CD-4 Ramblings and Tech Discussion

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
To match with the playback of Lou Dorren's new demodulator, we would need the reverse design as a new quadulator in order for this to work, not to mention his cutterhead improvements and re-designs for 3/4 speed cutting instead of the former half-speed or two-thirds-speed cutting previously required.

I'd suggest doing the CD4 modulation (and ANRS, and predistortion) on the digital domain and using a very high quality DAC. That way you can get everything under control. Of course, it won't be analog anymore but the end result may as well be superior than using a 1974 demodulator.
 
Neutrex-I and Neutrex-II originally came out of a 1960's RCA research project that had as its goal 'better' LP sound for all listeners, but especially on the low-cost phonograph 'console' units that RCA liked to sell (RCA could never ignore cheap now, could they?) - European, UK and Japanese LP's were being imported and sold in droves due to their high sound quality and RCA, (instead of just using better vinyl and mastering quality with more care in pressing the LP in clean-room conditions) chose to try a technical end-run around the problem. The RCA R&D project was code-named "Dynagroove", as the final LP system came to be called, and one part, the Neutrex groove pre-distortion, later got the name Dynagroove

Well, yes and no. Dynagroove's tracing compensation is for a spherical stylus. Neutrex is designed for a stylus of a parabolic side profile, if i understood it well. Which would make sense since the preferred stylus for CD4 is the Shibata. The patent for Neutrex is a separate one (i have seen it). Moreover, many other manufacturers had their own version of tracing compensation in the 60s and 70s: Telefunken/Neumann/Decca, and Toshiba, to name a few.

Later, Dynaflex paper-thin vinyl was added to the whole shebang

Which is a very good vinyl compound that RCA used to press badly mastered records.

And JVC refers to the pre-distortion system as Dynagroove in some of their very early CD-4 tech documents.

:O

Back to CD-4 and Neutrex... There's some great diagrams and explanations of how it works on Old Quad Guys Quad info site

I'll try to find the US patent number again.

Used either for CD-4 Quadradisc mastering or stereo/mono/matrix quad LP mastering, Neutrex worked very well - It basically came into standard use by all record companies on all LP's by the mid-70's or so, and it never really fell out of use that I know of.

I seriously contend this claim. Source? As far as i've read, it is a rare thing instead of the norm. The only series i recall that used tracing predistortion were RCA's Dynagroove and Telefunken Royal Sound. The Neumann VMS 66 system had a TS 66 b "tracing simulator", which could optionally be used to apply predistortion. But the later Neumann cutting lathes' electronics didn't come with any predistortion device, AFAIK.
 
Nope, it just got the name changed to Neutrex or "groove pre-distortion", etc... The pre-distortion used by Neutrex is an approximation, of course, but the playback tracking is much more accurate and the high frequency tracking distortion is an order of magnitude lower using the Neutrex (let's not call it Dynagroove) than on an LP cut without it. In the PDF I linked to, you can see how the stylus tracks the pre-distorted groove more accurately and why it became a standard part of LP mastering.

The problem is -as mentioned before- that (1) this isn't easy to do with precision in real life conditions (although i guess it was successful)
(2) tracing predistortion (of the classic type) works only if a spherical stylus of a certain radius is used. Conversely Neutrex works only if a Shibata stylus of standard dimensions is used, if i am not mistaken.
(3) A more advanced stylus doesn't need such predistortion since it reads the groove more accuratelly,. By the late 70s, elliptical styli of 0.3 and 0.2mil side radius were more popular. Thus no more need of predistortion compensation.
(4) Try playing a Telefunken Royal Sound record with a fine elliptical or with a shibata. It sounds terrible. The same record sounds mighty fine when played with a 0.7mil conical.
 
Nice!! Very nice!! But two suggestions:
(1) Use black vinyl, it is almost always quieter than colored vinyl.
(2) Forget about record weight, it is the quality of the record compound that matters. Philips 1980s records being a prime example of thin records that sound fantastically good.
(3) If you could get something similar to Q-540 record compound, it would be great. In any case, you need a vinyl compound that can withstand more pressure. That is, its resistance to plastic deformation should be higher.

+1

Black is better than or just as good as color, and 180gm does not guarantee fidelity. I have copies of the dreaded RCA Dynaflex vinyl that are hot stampers.
 
I'd suggest doing the CD4 modulation (and ANRS, and predistortion).
Any idea about how to do ANRS encoding and decoding in the digital domain?

I believe it is the only step missing to demodulate CD-4 recordings in the PC, starting from a two-channel high bandwidth recording.

Thanks in advance.
 
Any idea about how to do ANRS encoding and decoding in the digital domain?

I believe it is the only step missing to demodulate CD-4 recordings in the PC, starting from a two-channel high bandwidth recording.

Thanks in advance.
Saludos estimado Pablo,

Well, i'm no expert (nor amateur!!) at DSP processing but there are algorithms to do dynamics compression and decompression in the digital domain. If i recall correctly, for ANRS you'll want to filter and divide the input in two spectrums (easy to do in the digital domain), and then apply different compression ratios to them.

FM modulation is extremely easy in the digital domain. But i never understood the "PM" (phase modulation) part of the CD-4 encoding. If i recall correctly, on a certain part of the midrange (around 1KHz?) the carrier is phase modulated, not frequency modulated. But i don't fully understand how (or why!). Of course, let's hope that Lou Dorren reads this posts and help us.

For the Neutrex predistortion system, i guess it would be a case of reading the patents and understanding exactly what's being applied.

There are more steps: If you're going to cut at 1/3, 2/3 or 1/2 speed, you need to adjust your RIAA recording curves accordingly, or else the EQ will be non-RIAA when played back at 33RPM. As far as i know, this happened to the CBS guys when they started their "cbs masterdisc half-speed mastered" record series. The first releases had an EQ that was all wrong.

Mind you, I'm speaking from a theorical perspective only.
 
Any idea about how to do ANRS encoding and decoding in the digital domain?

I believe it is the only step missing to demodulate CD-4 recordings in the PC, starting from a two-channel high bandwidth recording.

Thanks in advance.

I'm not sure that the above approach does not take my thoughts into consideration.

The thing that matters most to me is the fidelity, tonality, and warmth of the finished product. I may be reading more into this than is implied, but working in the digital domain without those three things leaves me wondering if going to the trouble of recreating a CD-4 type vinyl is worth the effort?
 
I'm not sure that the above approach does not take my thoughts into consideration.

The thing that matters most to me is the fidelity, tonality, and warmth of the finished product. I may be reading more into this than is implied, but working in the digital domain without those three things leaves me wondering if going to the trouble of recreating a CD-4 type vinyl is worth the effort?

Well, this assumes that you can't get "fidelity, tonality and warmth" in the digital domain. Personally i find nothing wrong with digital audio, and i have some digitally-recorded LPs that sound fantastic. But of course, we're going off topic.

On the other hand, i understand that if we're going to start from a digital audio file, we might as well just deliver a 4-channel digital audio file instead of bothering to create a CD-4 record. On the base of this, perhaps you're 100% right in asking "Is it worth the effort?"

... but it can be also be done for fun!!

Plus, it is a different thing to play a LP cut from a digital recording, mastered using a top-line, very-high-quality DAC, than to listen to the very same digital recording on the average PC or the average CD player.
 
...we might as well just deliver a 4-channel digital audio file instead of bothering to create a CD-4 record...
And then what? Twin-track the disc with two grooves one-groove-width apart from one another and have a four-pole twin-stylus cartridge? Have each stereo pair be offset by 180-degrees and play it with two tonearms and two cartridges facing each other? Have one stereo pair go from the edge of the disc to halfway-through and the other two channels start halfway-through and go to the center? Use both sides of the disc at the same time in ``opposite'' directions and be only able to play the disc on its' edge?
... but it can be also be done for fun!!
Oh it was fun alright dealing with all those crazy formats. (Name them all and get the Last Two Wedgies and the Game, Match and the Million Dollars).

I believe it is the only step missing to demodulate CD-4 recordings in the PC, starting from a two-channel high bandwidth recording.
Except the ``high bandwidth'' would have to be ridiculous in terms of a sample rate.

As we all know, the MINIMUM sample rate required to capture an analog signal into digital is twice the highest frequency, one sample for the negative swing of the highest-frequency sine wave and one sample for the positive swing.

In reality though, trying to get an even CLOSE to accurate sampling of a sine wave with only two samples is absolutely ridiculous. Ten-times or a hundred still wouldn't give you THAT good of a sampling. A thousand times BEGINS to approach acceptability.

So, say you have your 20 Hz to 15 KHz bandwith-limited CD-4 baseband audio signal, processed in this manner in order to avoid interfering with the modulated carrier wave above it. This carrier wave, while centered at 30 KHz can swing down to as low as 20 KHz or as high as 45 KHz, again, bandwidth-limited in order not to interfere.

So, already we have not one but two stereophonic bandwidth-limited signals to deal with which we'll get to in a minute, and all those nice musical overtones which give recordings their depth of presence and height of clarity all stripped away by the limiting necessary for CD-4 in both the baseband as well as the carrier wave.

Which means CD-4 has no greater bandwidth on either of it's bands than FM-Stereo which is similarly limited to 15 KHz in order to avoid interfering with the 19 KHz unmodulated stereo beacon subcarrier, nevermind the 38 KHz Left-minus-Right subcarrier needed to recreate stereo radio.

But in order for the computer to EFFECTIVELY capture the CD-4 tone ALONG with all of it's modulations, the computer would have to sample at a minimum of 900 KHz, 9,000 KHz or 90,000 KHz (of ten times, a hundred times and a thousand times respectively of twice-the-highest-frequency needed for sampling).

The only computer that can do that on a regular basis is the Cray Supercomputer Arrays at Government installations such as NASA, Jet Propulsion Laboratories or the Lawrence Berkeley Research Center. And then, not only would the CRAYS have to be ganged together in order to do it with any degree of speed, but all their hard drives and other storages would have to be assembled in a huge RAID as well JUST to be able to save it to disc fast enough.

Forget real-time recording on THAT one. You'll be there forever, like the first digital recording engineers were back in the mid `50's trying to record a three-minute selection of the First Movement of Beethoven's Fifth Symphony onto IBM punch cards.

The resulting three-minute selection was in fact transcribed, onto millions and millions of punch cards occupying floor after floor after floor of card-catalog style storage. It took three years to punch all the cards, and another year to load them all back into the computer for playback.

At eight-bit word lengths and a 10 KHz sample rate.

Playback was done for three minutes of music, and it sounded like extremely clear AM radio.

Nothing more. Baby simple. But out of that has grown..... .

Oh, sorry, what were we talking about?
Dropped into Bernstein's Omnibus show from the 50's there for a little bit.
http://www.youtube.com/watch?v=xK1BkymfaM4 Slide over to 3:25 and begin.

And then back to our other topic, the main problem with the ORIGINAL CD-4 system was NOT the mis-placement of the RIAA curve or the NAB curve from the tape being wrong, it was the fact that when you divide frequencies in a cutterhead of the period down, the result is what's known as a bass contour effect.

The bass frequencies become lost in the dirt of the bottom end of the cutterhead's frequency response (which now must engrave 10 Hz, 12 Hz, or 15 Hz respectively to accurately reproduce 20 Hz later on).

Then, due to various ``bumps'' and ``valleys'' in the not-perfectly-flat response of the cutterhead in question, rather than adding a nice pleasing effect to a vocal or lead instrument, this will now mess with the frequencies in the high-bass or low-midrange areas, making the recording sound terrible in comparison to standard stereophonic releases cut on the same lathes in the same period.

Which, especially in the early days of CD-4, they often did, but people forgave a lot in order to have double the channels.

So that's why you really couldn't master the ``two-channel modulated program'' into the computer the conventional way and retrieve the signals digitally, even if you HAD a program to do so.

Trying to load master tapes into a computer with their original supersonic bias frequency tracks intact in order to create an inverse to cancel out the miniscule analog speed variations on the original recording date creates a similar problem.

But Jamie Howarth at www.plangentprocesses.com does similar work with their proprietary-at-the-moment technology for retrieving bias frequencies off master tapes and using that to create an inverse sync and run a computer off that newly-created inverse sync without actually having to capture the original supersonic bias to do so.

So, while excellent CD-4 retrieval will still be for the foreseeable future accomplished by Lou's new demodulators, master tapes can now be restabilized using similar technology as the sync on tape carries no music modulated within it as does CD-4.

And then, you never know. New developments could be just around the corner to take care of the bass-contour problem and midrange problems incurred from mastering at half-speed, 2/3rds speed or 3/4 speed. Add to that, mastering onto DMM and mastering for playback at 45, and any way you slice it, you'll have a considerably better product than you had in 1974.
 
Last edited:
Except all Klystrons are pretty much the same and have pretty much the same non-linearities. But all the ubiquitous Astatic or Tetrad or Electro-Voice or any other ubiquitous inexpensive ceramic cartridges mounted upon the equally-ubiquitous all-plastic Garrard or BSR record changers upon which the vast majority of Dynagroove pressings were played, do not have even CLOSE to the same distortions

Dynagroove (and other pre-distortion methods) correct for the distortion created by a conical stylus of a specific radius (typically 0.7mil). The ubiquitous ceramic cartridges and the ubiquitous cheap magnetic cartridges and other ubiquitous cartridges use EXACTLY the same ubiquitous 0.7mil conical stylus, which will produce EXACTLY the same tracing distortion that the pre-distortion system is trying to cancel.
 
And then what? Twin-track the disc with two grooves one-groove-width apart from one another and have a four-pole twin-stylus cartridge?
Non-sequitur. I never proposed such a thing. I was talking about a four-channel digital audio file as a quad medium. A digital file to be played with a PC and a DAC.

Except the ``high bandwidth'' would have to be ridiculous in terms of a sample rate.

As we all know, the MINIMUM sample rate required to capture an analog signal into digital is twice the highest frequency

Yes. Thus you'd need 100KHz sampling rate to safely capture non-decoded CD-4 audio.

In reality though, trying to get an even CLOSE to accurate sampling of a sine wave with only two samples is absolutely ridiculous.

No. Unless you want to believe what some audiophile magazines (devoid of science and common sense) publish.

Ten-times or a hundred still wouldn't give you THAT good of a sampling. A thousand times BEGINS to approach acceptability.

No, that isn't true. Review the nyquist sampling theorem.

But in order for the computer to EFFECTIVELY capture the CD-4 tone ALONG with all of it's modulations, the computer would have to sample at a minimum of 900 KHz, 9,000 KHz or 90,000 KHz (of ten times, a hundred times and a thousand times respectively of twice-the-highest-frequency needed for sampling).

No, that isn't true. Review PCM and its theory.

The only computer that can do that on a regular basis is the Cray Supercomputer Arrays(...) recording engineers were back in the mid `50'(...) millions and millions of punch cards (...) Nothing more. Baby simple. (...) Dropped into Bernstein's Omnibus show from the 50's

That is not relevant. Please stay on topic.

And then back to our other topic, the main problem with the ORIGINAL CD-4 system was NOT the mis-placement of the RIAA curve or the NAB curve from the tape being wrong,

I never said such a thing.

Then, due to various ``bumps'' and ``valleys'' in the not-perfectly-flat response of the cutterhead in question, rather than adding a nice pleasing effect to a vocal or lead instrument, this will now mess with the frequencies in the high-bass or low-midrange areas, making the recording sound terrible in comparison to standard stereophonic releases cut on the same lathes in the same period.
Which, especially in the early days of CD-4, they often did, but people forgave a lot in order to have double the channels.

1. Feedback cutters, which exist since 1958, do have a perfect flat response from 40Hz to at least 16KHz, providing you use enough feedback.
2. Half-speed cutting was already perfected by the Decca/Telefunken team in 1958. That's what they used to master the Decca FFRR records.
Let me say it again. Half speed cutting problems were already solved.
In 1958.
With a prehistoric Neumann cutting head.
Which was so old it wasn't even branded "Neumann" but "Telefunken-Decca"
Which was so old its armature wasn't even 45-45 degrees but horizontal-vertical.
Go listen to those Decca/London FFRR records. They sound beautiful. They do have enough bass.

So that's why you really couldn't master the ``two-channel modulated program'' into the computer the conventional way and retrieve the signals digitally, even if you HAD a program to do so.

You were talking about cutting records at half speed and now you're talking about transcribing CD4 records to digital. Non sequitur!

But Jamie Howarth at www.plangentprocesses.com does similar work with their proprietary-at-the-moment technology for retrieving bias frequencies off master tapes and using that to create an inverse sync and run a computer off that newly-created inverse sync without actually having to capture the original supersonic bias to do so.

Well, they DO capture the supersonic (>60KHz) bias. Which, curiously enough, is what you just said it was practically impossible to do on your last post. And no, they don't use a hyper expensive Supercomputer to do it. It can be done with a 192KHz DAC if you play the tape at half-speed. So there you are, they capture bias frequencies of over 100KHz with a 192KHz DAC. Don't believe me, believe the very same Jamie Howart:

http://www.gearslutz.com/board/6302179-post19.html

You just posted that for capturing the 50KHz bandwidth of CD4 you need 90000KHz of sampling frequency. This isn't true. Thus, i insist you should check out the Nyquist sampling theorem. Science, it works.
 
I hate to say it...

There is no rhyme or reason to an Analogue CD4 medium today.

Producing new multitrack recordings - you might as well use the DVDA format... or any of the current PC audio multitrack container formats.

Any CD4 on vinyl will struggle with all the compromises involved, and will be compromised by it in the end.

Which is not to say we should not have CD4 rigs - the vintage material is there and should be enjoyed!

If someone is already making a multi-track recording and wants to issue it in CD4 for those who prefer the rituals of analogue to those of digital - cool!
But it is just a fun thing to have and not an attempt to produce something superior.

Standard stereo LP's can sound superb, and do so to a substantial degree because of the efforts that went into making CD4 work - they led to a quantum jump in record technology/capabilities throughout the chain.

Let's face it - Digital is superior in 99.9% of cases and almost all top end recordings today are done digitally...

Producing CD4 recordings today is like producing Pianola rolls.... nothing wrong with it and good fun. But lets not stress about it!

bye for now

David
 
Hi David, i understand your point. But for example we can say "what is the point on releasing LPs from a digital master today?" Yes, there is a point, as i've said before: First, the LP is a good enough medium to transport such audio with no subjective loss of quality. Second, people who have a high-end turntable system not necesarily have a high-end D/A converter. Third, the LP medium is aesthetically more pleasant.

So that above applies to stereo records, and i'd venture to say it can also apply to CD4 records. It looks as a complex system, David, but isn't THAT complex. In any case, yes, let's do it for fun.

Greetings! Nice to see you in this forum too!
 
Saludos estimado Pablo,

Well, i'm no expert (nor amateur!!) at DSP processing but there are algorithms to do dynamics compression and decompression in the digital domain. If i recall correctly, for ANRS you'll want to filter and divide the input in two spectrums (easy to do in the digital domain), and then apply different compression ratios to them.
Hola de nuevo, Flavio.

It seem to be more complex than compressing and decompressing, as the processing is frequency-dependent.

I did some searching on the basics and got stuck at ANRS.
 
Except the ``high bandwidth'' would have to be ridiculous in terms of a sample rate.
Hello ndiamone.

I don't follow this.

Won't 24/96 capture up to 48 Khz?

I believe there are people here that have made digital files of CD-4 records and played them through a demod with success.

Would that prove anything?
 
I'm not sure that the above approach does not take my thoughts into consideration.

The thing that matters most to me is the fidelity, tonality, and warmth of the finished product. I may be reading more into this than is implied, but working in the digital domain without those three things leaves me wondering if going to the trouble of recreating a CD-4 type vinyl is worth the effort?
Hello ress.

Looking for aural bliss in CD-4 reproduction may not be a good bet.

Hopefully, we will soon see if Lou's demodulator brings us a step closer.

Given the advances in digital recordings and players, maybe the optimum path is to process existing mixes and play back via DVD-A/SACD media. Or digital files. Just as the generous converters already do now.

Anally, I'd love to have digital recordings done from two-channel, but I'm sure using a four-channel recording from a good demod would be virtually the same or better, as a starting point. What would be needed is to assess the compromises made for CD-4 cutting and compensate for them as possible, all in the digital domain.
 
hmm and no need to waste too much space - CD4 is 15k bandwidth limited - so digital version would be either 1x24/96 (to 48kHz) or 2x24/32 (to16kHz)

2 x 24/32 will take less space and be easier to play back (no decoding) - but harder to create (has to be decoded into the file packages)
 
Back
Top