dynaquad diamond

[MidiMagic]

would that still be correct when we change the conditions as follows: mono- and stereo compatibilty is not required and our mixes are done likewise.

(for example, as i mentioned, when i want to use the 4:2 encoding mainly for easy distribution and it can contain binaural rendered content already, so that we use the 4 speakers to have full 360 degree positioning, which is not possible when you use these formats as intended.)

So are you wanting a huge number of tracks in a recording so you can use any playback method you want by choosing the correct tracks?

Or are you somehow expecting the recording to be made compatible with more things the way software is made more compatible by adding drivers or something?

I am sorry to say that a 2-channel analog recording (or even a digital recording) is limited to the actual recording it was recorded with. You can't put multiple systems into a single recording.

I can record an RM recording that plays on the Hafler Diamond, Dynaquad, Stereo-4, Scheiber, RM, QX, Sansui QS, Dolby Surround, Dolby PL-I, and Dolby PL-II only because all of these various products actually are looking for the same kind of encoding.

This encoding is (as seen in a phonograph record groove) is:
- Left is encoded as the left pickup channel (45 degree stylus motion (high near label).
- Right is encoded as the right pickup channel (45 degree stylus motion (high near rim).
- Front is encoded as both channels in phase (horizontal stylus motion).
- Back is encoded as both channels our of phase (vertical stylus motion).
- Other angles (e.g. left front) are changes in the stylus motion angle.

Here is a diagram of those record groove angles. The arrowhead direction indicates the stylus motion making the positive polarity of the decoded signal:

Note that SQ and other matrix systems use different encoding methods of putting the sounds into the record grooves. They are not compatible with these decoders. For example, SQ uses these stylus motions instead. Notice that some stylus motions are circular.

There is no way to make an RM decoder decode SQ correctly and there is no way to make an SQ decoder play RM correctly.

I did find a way to record RM and binaural into the same recording by using a special mic array. But that is an exception to the rule that different matrix systems are incompatible with each other.

I actually proved mathematically that it is impossible to put all of the possible phonograph stylus motions into a monophonic signal. Some of them will always be diminished or entirely lost,.

 

[MidiMagic]

I was making Hafler Diamond recordings in late 1970. Most of them at the time were sound effects. But some of them were recordings of friends' bands.

 

[Quadrobineg]

So are you wanting a huge number of tracks in a recording so you can use any playback method you want by choosing the correct tracks?

Or are you somehow expecting the recording to be made compatible with more things the way software is made more compatible by adding drivers or something?

i have two open questions. the one i asked here for is:

1.) i want to re-eralease my ensocder/decoder plug-ins and nee dto check if thes are correct. so what are the correct formulas for all these encoders and deocders?
i know tat some of them suck, and i know that some of them are similar to each other, but i want to follow the historical ideas and implement it as it was meant.

also on my to do list on the field of research is:

2.) how can i utilize those historcial formats to provide custom recordings (which are of ambisonic/vbap style) to create 2-tracl audio CDs to peopel which own one of these historical decoders.

I can record an RM recording that plays on the Hafler Diamond, Dynaquad, Stereo-4, Scheiber, RM, QX, Sansui QS, Dolby Surround, Dolby PL-I, and Dolby PL-II only because all of these various products actually are looking for the same kind of encoding.

my weak point is that i have no actual experience with any of these quad formats.

however i am coding DSP and know about the theories behind it, and i can assure you that in none of the quad or dolby formats you can actually place virtual sound sources freely. they have simply not enough speakers to do that. a conversion from a soundfield orientated format would be lossy and always be only a compromise. yet it would enable you to have a third option between the original format and a binarual downmix to stereo/headphones.

to find a working link between these two worlds, i need to uderstand how the historic formats work.

the answer to my question about DQ diamond was already given in post #2. and what i did in my code 20 years ago seems to be wrong.

however it means that i can more or less forget about this format. when it comes to question 2.), because it seems to be basically what we call pseudo quadrophony.

I actually proved mathematically that it is impossible to put all of the possible phonograph stylus motions into a monophonic signal. Some of them will always be diminished or entirely lost,.

while it is good to learn about these things, for my personal application it wont matter if a whatever-quad format CD would be stereo- and mono.-compatible. making an addition CD for "SQ only" and add it to the stereo mix costs 50 cents more. that "no-stereo-compatibilty" is against the specs is not a problem. i´d just wanted people with an existing historical decoder to be able to play the material.

 

[kfbkfb]

i´d just wanted people with an existing historical decoder to be able to play the material.

One advantage of using Hafler/DynaQuad surround sound encoding is that pretty good decoding can be done with only a stereo amplifier and 1 additional speaker (+wire), no special hardware decoder and another stereo amplifier are needed.


Kirk Bayne

 

[par4ken]

IMHO all matrix systems are stereo and mono compatible. The exception is centre back which will cancel in mono on every system (exception of Matrix H and UHJ). That should not be a big concern, just don't put anything important back there if mono playback is a concern.

SQ was always the most popular vintage system but is a bit less compatible with the others. Dynaco diamond encoding will still decode properly via SQ, (Lf,Cf,Rf,Cb). SQ encoding is not very effective decoded via RM (QS), EV4 or Dyna.

As per MidiMagic RM type encoding is easy to do, including the EV and Dyna variants. For QS you have to phase shift the rear signals by +-90 ° which is easy to do with software. QS has the advantage of no cancellation of any signals within the encoding process.

https://en.wikipedia.org/wiki/QS_Regular_Matrix
https://en.wikipedia.org/wiki/Dynaquad
https://en.wikipedia.org/wiki/Stereo_Quadraphonic

 

[Quadrobineg]

those who are documented somewhere were easy. but someone must have been deleting the magic diamond info from the internet.

only matrix-h is a bit a b*tch because it requires FFT for the phases. (=latency)

 

[par4ken]

those who are documented somewhere were easy. but someone must have been deleting the magic diamond info from the internet.

only matrix-h is a bit a b*tch because it requires FFT for the phases. (=latency)

There really is nothing much to document with "the magic diamond". It is as simple as it gets. Just stereo with out of phase information added for the back! Kirk already supplied an "Audio" magazine link, describing the process.

The only thing to note is the that the back signal is mixed out of phase in the right stereo channel, while in phase in the left. It could have been done the other way around. Also (F) should be mixed 3dB lower than (L) and (R) to make the sound level balanced the same as is done with regular stereo. Likewise for the (B) signal.

I always thought of Matrix H as a totally unnecessary science experiment supported by the British public!

 

[MidiMagic]

IMHO all matrix systems are stereo and mono compatible. The exception is centre back which will cancel in mono on every system (exception of Matrix H and UHJ). That should not be a big concern, just don't put anything important back there if mono playback is a concern.

SQ was always the most popular vintage system but is a bit less compatible with the others. Dynaco diamond encoding will still decode properly via SQ, (Lf,Cf,Rf,Cb). SQ encoding is not very effective decoded via RM (QS), EV4 or Dyna.

As per MidiMagic RM type encoding is easy to do, including the EV and Dyna variants. For QS you have to phase shift the rear signals by +-90 ° which is easy to do with software. QS has the advantage of no cancellation of any signals within the encoding process.

- The center back position is useful for extra reverb, an ambience pickup, or sounds from a live audience.

- SQ was not "the most popular" system. It was the most prolific due to the clout of Colombia Records. It was not popular to me because it was the worst system for recording concert hall ambience. First, there was only 3dB separation between the ambience and the program material. Then second, the gain riding in the decoders removed most of that. They faked ambiance with a lot of reverb.

The real purpose of the phase shift in the QS back channels was to make encoding an already mixed discrete quadraphonic tape work. When live music is being encoded, the phase shift is not necessary.

There really is nothing much to document with "the magic diamond". It is as simple as it gets. Just stereo with out of phase information added for the back! Kirk already supplied an "Audio" magazine link, describing the process.

The only thing to note is the that the back signal is mixed out of phase in the right stereo channel, while in phase in the left. It could have been done the other way around. Also (F) should be mixed 3dB lower than (L) and (R) to make the sound level balanced the same as is done with regular stereo. Likewise for the (B) signal.

I always thought of Matrix H as a totally unnecessary science experiment supported by the British public!

I thought the phase reversal between the back and the right speaker (the "hole" in quadraphonic encoding jargon) would be a major problem. But when I actually encoded and played material using the system, it proved to not be a problem.

Every matrix system has at least one hole. SQ has two. But iof you have the right equipment, the hole can be moved away from the sound being encoded for each sound separately.

The purpose of BMX, UMX, UD4, H, and UHJ is avoiding the cancellation if center back in mono play.

 

[par4ken]

The original Dolby Surround built upon the Dynaco Diamond configuration. They added undesirable (for music) features like limited bandwidth surround channel and Dolby B. The Circle Surround 4.2.4 encoder used the same idea but with full bandwidth channels.

The notable feature of both was the inclusion of 90° phase shifts for the surround channel. As with QS that modification permits the panning from front to back, which is not possible without it!

My Circle Surround Encoder experiments are described in this thread.

https://www.quadraphonicquad.com/fo...universal-synthesizer.5331/page-6#post-557749

 

[par4ken]

 

[MidiMagic]

The original Dolby Surround built upon the Dynaco Diamond configuration. They added undesirable (for music) features like limited bandwidth surround channel and Dolby B. The Circle Surround 4.2.4 encoder used the same idea but with full bandwidth channels.

The notable feature of both was the inclusion of 90° phase shifts for the surround channel. As with QS that modification permits the panning from front to back, which is not possible without it!

My Circle Surround Encoder experiments are described in this thread.

https://www.quadraphonicquad.com/fo...universal-synthesizer.5331/page-6#post-557749

Actually, what the 90 degree phase shift does is allow the f-to-b panning from a discrete quad recording to be encoded. It also allows a discrete quad output from a mixer to be encoded.

Using my mixer as the encoder (with my encfoder insert) lets me pan f-to-b (and b-to-f) without the 90 degree phase shift. See my diagrams in an earlier post.

 

[MidiMagic]

would that still be correct when we change the conditions as follows: mono- and stereo compatibilty is not required and our mixes are done likewise.

(for example, as i mentioned, when i want to use the 4:2 encoding mainly for easy distribution and it can contain binaural rendered content already, so that we use the 4 speakers to have full 360 degree positioning, which is not possible when you use these formats as intended.)

The recordings I make are totally compatible with stereo.

Any sound source placed at center back will disappear in mono play. I just don't put any primary material at center back if I expect mono playback.

This system allows RM plus binaural:

SURROUND FIELDS

 

[par4ken]

Actually, what the 90 degree phase shift does is allow the f-to-b panning from a discrete quad recording to be encoded. It also allows a discrete quad output from a mixer to be encoded.

Using my mixer as the encoder (with my encfoder insert) lets me pan f-to-b (and b-to-f) without the 90 degree phase shift. See my diagrams in an earlier post.

You might be able to pan from L to B or R to B but pans from Cf to Cb (F to B) are not possible without the phase shifters. You can not encode a signal in phase and out of phase at the same time. The signal will cancel in the right Rt channel so that panning to or through the (full) centre of the room is not possible.

Lt = L + .707F + .707B
Rt = R + .707F - .707B

In the case of a F to B pan F = B let's call that C
We want to be 3dB down at the mid position so that gives us .5 instead of .707.

Lt = L + .5 C + .5C = L + C
Rt = R + .5C - .5C = R

The C signal totally cancels out in the Rt channel at the mid pan position. It remains in the LT channel however.

 

[MidiMagic]

You might be able to pan from L to B or R to B but pans from Cf to Cb (F to B) are not possible without the phase shifters. You can not encode a signal in phase and out of phase at the same time. The signal will cancel in the right Rt channel so that panning to or through the (full) centre of the room is not possible.

Lt = L + .707F + .707B
Rt = R + .707F - .707B

In the case of a F to B pan F = B let's call that C
We want to be 3dB down at the mid position so that gives us .5 instead of .707.

Lt = L + .5 C + .5C = L + C
Rt = R + .5C - .5C = R

The C signal totally cancels out in the Rt channel at the mid pan position. It remains in the LT channel however.

Here is how I do it:

Pan LF to LB:
Start at the LF diagram. Turn pan left to get to LS. Depress 3/4 button to go to back half. Turn pan right to get to LB position.

Pan LB to LF:
Start at the LB diagram. Turn pan left to get to LS. Release 3/4 button to go to front half. Turn pan right to get to LF position.

Pan RF to RB:
Start at the RF diagram. Turn pan right to get to RS. Depress 3/4 button to go to back half. Turn pan left to get to RB position.

Pan RB to RF:
Start at the RB diagram. Turn pan right to get to RS. Release 3/4 button to go to front half. Turn pan left to get to RF position.

Pan clockwise LS to LS:
Start at the LS diagram. Release 3/4 button to go to front half. Turn pan right to get to RS. Depress 3/4 button to go to back half. Turn pan left to get to LS position.

 

[par4ken]

Here is how I do it:

View attachment 89709

Pan LF to LB:
Start at the LF diagram. Turn pan left to get to LS. Depress 3/4 button to go to back half. Turn pan right to get to LB position.

Pan LB to LF:
Start at the LB diagram. Turn pan left to get to LS. Release 3/4 button to go to front half. Turn pan right to get to LF position.

Pan RF to RB:
Start at the RF diagram. Turn pan right to get to RS. Depress 3/4 button to go to back half. Turn pan left to get to RB position.

Pan RB to RF:
Start at the RB diagram. Turn pan right to get to RS. Release 3/4 button to go to front half. Turn pan left to get to RF position.

Pan clockwise LS to LS:
Start at the LS diagram. Release 3/4 button to go to front half. Turn pan right to get to RS. Depress 3/4 button to go to back half. Turn pan left to get to LS position.

That is all well and good, panning left to right and front to back across the sides. The problem is how to pan from centre front to full centre and to centre back. You did not describe that process. Your diagram is showing "overhead cables required" and the Aux button pressed which I assumes is a separate microphone. That still cannot be mixed in without adding a phase shift as it can not be in phase (front) nor can it be out of phase (back) in the final mixdown.

i don't mean to sound overly critical, you are showing how an awful lot can be achieved using a simple matrix, avoiding the all pass phase filters can only have a positive effect on overall sound quality.

 

[Quadrobineg]

it is totally legit to use an analog mixers to create a surround panner, but you can still not properly place a sound source when speakers have distances of 60 degrees or more.

which is why old school quad setups arouse my interest still far more than f.e. prologic I, which simply "does not do what it is told to". (most) quad at least have equal distances.


regarding the phase shifts in some systems: this is where it makes the most sense to use digital encoding tools, because that works a dozen times better there - and almost free of cost.
same with circle surround vs. dolby surround; i bet one of the reasons why they badlimited at 7khz the rear is that they used bad analog filters for the phase stuff.

 

[par4ken]

it is totally legit to use an analog mixers to create a surround panner, but you can still not properly place a sound source when speakers have distances of 60 degrees or more.

which is why old school quad setups arouse my interest still far more than f.e. prologic I, which simply "does not do what it is told to". (most) quad at least have equal distances.


regarding the phase shifts in some systems: this is where it makes the most sense to use digital encoding tools, because that works a dozen times better there - and almost free of cost.
same with circle surround vs. dolby surround; i bet one of the reasons why they badlimited at 7khz the rear is that they used bad analog filters for the phase stuff.

Circle surround didn't bandlimit the surround. Dolby did because of noisy optical movie soundtracks.

I agree that if you have to use phase shifting for encoding then digital is the way to go.

I will always disagree with those that claim that sounds can't be properly panned in quad. There are many discussions about this in other threads but if the speakers are placed close enough together around the listener then PWM works just fine. Placed far apart and you have holes especially to the sides just like hole in the middle stereo when the speakers are too far apart. In a typical listening room the speakers are almost certainly too far apart to produce proper side images, but to me that is a moot point, outweighed by just what can be accomplished with just four speakers.

If speakers are placed the way that I often advocate with the back speakers off to the sides then the "sixty degree rule" can be satisfied except for the rear speakers, but they would fall under the "headphone rule". Placed as such there are no holes. You do sacrifice imaging behind you however, imaging is fine all around in front of you and sideways over your head.

With modern surround they always talk about speaker placement in terms of degrees. In the seventies speaker placement was never described in that way. Many different speaker arrangements were once suggested, none right or wrong. No hard and fast rules (I hate rules) it was always suggested to arrange the speakers dependant on your particular listening room and your own personal preference. It would seem that today you require a special home theatre room with the proper Dolby approved speaker placement!

 

[MidiMagic]

That is all well and good, panning left to right and front to back across the sides. The problem is how to pan from centre front to full centre and to centre back. You did not describe that process. Your diagram is showing "overhead cables required" and the Aux button pressed which I assumes is a separate microphone. That still cannot be mixed in without adding a phase shift as it can not be in phase (front) nor can it be out of phase (back) in the final mixdown.

i don't mean to sound overly critical, you are showing how an awful lot can be achieved using a simple matrix, avoiding the all pass phase filters can only have a positive effect on overall sound quality.

That image is from an article that describes using a short delay to provide the overhead effect (using B. Bauer's acroperiphony effect). The aux send feeds the delay, which then feeds the stereo aux return as a mono signal. It is harder to pan because you have to adjust the pre-fader send and the fader at the same time.

I usually just don't provide the send when I mix a piece. I keep all of the instruments in a circle around the listener. I reserve center back for reverb and audience pickup feeds.

 

[MidiMagic]

it is totally legit to use an analog mixers to create a surround panner, but you can still not properly place a sound source when speakers have distances of 60 degrees or more.

which is why old school quad setups arouse my interest still far more than f.e. prologic I, which simply "does not do what it is told to". (most) quad at least have equal distances.


regarding the phase shifts in some systems: this is where it makes the most sense to use digital encoding tools, because that works a dozen times better there - and almost free of cost.
same with circle surround vs. dolby surround; i bet one of the reasons why they badlimited at 7khz the rear is that they used bad analog filters for the phase stuff.

Remember that all of this digital sound stuff did not exist for the general public when Dolby Surround debuted in Star Wars. They "badlimited" (huh?) to 7KHz because they wanted to keep the dialog sibilants out of the surround channel.

Pro Logic I does what I tell it to. I turn the pan pot and set f/b and the image goes wherever I turn the pot.

 

[fizzywiggs41]

Just to add to this discussion, thought I'd include the bit from this article announcing the first quad matrix records , from Audio magazine February 1971.

@par4ken , notice the Dynaco album ..re release date ..could have been as early as 1970 , but at least Jan 71, the month prior to this article.
In any case , very early pressing.

This article is worth the full read , as it also discusses Duane Coopers work with his system. On page 2.

(from the fizzy library )