rear Left + 0.707 j Rear Right

Rear Left’ = j Front Left + 0.707 Rear Left + 0.707 j Rear Right

Rear Right’ = -j Front Right + j-0.707 Rear Left + 0.707 Rear Right

This is just one of many variations of what has been done, sometimes there’s only one of the phase shifts used, and there’s even been times when there’s been no phase shifting done at all!

The situation above is that the rear channel information is being output from both the fronts and rears, and at the same levels. This causes a somewhat foggy phase mess where sounds may come from the approximate direction it is supposed to come from, but can move from the rears to the fronts, or visa-versa, with the movement of the head of just leaning to the front or back.

I’m sure, buy now, you’ve noticed the elephant in the room. The major drawback to SQ is that the front channels receive no processing at all, so what you hear is the entire four channel encoded stereo information. And it’s this that limits SQ’s ability to produce a clean image with good levels of low level detail and separation.

Every single decoder from the basic 10/40 type to the king of SQ decoders, The Tate, suffer in this respect. There was nothing that could be done about it, so it was NEVER discussed. Even in our technically enlightened times there’s no way of dealing with the problem using hardware.

But, as previously mentioned, a few years ago i stumbled upon a way of dealing with this issue, and although the process is only working at 50% of its capacity, the difference it makes is quite remarkable, allowing SQ decodes to be heard without the limitations that have been accepted as the norm since it was released 47 years ago.

Oh, and the process is called “Phoenix”, and has come into use with all of the other matrix decoding process’s, with differing levels of improvement, to boot.



The Importance Of Accurate Decoding: QS

26 Jul

As a further extension of my look into the matrix systems, i thought it might be helpful, for those not aware of their inner workings, to show the importance of accurately decoding these AM sources, and what happens when it’s not done correctly.

To reduce the mathematical load on you all,  i’ll not go too deeply into those inner workings, but i must state that i will be using the terms Fc (Front Centre) and Rc (Rear Centre) which are not part of any quadraphonic system but will be used to (hopefully) help show how this area is affected in the encode/decode process.


The most popular, and successful, of the three Japanese matrix systems that were based on the RM standard, is one of the simpler PAM systems: (j = +-90 degrees)

Lt = (0.924 Lf + 0.383 Rf) + (0.924 Lb j + 0.383 Rb j)

Rt = (0.383 Lf + 0.924 Rf) + (0.383 Lb -j + 0.924 Rb -j)

Above is the QS encode equation, from which the encoder used to create the stereo compatible signal is based on. The way this system works means that each of the stereo channels contains an amount of all four channels.

The Front Channels do not have any adjustment to their phase, only having the Left and Right ‘blended’ to a predetermined amount (which is actually THE MATRIX). The effect on the Fc is that it remains in phase with the front left and right channels.

The Rear channels are similarly blended, but only after having ther phase shifted by +-90 degrees. It is this phase shifting that finally allowed four channels to be encoded and decoded without the severe limitations that crippled the  previous two AM systems. The effect on the Rc is that, because rear left is shifted 90 degrees and rear right is shifted -90 degrees, the “imaginary” point Rc is actually 180 degrees out of phase with Fc.

The “blending” is not just figures plucked out of thin air, but are the result of a great deal of work by the systems designers and is the heart of the whole matrix. The phase shifting is purely a means to an end, nothing more.

There’s one small point that needs to be mentioned here, and that is mono compatibility. I’ve read a few times now that SQ isn’t mono compatible but QS is. That is pure poppycock!

Any matrix system that uses +-90 on the rears, as both QS and SQ do, can never be mono compatible. Why? Quite simply, when the two channels are summed together to create a mono signal, any information that has been panned to be in, or around, the Rc area is cancelled out. It’s basic maths and 100% pure fact.

So, hopefully that gives something of a slight overview of what is actually on each of the channels when playing a QS record, but what about decoding?

This is where it can get very mathy, i’m afraid, so what we’ll look at is what we get out of a basic decode and a none accurate decode. The non-accurate version is just an idea because there are so many ways to do it wrong, but only one way to do it right.

So, if a particular decoder is fitted with some form of Separation Enhancement circuitry then it needs to be fed with four accurately decoded signals. What is to be expected from an accurate QS decoder is:

Front Left’ = Front L + j Rear L + j Rear R

Front Right’ = Front R = -j Rear R + -j Rear L

Rear Left’ = Rear L + -j Front L + -j Front R

Rear Right’ = Rear R + j Front R + j Front L

OK, how many thought that all you got was the main (wanted) channel? Afraid not. This is what you get when the decoder is mathematically correct, which is rather like the dodo, not seen in these times.

So, lets look at a perfectly normal output from a decoder that doesn’t follow the QS rulebook:

Front Left’ = Front L + <>Front R + j <>Rear Left + j <>Rear R

Front Right’ = Front R + <>Front L + -j <>Rear R + -j <>Rear L

Rear Left’ = Rear L + <>Rear R + -j <>Front L + -j <>Front R

Rear Right’ = Rear R + <>Rear L + j <>Front R + j <>Front L

The above assumes the correct usage of +-90 phase shifting. If that is not correct then the picture gets very messy. The use of <> indicates a variable level, which is hard to work out due to the non-correct method of decoding. Also not shown is the effect (even on the Front channels) of phase shifting caused by the mutual and un-mutual phase addition/cancellation.

Feed the above into some form of Separation Expansion circuitry and you get poorly focused imagery that wanders with movement of the head, amongst other things. Unfortunately, this is quite common.

I understand that most would just shrug their shoulders and say that it doesn’t mean anything to them, so you will be able to actually hear the difference later.

%d bloggers like this: