Working CD-4 (software) Demodulator!

[Brice]

It may be of help to read the following links regarding making CD-4 needle-drops:

General advice: http://pspatialaudio.com/needledrop.htm
CD-4 specific: http://pspatialaudio.com/JVC_CD-4.htm#levels

Best,

Richard

 

[winopener]

Very useful, thanks.

 

[kfbkfb]

I have no experience with needle drops,
what sort of amplifiers are those of you
doing successful needle drops using?

(I was thinking of building myself a battery
powered OPAMP based amplifier for
needle drops)

Kirk Bayne

 

[Brice]

Building your own battery powered n'drop preamp is certainly possible. Remember, all that is needed is some gain: it's not like it needs precision equalisation. RIAA EQ on the baseband signal is all accomplished in the software.

I have used something as simple as the attached little circuit to very good effect. It's built around a dual op-amp which could be something as modest as a NE5532. I prefer the recommended OP284EPZ. It's a lot more expensive (10 Euros), but it's rail to rail which gives better headroom and has a low quiescent current which prolongs battery life.

The gain is 33dB which is more than adequate to drive sound cards from the majority of MM cartridges. Note that the value of R1 can be changed. I know that some older CD-4 cartridges prefer loads >47k. This can go as high as 470k to give an input R of 100k.

Best wishes,

Richard

 

[Brice]

All,

I am delighted to inform you that the first version of our CD-4 demodulator software is now available in Version 3.1.8 of STEREO LAB at,

http://pspatialaudio.com/

If you request a free demo version of the software, you will be sent a link to download the installer. The demo version includes the full functionality of the CD-4 decode just with a limit on the number and duration of the conversions. A reminder that Stereo Lab only runs on Mac OS X.

This first release of the software CD-4 demodulator is a very general decode intended for CD-4 material in a wide range of physical condition. In fact, it's based on my own rather haphazard library of CD-4 records, all of unknown provenance!

Maximum channel separation has not been a specific aim of this development, rather that all the records used for development should decode to something acceptable and - above all - enjoyable. By decoding to an intermediate first-order Ambisonics format, the final signals are designed to drive a modern 5.1 loudspeaker system (Center and LFE are silent).

Source needle-drops need to be non RIAA equalised: both decoding and RIAA equalisation are all performed in the digital domain.

Where the "rock meets the record", CD-4 remains a physical medium, so the usual hardware requirements of Shibata (or similar) stylus and a cartridge with adequate bandwidth and low crosstalk are just as important when using the software decoder as they are with a conventional setup.

I have added a decode mode ("Rears as Subcarrier") in which the rear channel audio is replaced by the recovered, filtered subcarriers from the record. This is intended only for engineering purposes. For example, it could be used to establish the quality of (or lack of damage to) the record or to check for up-talk from the baseband. It could aslo be used to measure cartridge crosstalk using the method developed by Shure (Carrier Crosstalk Considerations in the CD-4 System. Anderson, C.R. JAES July/August 1977).

Best,

Richard Brice

 

[Jerfo]

I am delighted to inform you that the first version of our CD-4 demodulator software is now available in Version 3.1.8 of STEREO LAB at,

http://pspatialaudio.com/

This is great news!! I can't wait to give it a spin! I've collected a handful of CD-4 LPs just for this occasion.

A reminder that Stereo Lab only runs on Mac OS X.

D'oh! Is there any chance of a Windows release in the future? In the meantime, maybe I'll attempt to spin up a OSX VM and see if it'll function in that environment.

 

[Brice]

Hi Jerfo,

A Windows release isn't planned. My apologies. We made the decision to develop Stereo Lab solely on the OS X platform several years ago.

Richard

 

[Jerfo]

Hi Jerfo,

A Windows release isn't planned. My apologies. We made the decision to develop Stereo Lab solely on the OS X platform several years ago.

Richard

That's a bummer, but understandable. I imagine that you guys aren't exactly a huge shop and have to be selective about what you can and can't support.

 

[Brice]

I imagine that you guys aren't exactly a huge shop and have to be selective about what you can and can't support.

Exactly.

Richard

 

[fredblue]

All,

I am delighted to inform you that the first version of our CD-4 demodulator software is now available in Version 3.1.8 of STEREO LAB at,

http://pspatialaudio.com/

If you request a free demo version of the software, you will be sent a link to download the installer. The demo version includes the full functionality of the CD-4 decode just with a limit on the number and duration of the conversions. A reminder that Stereo Lab only runs on Mac OS X.

This first release of the software CD-4 demodulator is a very general decode intended for CD-4 material in a wide range of physical condition. In fact, it's based on my own rather haphazard library of CD-4 records, all of unknown provenance!

Maximum channel separation has not been a specific aim of this development, rather that all the records used for development should decode to something acceptable and - above all - enjoyable. By decoding to an intermediate first-order Ambisonics format, the final signals are designed to drive a modern 5.1 loudspeaker system (Center and LFE are silent).

Source needle-drops need to be non RIAA equalised: both decoding and RIAA equalisation are all performed in the digital domain.

Where the "rock meets the record", CD-4 remains a physical medium, so the usual hardware requirements of Shibata (or similar) stylus and a cartridge with adequate bandwidth and low crosstalk are just as important when using the software decoder as they are with a conventional setup.

I have added a decode mode ("Rears as Subcarrier") in which the rear channel audio is replaced by the recovered, filtered subcarriers from the record. This is intended only for engineering purposes. For example, it could be used to establish the quality of (or lack of damage to) the record or to check for up-talk from the baseband. It could aslo be used to measure cartridge crosstalk using the method developed by Shure (Carrier Crosstalk Considerations in the CD-4 System. Anderson, C.R. JAES July/August 1977).

Best,

Richard Brice

(referring to the part of your quote highlighted in bold)

while this is a superb effort and one which I applaud and support, I urge the developers to consider including a "maximum separation" setting in their software, not least because the CD-4 system was by definition the only commercially widely available discrete 4-channel vinyl retrieval method and it would be a shame to only have the ability to demodulate a CD-4 record in software to lesser separation levels than the system is indeed capable of. anyhoo, not a grumble just a suggestion, all just imho and good luck with it all, super commendable!

 

[Brice]

Hi Fredblue,

Thanks for your support of the idea. I'm happy to say that we are already working on a new demodulator algorithm. The FM demodulator in the first release is bullet-proof: it won't unlock (even if the speed is 10% wrong or if there's wow and flutter); it won't "blast" if the carrier dissappears; and it re-acquires the FM signal in a few cycles of it re-appearing. But, it pays for these robust qualities with certain sacrifices in fidelity.

The new demodulators are much more "fussy". They give better results on good material .... and much worse results on poor-quality material!

I wanted to develop the "safe" decoder first. So that the high-quality option is just that... an option.

Best wishes,

Richard

 

[fredblue]

Hi Fredblue,

Thanks for your support of the idea. I'm happy to say that we are already working on a new demodulator algorithm. The FM demodulator in the first release is bullet-proof: it won't unlock (even if the speed is 10% wrong or if there's wow and flutter); it won't "blast" if the carrier dissappears; and it re-acquires the FM signal in a few cycles of it re-appearing. But, it pays for these robust qualities with certain sacrifices in fidelity.

The new demodulators are much more "fussy". They give better results on good material .... and much worse results on poor-quality material!

I wanted to develop the "safe" decoder first. So that the high-quality option is just that... an option.

Best wishes,

Richard

excellent news!
thank you for the info, Richard.
i look forward to playing around with your "unsafe" demogitator in the fullness of time!

 

[Soundfield]

"unsafe" demogitator

I've been trying to rid my garden of moggies for years - this sounds like a particularly effective device (I'm thinking laser guided cattle prod perhaps?) !

 

[kfbkfb]

So...

Has anyone tried this (all software) CD-4 decoder?

If so, what do you think of it?


Stylus Care Information:
http://www.pspatialaudio.com/stylus_care.htm

Kirk Bayne

 

[Jerfo]

Funny you should mention it...

I attempted to give it a try this morning, but it didn't work quite properly...through no fault of the software. I'll explain:

Since I don't have a Mac, I installed a macOS High Sierra virtual machine within Virtual Box. From there, I acquired the Stereo Lab demo software and it installed in the VM. It seemed to have no problem installing and running in a virtual environment.

I'd already done a rip of the Brecker Brothers' "Back to Back" CD-4 record a few months ago, so my intent was to use that rip as my test bed. However, I ran into two problems:

1. The rip was done at 192/24. Stereo Lab doesn't support this sample rate. I used Reaper to downsample it to 96/24, so no big deal.
2. The rip was not a flat, RIAA curve-less rip. This was much more of a problem, as Stereo Lab applies the RIAA curve itself. This yielded a muddy, incorrect result.

When I have a few minutes, I'm going to do a re-rip of the first track on that album with no RIAA EQ applied. I'll let you know how it goes.

As far as the software itself, it took me all of thirty seconds to figure out how to get it to do what I wanted to do. So, no complaints there at all.

 

[Brice]

Hi Jerfo,

Not having a non-RIAA rip is a problem. However, there is a work-around in the Stereo Lab software. One of the equalisation options is inverse-RIAA. If you pre-equalise the rip using this feature and then subsequently process for the CD-4 decode, at least you won't get the horrible bass heavy baseband decode.

This isn't an ideal work-around because the FM carriers are badly attenuated by the hardware RIAA and boosting their level doesn't guarantee their accuracy and freedom from noise. But, given that you have a 24 bit rip to start with, lifting the carriers by 30 to 40dB should stll be acceptable.

Stereo Lab will issue a warning on the second process (the CD-4 decode) saying that "are you sure, because this is a cascade process". Just ignore that.

Best,

Richard

 

[Jerfo]

Not having a non-RIAA rip is a problem. However, there is a work-around in the Stereo Lab software. One of the equalisation options is inverse-RIAA. If you pre-equalise the rip using this feature and then subsequently process for the CD-4 decode, at least you won't get the horrible bass heavy baseband decode.

This isn't an ideal work-around because the FM carriers are badly attenuated by the hardware RIAA and boosting their level doesn't guarantee their accuracy and freedom from noise. But, given that you have a 24 bit rip to start with, lifting the carriers by 30 to 40dB should stll be acceptable.

Stereo Lab will issue a warning on the second process (the CD-4 decode) saying that "are you sure, because this is a cascade process". Just ignore that.

Cool! This worked as a proof of concept. It decoded and sounded OK, although I have no means of determining how accurate the decode was. The right rear speaker seemed like it was mostly cowbell, which was:

A. Funny.
B. An indicator that some sort of discrete speaker mapping was happening here.

The anti-RIAA process took my source down to 2 minutes, and the CD-4 then took it down to 1 minute, so the sample size was very limited. I'll try to re-rip that first track today without RIAA and let you all know how it goes. That'll at least give me two minutes worth of material to demo.

 

[Jerfo]

OK, I've had a chance to do some testing with this. I have my AT LP120 (internal preamp removed) with Ortofon 2M Blue stylus / cartridge feeding directly into a Behringer UMC202HD audio interface. (I have some nicer audio interfaces in my music studio, but this will suffice for now.) I realize that my stylus isn't a shibata or microline stylus, so it's not optimal for reading the CD-4 carrier information. (It's there, though. Keep reading.)

I tried clips from three CD-4 LPs: The Brecker Brothers' "Back To Back," Seals and Crofts' "Diamond Girl," and Floyd Cramer's "In Concert." In each case, Stereo Lab gave me the "No RADAR detected" error, but it did deliver a result:

The end results all sounded just so-so, lacking in detail, with the Cramer clip sounding the best of the three. This setup was clearly not optimal from a sonic standpoint; the overall sound doesn't come close to comparing to what this cartridge sounds like through my Pro-ject Phono Box S pre-amp. From a strictly CD-4 / quad standpoint, there's an audience clap-along section in the first Cramer track, and that seemed to get placed nicely into the rear speakers. That seems like it's accurate at first glance, but I decided to dig deeper...

I rendered the original Cramer stereo track into individual mono files and fed them one at a time into SDR#. (For those of you not familiar with this software, check out this link: https://thethinkingmatt.wordpress.c...g-quadradisc-cd-4-via-dsp-a-proof-of-concept/ . Also, look for a VWestlife YouTube video on it.) In this tool, I can very clearly see (and hear) the demodulated 30k signal:

The red stuff in the middle is all of the audible frequency range stuff. Once you leave the audible range, you get a whole bunch of nothing (dark blue) until you see a thin band of audio in red at 30k. There's our modulated carrier frequency. Pretty cool, right? SDR# lets you demodulate that and record it to a WAV file, which I did.

I went into Reaper, imported all of the original and demodulated files, normalized everything, applied RIAA to the master bus, and rendered four separate output files: Original Left plus Difference Left, Original Left plus Inverted Difference Left, Original Right plus Difference Right, and Original Right plus Inverted Different Right. Ignoring some of the obvious differences - I shouldn't have normalized everything because the output file was waaaaayyy too hot - the 4 channel separation seems very similar between the Stereo Lab version and my makeshift version.

I've uploaded some files up to my Google Drive so that you can compare and contrast. Here's the link: https://drive.google.com/drive/folders/1imbDsPZOQzOmyrZyYLBin01sq-uU5rX5?usp=sharing . You'll find a stereo rip of the original Floyd Cramer track ("On The Rebound") done through my Pro-ject preamp, the CD-4 conversion done through Stereo Lab, and the manual hackjob CD-4 conversion that I did using SDR# and duct tape.

I hope that folks find this helpful. This was fun!

 

[sjcorne]

I need to give this a shot...

It’d be really sweet to get maximum seperation out of all those rare Japanese CD-4 only titles

 

[kfbkfb]

Jerfo:

Please post the exact software name and version number of the Mac emulator/virtual machine
and also please the version number of Windows 10 you're using.


(this looks like my most promising option for archiving my CD-4 discs using my Windows 10 PC)


Although extended (phono cartridge) frequency response (to 45kHz) is important for CD-4 decoding,
channel separation in the 20kHz to 45kHz range needs to be at least 12dB and preferably 20dB.

Kirk Bayne