Working CD-4 (software) Demodulator!

[tomitaisao]

Hi all. New to this but I’m an engineer and fascinated by the software vs hardware approach. I have just ordered Tomita Bermuda Triangle on CD-4 and wonder has anyone tried anything with this album before? Richard, I’ll happily send it on to you when I get it if you think you can do something (assuming it is in okay condition).

Most of Tomita’s CD-4 works (of which I believe there are 7) have made their way to SACD - most recently Daphnis et Chloe (also it was the last CD-4 release I believe according to this forum).

Firebird and Bermuda Triangle (apart from some tracks on Space Fantasy SACD) have not in full. See https://www.quadraphonicquad.com/fo...ta-sacd-and-their-quadraphonic-sources.24877/ by @winopener

This much quoted review gives a good overview of the SACDs https://www.amazon.com/gp/customer-...ef=cm_cr_dp_d_rvw_ttl?ie=UTF8&ASIN=B01GUHGWR6

John from isaotomita.net

 

[kfbkfb]

Page 2 post 24 has a circuit design for a Needledrop Amplifier.

Page 3 post 41 has information about running the Stereo Lab CD-4 Decoder with Windows 10.

Kirk Bayne

 

[Brice]

tomitaisao Thank you for your offer. As a general point to all, I should be delighted to perform demo decodes of CD-4 material; either from the original medium or from recorded, non-equalised needle-drops.

Remember too that a free demo version Stereo Lab exists which you can download and use for up to ten decodes.

Richard

 

[quattro64]

This is very exciting, I think I can get this working on my system and test it out. I'd also eventually like a version of this for max separation and max fidelity. For now though just getting it to work ok for most material is fine. I do wonder though, Will there be a version, or a setting in some future version, for those of us using the Panasonic SG cd-4 carts?? SG= Strain Gauge.

 

[Circular Vibes]

This is very exciting, I think I can get this working on my system and test it out. I'd also eventually like a version of this for max separation and max fidelity. For now though just getting it to work ok for most material is fine. I do wonder though, Will there be a version, or a setting in some future version, for those of us using the Panasonic SG cd-4 carts?? SG= Strain Gauge.

I have strain gauge cartridges as well, but it requires a voltage across each channel to work with resistance. I seriously doubt it can be applied in software, you would need the preamp for your cartridge to go into the computer this way. If you go this way, you already have CD4 I have to assume, short of Sound smith or Win Labs, to my knowledge.

 

[quattro64]

Well my plan was to build my own high end version of a stereo pre for my pana SG cart, I understand it might not be compatible with this software. I still have old CD-4 hardware that I can use for playback. I have other normal, high end carts and flat mm phono stage where I could use this. Hoping there will be a great 'fussy' version of this software available some day for a dedicated setup to do rips.

 

[Brice]

Dear Quattro64 and Circular Vibes,

There are three things which distinguish an SG cartridge from a standard MM/MC:

1) As you corectly say, they require a bias voltage to the semiconductor elements.
2) They do not require RIAA correction because they sense stylus displacement (and not velocity).
3) One channel is polarity inverted with respect to the other.

You're right, I can't help with the polarising voltage: that's hardware. But I could adapt the software to NOT implement the RIAA correction on the baseband signal, and have it invert the polarity of one of the channels.

At the very least, that would simplify the SG preamp.

Let me know if this would be of interest and I will put it in the development schedule.

Best wishes,

Richard

 

[quattro64]

Hmm. Thanks Brice! After reading about how your software tech worked yesterday I was thinking this wouldn't be compatible with SG because the rip wouldn't have the right carrier signal, cause of the RIAA that is inherent in the SGs eq curve. If you could make it compatible with SG carts that would be wonderful. I am planning to build a discrete SG pre circuit based on a thread on diyaudio. I think it is entitled Balanced FET Phono Pre for Strain Gauge(or something like that). A guy named Hazard. A software solution after this for demoding CD-4? Holy Grail! thanks

 

[Brice]

Hi Quattro64. No, the Panasonic cartridges are very suitable for CD-4 rips. The only issue is that (at present) we apply RIAA to the baseband signal which we would simply bypass when a strain-gauge cartridge was used for the needle drop. We can add this as an option in the CD-4 preferences tab in the next version.

 

[quattro64]

Thank you, that would be great!

 

[Brice]

Looking around here and elsewhere on the Internet for a preamp design to test a strain-gauge Panasonic cartridge, I realised the very simple pre-amp I designed (on page 2 of this thread) may be easily adapted for an SG cartridge. The limitation is the very low rail volts dictate and low value of bias resistor (1.2k for 4mA on 9V rail). Because this appears effectively in shunt across the cartridge, this reduces gain which is a shame in a low noise stage. Obviously I could arrange a couple of current sources, but - since the aim of the design was to be a minimal as possible - I tried to think of a simpler way. There is...

The bias resistor may be bootstrapped to may it "look" like a much higher value than it really is. The output of the non-inverting gain-stage is an ideal place to derive the bootstrap EMF as shown in the diagram attached.



Positive feedback must - of course - be treated with caution, but, provided the fraction rr/( Rp +rr) is smaller than r/(Rf+r), the circuit stays stable when there is no cartridge fitted. I found that, if the bootstrap fraction was some 16% below the op-amp negative-feedback fraction, the gain reduction due to the bias resistor was reduced from 6dB to less than 1dB with no sign of instability from the preamplifier.

 

[quattro64]

Very nice! A simple solution to those looking to use their Technics/Pana SG carts. Your site by the way is excellent, plenty of reading there for Vinyl newbs and old collectors as well.

 

[Brice]

Very nice! A simple solution to those looking to use their Technics/Pana Sg carts. Your site by the way is excellent, plenty of reading their for Vinyl newbs and old collectors as well.

Thank you Quattro 64.

 

[mattdp]

Hey folks! I'm the author of the aforementioned blog post detailing how to decode CD-4 via SDR#.

Very glad to see someone develop decoding into a standalone piece of software! Although I'm a PC user, I might have to look into getting a virtual machine together to run Mac OS for this. Stereolab - you guys have done a fantastic job with the documentation on CD-4 playback and such. I did want to comment that I have cleaned my entire collection via ultrasonic cavitation and cannot recommend it highly enough. It can really get the crud out.

I would also be in favor of a "discrete" decoding option, rather than Ambiasonics.

I'm curious to know if Stereolab implements (or has plans to implement) the ANRS noise reduction system present on the subcarriers (documented on page 7: http://www.cieri.net/Documenti/Altri marchi/JVC - All You've Ever Wanted to Know about the CD-4 Disc System (01-1973).pdf). Also, does it factor in whatever delay time exists between baseband and subcarrier. Edit: I found it is also documented here in Lou Dorren's papers: https://www.quadraphonicquad.com/forums/index.php?attachments/cd-4paper4-pdf.28570/

This goes without saying, but some material released on CD-4 has since been reissued (presumably from the original master tapes) onto digital formats. I found that test piece I used (Love Theme From The Godfather - from the Panasonic CD-4 Demo album) is on Vocalion's 2016 Hugo Montenegro SACD in both 2.0 and 4.0. Could certainly be useful for the purposes of reverse-engineering the noise reduction, calibration levels and such. (For example, one could derive the difference information from the digital 4.0 files and compare with the difference carriers from the record, etc..)

 

[Brice]

Hi Mattdp,

Thanks for the information on cavitation cleaning on CD-4 discs. I'll update our cleaning page to reflect your experiences.

Rest assured that the ANRS is implemented in our decoder. Quite honestly, a decent CD-4 decode requires that it is because the "raw" demodulated signals from the sub-carriers are so noisy and compressed that ANRS is necessary for an acceptably decent decode. I have submitted an Engineering Report on our decoder to the AES. I hope that this will be published soon in the AES Journal. That will include a full description of the decoder.

In fact, the Automatic Noise Reduction System (ANRS) proved the most complicated and difficult hardware system to transmute into software; mostly because of a lack of clarity in the desired performance. In fact, it became necessary to analyze and measure real hardware to comprehend what was required. Interestingly, real decoders actually do something rather different to the technical descriptions in the CD-4 literature. The ANRS system is really an LF and MF noise-gate.

We incorporate both modern correlation-based noise reduction techniques and frequency-based noise-reduction (the ANRS) within the software decoder. We base these techniques on software emulation of four hardware decoder references; the first is based on the Marantz CD-400 (JVC 4DD-5) and two are based on the QSI-5022 integrated circuit decoder designed by Louis Dorren. Finally, the late JVC CD4-50 has had a great influence on the software design. The CD4-50 is a rare hardware decoder but is widely reckoned to be a decoder which behaves where others fail. We discovered that this is due to the rather different alignment of the ANRS circuits in this late implementation. We have incorporated these changes in the Stereo Lab software decoder.

Richard

 

[DanielTheGreat]

Richard, I chanced upon this thread while searching for information about CD-4 carrier levels on the record. I don't use any crApple equipment, so won't be looking at your software (no reflection on your software, of course; just your choice of hardware ). I have a hardware decoder (SH-400) that is currently malfunctioning. When I get round to trying to fix it (I'm a professional EE), I'll need test input signals. My e-workshop is too small to even temporarily install a turntable as well as the SH-400, DSO, circuit diagrams, etc, so I'm looking at designing and building a small 'CD-4 signal simulator'.

My simulator will generate fixed (different) sine wave frequencies for each of the four channels, perform the required matrix/modulation for the carrier and baseband signals, and output the composite L & R signals at CD-4 cartridge levels to go straight into the SH-400 cartridge input sockets.

I'm looking for information on where to set the carrier signal level relative to the baseband signal maximums. Lou Dorren mentioned that the carrier was recorded at -20dB relative to peak baseband, but it's not clear to me whether this is prior to the inverse-RIAA used in cutting records (if indeed that standard curve was used for CD-4, which I doubt), or the resulting 'physical' level resulting on the vinyl.

From your tests with 'raw' CD-4 cartridge outputs, are you able to offer any guidance on how I should set the carrier level wrt the baseband maximum? Any advice would be much appreciated.

 

[Brice]

Dear DanielThe Great,

I'm impressed by your ambition to make a CD-4 signal generator. Being (originally) a hardware design engineer myself, I considered doing this. It isn't easy. The baseband is easy enough. But you need two FM modulators which (if they're PLL, which is easiest) must be locked together to generate the difference signals. This is somewhat tricky. In the end, I used software to generate our CD-4 test files. These, of course, could have been fed to hardware via a sound card.

I believe the answer you request concerning carrier level is covered in footnote. 6 of this page:
http://pspatialaudio.com/JVC_CD-4.htm
To quote the end of the footnote:
Each, unmodulated carrier is recorded at a velocity of 3.54cm/s and thus appears (because it appears before the baseband moduation) as a signal at exactly reference level when recording a non-equalised needle-drop.

Let me know if that answers your question. When you do get your generator going, perhaps you'd share the design. I'm sure lots of folks here would find it a useful box to have.

Best,

Richard

 

[Brice]

It might help in your hardware design to know of two references I have to the original CD-4 modulator hardware. The first is:
The CD-4 Mark-II Modulation System by YASUMORI KOKUBUN AND SADAHIKO MURAMOTO,JAES. JULY/AUGUST 1974, VOLUME22, NUMBER6.

They used a (very involved!) serrasoid FM modulator.

The second is: THE RCA QUADULATOR by G. A. BOGANTZ AND J. F. WELLS. JAES. MARCH1977, VOLUME 25, NUMBER 3.

This design uses locked PLLs.

 

[kfbkfb]

https://www.americanradiohistory.com/ARCHIVE-RCA/RCA-Engineer/1976-10-11.pdf
An overview of the RCA Quadulator CD-4 encoder, there might be some useful tidbits
of information to help fix a hardware demodulator.

Kirk Bayne

 

[DanielTheGreat]

Richard and Kirk,

Thank you both for your technical references, which are both interesting. I was initially puzzled by Richard's mention of: "must be locked together" for the FM modulators. "How can this be possible," I thought, given that they are each being modulated by different signals and thus will have different instantaneous frequencies at various times! But Kirk's reference cleared this up by stating that: "The purpose in synchronizing the two oscillators at all is to eliminate the annoying heterodyning which is heard if the two carriers are not matched in the absence of modulation" (my emphasis).

And the recording amplitude numbers from Richard's reference suggest the carrier amplitude is in the same ballpark as the baseband amplitude (actually velocity, but for magnetic cartridges this becomes ~amplitude). So if I just make my carrier and baseband signals about the same peak amplitude, that should be close enough for any self-respecting decoder?

When I first thought of doing this, I too (like Richard) thought in analogue terms, using a handful of XR-2206 function generators to generate the four different channel sinusoids, plus two more for the FM modulators. And of course, op-amps for the sum and difference matrix circuits. But then I thought: "This is getting a bit messy."

I happen to have on hand an MCU (microcontroller unit), which can run at up to 54MHz, and has internally:
* A touch controller, for a possible user interface;
* Multiple 16-bit timers (at least eight), which can be used to time signal generation and outputs;
* Two 12-bit DACs, for outputting the composite L and R signals;
* DSP hardware, perhaps useful for generating sine waves algorithmically [as opposed to using a look-up table (LUT), the usual fast approach. Will need to do experiments to determine if algorithmic sine generation is fast enough with this hardware].

There will most likely be some analogue low-pass filtering on the DAC outputs, but still this ends up being hugely simpler hardware than the bucket of chips and passives used in any fully-analogue solution.

It's not a high priority project for me, and will take some (spare) time, but I'll certainly post information here when I've made some progress.

Daniel