1/2 speed remastering CD-4 records

[Doug G.]

Believe me/us, Pablo, you are NOT an *****.

An intereting thing you can do with a CD-4 record, even on a regular two channel turntable, is, place the record on the platter and the stylus on the recorded area. Just move the platter back and forth a little bit, by hand, and you will hear the carriers. Even just barely moving the platter, it will be a fairly high frequency. Figure, at 33 1/3 RPM, it's 30 kHz, at 16 2/3 RPM, it's 15 kHz, at 8 1/3 RPM, it's 7.5 kHz, at 4 RPM, it's approx. 3.75 kHz .

Doug

 

[proufo]

Believe me/us, Pablo, you are NOT an *****.

Thanks Doug. You are so kind.

I've been pondering this half-speed playing thing for a long time now. Of course I know that the whole point is to halve the frequencies, and yet I jumped and wrote what I wrote. If that is not idiocy...

Sometimes I wonder if ripping records playing them backwards makes a difference. Sort of a notion that the trailing edge of the modulations is less worn than the leading edge. If that is not craziness...

Thanks again.

 

[par4ken]

While I'm interested to see how this turns out, I think that half speed mastering would be a waste of time as it could cause as many problems as it solves. If the cartridge output at 20Hz is say 3 dB down it will now be 3dB down at 40Hz rolling off the bass. also any rumble from the turntable will be doubled in frequency and become more audible.

Audio-Technica still makes cartridges that although not specified for CD-4 apparently work well. I purchased a Sony moving coil in the mid to late eighties with specs out to 50 or 60Khz, it works much better than the Grado and the Empire that I previously used, which both were made for CD-4. I would invest in a cartridge.

As an experiment I recorded one side of a CD-4 LP @96KHz 24 bit from my pre-preamp into the sound card. The carrier and modulated difference signal are clearly visible in the spectral display in Adobe Audition.

I recall someone in a post here (I believe) tapped the base-band signal from his demodulator, recorded it to DVDA, then played the disc's back through the demodulator, feeding back into the same point. I always have trouble searching for posts on this site, so perhaps someone else could post the link.

Would be nice if someone could/would develop a method of demodulating the signal though software, not impossible just not easy.

 

[Groovy Daniel]

It's worth remembering that all CD-4 discs were mastered at 1/2 speed, and by the same major labels that were otherwise churning out wafer thin and partly recycled standard product. Low frequency roll off can be corrected. Surface noise, rumble and other distortion that doesn't vary with the musical content can be effectively removed without objectionable artifacts by software such as Izotope or Audition, which can remove just the noise that matches a sonic print of noise from, say, an outer or inner groove. What can't be fixed in the software without affecting the sonics is tracking distortion, and no cartridge can track at 30k as reliably as it can at 15k.


Sent from my iPad using Tapatalk

 

[bmoura]

It's worth remembering that all CD-4 discs were mastered at 1/2 speed

Actually some of the early CD-4 discs were cut at 1/3 speed.
Also done that way to meet the challenge of cutting such a high carrier frequency with early record lathes and equipment.

 

[ingresman]

Over the years we've had a lot of talk of software Cd-4 demodulation. I'll just add noise reduction into the mix as well. JVC used a modified version of their own ANRS system (similar to Dolby B) so that's adds another level of hardness to the software.

 

[edunkefer]

There appears to be a lot of misunderstanding regarding CD-4 so thought I would try to lay out a simple technical description of how CD-4 works and obstacles to ½ speed remastering. All of this information is from JVC technical literature.
JVC engineers had two guiding principles, Compatibility and Discreteness (fully discrete 4-channel) thus the system name CD-4.
SQ, QS systems are matrix 4-2-4 systems because of the encoder/ transmission / decoder configuration. The major drawback is that interchannel crosstalk cannot be eliminated and as crosstalk increases, channel separation deteriorates. CD-4 overcame the matrix systems issues with its 4-4-4 encoder/ transmission / decoder configuration with discrete signals. (Not saying there isn’t other issues)

The four discrete signals which are recorded are:
(CH1 + CH2) Left wall, audio signal
(CH1 - CH2) Left wall, frequency modulated signal
(CH3 + CH4) Right wall, audio signal
(CH3 - CH4) Right wall, frequency modulated signal

When a CD-4 record (disc) is played on a stereo system you hear all 4 channels of recorded audio content (Compatibility). The left and right wall audio signals contain both Front and Back content, (CH1 + CH2) and (CH3 + CH4).
By providing circuitry in the CD-4 demodulator which can perform a simple algebraic operation on these signals, the four original independent signals can be obtained. (after demodulating the FM signals)

1/2 [(CH1 + CH2) + (CH1 - CH2)] = CH1, Left Front
1/2 [(CH1 + CH2) - (CH1 - CH2)] = CH2, Left Rear
1/2 [(CH3 + CH4) + (CH3 - CH4)] = CH3, Right Front
1/2 [(CH3 + CH4) - (CH3 - CH4)] = CH4, Right Rear

Frequency Modulated 30 kHz Carrier
Technically this is not an audio signal but an RF signal (RF low frequency band). Basically it is the same as an FM radio signal. The 30 kHz carrier is modulated +15 kHz and -10kHz i.e. 25 kHz audio bandwidth. The 20 to 45 kHz frequency range takes on RF electrical characteristics and requires the use of RF coax cables between turntable and CD-4 demodulator and low capacitance tone arm wiring. I have a JVC 4DD-5 I purchased new and it came with a turntable to demodulator coax cable. If you use a standard audio cable here you will have problems with the carrier signal. To further make the point that this isn’t just an audio signal the 4DD-5 user manual has a warning that playing CD-4 records too close to a TV could cause interference (RFI) in the TV. (no FIOS in the early 70s)

Modulation
For the sake of simplicity, the modulated carriers which holds the difference signal, (CH1 - CH2) and (CH3 - CH4), are referred to as "frequency modulated" but this is not entirely true; it is actually a combination of frequency modulated signals and phase modulated signals. Below 800Hz frequency modulation is used, from 800Hz to 6,000Hz phase modulation is used and then above 6,000Hz frequency modulation is again used. This takes full advantage of the nature of both these systems of modulation. There is reduced crosstalk and an improvement in the S/N in the mid, phase-modulated range and more linearity in the low and high frequency, frequency modulated, ranges.

PLL Demodulation
A Phase Locked Loop (PLL) IC in the CD-4 demodulator is used to demodulate the frequency and phase modulated 30 kHz carrier back to the audio difference signals. (Same as an FM radio receiver) The difference signals are then fed to an ANRS system.

ANRS
As ingresman mentioned in previous post JVC designed an ANRS (Automatic Noise Reduction System) to reduce noise in the FM signal. The 30 kHz carrier is recorded 19dB lower than the audio signal to avoid interference between the audio signal and the FM difference signal. This makes it susceptible to noise. The ANRS system uses variable gain amplifiers relative signal level and frequency which adds another layer of complication. The sum and difference signals must be in the proper level to each other for the algebraic operation to derive the 4 discreet outputs that match the original inputs.

Delay Circuit
The last quirk is the delay circuit. In playback, the modulated signals must be picked up 40µsec earlier than the corresponding sum signals. This is so that the demodulation process can be performed without altering the phase relationship between the sum and difference signals in playback. It is done by using a delay circuit in the sum signal's path during the cutting process. The four signals must be in the proper phase to each other for the algebraic operation on sum and difference signals to derive the 4 discreet outputs that match the original inputs.

Hope this helps and provides some insight to the complexity of CD-4.

 

[steelydave]

That's great info. Even though I don't have a CD-4 setup, I somehow feel better knowing it.

The question is (and I posed this in another thread) if you were doing some/any/all of this in the digital domain, is there any way to increase the quality of a CD-4 decode?

It's my understanding that CD-4 (with demodulators of the time) offers something like 30dB of separation between front and rear channels, and I guess the perfectionist in me wonders if there's a way to use the precision of computers to raise that number, sort of akin to what the Surround Master and script decoding has done for QS, SQ and other matrix-based formats in the last few years.

 

[Groovy Daniel]

I must have missed where there was misunderstanding.


Sent from my iPad using Tapatalk

 

[proufo]

There appears to be a lot of misunderstanding regarding CD-4 so thought I would try to lay out a simple technical description of how CD-4 works and obstacles to ½ speed remastering.

Hello edunkefer.

What additional difficulties you think would be involved in half-speed playback?

 

[edunkefer]

Pablo, for ½ speed capture and then play back through a CD-4 demodulator (at full speed) you would probably lose the Sum Signal audio below 40Hz which at half speed would be under 20Hz and would not be captured. The FM Diff signal would not lose the low end. This may not be significant but when the Sum and Diff signals go through the final matrix this could change where the 20-40 Hz audio is heard relative to front or rear because the Sum signal would be missing the 20-40 Hz audio that would still be on the Diff signal.

I just got a TASCAM US 4X4 audio interface this week. I’ve only had time to test it with a function generator but it will capture a 45 kHz signal with a 96 kHz sample rate, likewise testing with a software tone generator it will output 45 kHz. So doing a straight capture of a CD-4 disc for use with an external CD-4 demodulator should not be a problem other than working with low output from the turntable (approx. 2.0mV) as well as controlling the 2.0mV level back into a CD-4 Demodulator.

In regards to doing a full digital process thought the easiest way to answer this would be to follow one set of the sum and difference signals through a demodulator then discuss how you might accomplish the same digitally. So let’s take the left channel signals. The player output enters the first stage of RIAA equalization, (CH1 + CH2) Left wall (Sum Signal) audio signal (20 Hz to 15 kHz) and (CH1 - CH2) Left wall (Difference Signal) frequency modulated signal 30 kHz Carrier (20-45kHz modulated).
This first stage of equalization has a RIAA standard turnover curve characteristic.
The Sum and Diff signals now split and take separate paths through the demodulator. A low-pass filter cuts off signals above 15 kHz so on this path only the Sum signal passes to a second stage of RIAA equalization. The second equalizer has an equalization curve corresponding to the RIAA roll-off characteristic. In conjunction the two stages of equalization give complete RIAA equalization to the Sum signal.
The Sum signal is now setting at the matrix circuit waiting on Diff signal. Also remember it was delayed 40µsec from the Diff signal during the recording process.

After the RIAA turnover curve equalization the Diff carrier signal goes to the PLL demodulator circuit. The output of the PLL detector is an audio Diff signal. The demodulated difference signal then passes through a low-pass filter which removes any residual carrier component to the audio difference signal.
The Diff signal is now fed through an FM-PM compensation circuit. This circuit equalizes the difference signal which was phase modulated in the recording system for the purpose of improving the S/N ratio (modulation below 800 Hz is FM, 800Hz – 6 kHz is PM, above 6 kHz FM).
The signal is also applied to an ANRS expander circuit which compensates for the ANRS compression applied in the recording system.
The difference signal from the expander circuit is now in phase with delayed Sum signal (40µsec through the PLL detection PM equalization circuits) and is applied to the matrix circuit where it is added to or subtracted from the sum signal to generate the front and rear channel output:
CH1, Left Front
CH2, Left Rear
CH3, Right Front
CH4, Right Rear
In the demodulator channel separation (front to back) is controlled by adjusting the sum signal level. While the sum signal level varies with the output of the cartridge or stylus, the difference signal level is determined by the degree of FM and PM modulation in the recording system. Therefore, the separation only has to be adjusted when the cartridge or stylus is replaced with a new one.

Capturing CD-4 and digitally processing would be very challenging. Keeping in mind each cartridge output, left and right, has two discrete signal elements that are being captured, the Sum signal and the FM Difference signal each require a different set of process steps so they would need to be isolated first. Digitally processing the Sum signal would be straight forward just RIAA equalization.
The FM Difference Carrier signal is more complex. First would be compensating for it being recorded 19 db below the Sum signal and then applying RIAA turnover curve equalization.
Next would be PLL detection/demodulation of FM Difference 30 kHz carrier signal. I am not aware of any software FM demodulation application currently available. If demodulation was possible at this point you would have an audio Diff signal (20 Hz – 15 kHz range).
Next steps would be digital compensation for PM modulation and ANRS expansion. These are not off the self apps and would have to be developed. Compensation for the 40µsec timing shift between the Sum and Diff signals would also have to be applied to ensure the Sum and Diff signals are in proper phase to each other (phase affects the algebraic summation).
The last step is digitally performing an algebraic mix of the Sum and Diff audio signals to create the 4 discrete output channels equal to the original 4 discreet inputs.
A very complex process.

 

[quicksrt]

And then one last thing, one would need ClickRepair software in there after the decoded four chans, but before playback. An absolute must imo.

 

[proufo]

A very complex process.

Indeed. Many thanks, edunkefer, for such a comprehensive description of it.

It should be kept in mind, though, that virtually all of the complexity applies also to the regular 33 1/3 playback. On top of the advantages of not requiring a fragile CD-4 playback setup, digitizing the records opens a door for preprocessing the source signals manually and simplifying the downstream processing, probably improving the overall result.

Also, I do not believe that cartridges that reach 20Hz show a sharp cutoff below that. Many preamplifiers of the past had a subsonic filter to prevent the below-20Hz signals reaching the power amp. That should indicate that it is common that cartridges reproduce signals below 20 Hz. Also, given the usual narrow-groove cutting of CD-4 records (tracking 45 kHz signals at the inner grooves is probably very difficult), it is likely that there's not a lot of low-frequency content there. I could be wrong.

Thanks again.

 

[proufo]

.