Peufeu's Electronic Stuff Unfinished projects since 1912

The case of the CD723 is a bit special. Like all transports, it adds a some jitter to the signal, but it also corrupts the bits. However today we will look at the jitter part of the problem.

Take a low-end CD player, like a CD723. It has a SPDIF output, which will send a SPDIF signal even if no disc is playing. Therefore, we can make an interesting experiment :

• Watch the SPDIF output while a disc is rotating
• Watch the SPDIF output while the CD player is idle

This CD player is clocked from the low jitter clock in my DAC, through a cable. Therefore, by comparing the input clock signal and the output SPDIF, we can learn about the jitter added by this particular player.

It is not easy to display a proper SPDIF signal because you'd have to sync on the beginning of a frame, and then it's difficult to zoom in.

So, I chose to sync the scope on the Bit Clock recovered by the CS8412 chip inside the DAC, and display the Master Clock signal received by the player. It gives the same amount of visual information as syncing the scope on the Master Clock and displaying the Bit Clock, but it's easier to zoom in.

So, what we will see is the jitter at the output of the CS8412, caused by the CD player + receiver chip + cabling.

Scope settings : vertical, 0.5 V/div. Horizontal, shown on images. I then took a photo of the screen, in the dark, with a 8 second exposure.

When the CDP is idle, it generates SPDIF containing digital silence :

We see a neat edge. It is duplicated because of the asymetrical nature of SPDIF, which combines with the lowpass characteristics of the cable to generate a bit of jitter, even when idle.

Now, when the disc is rotating, we see about 1.5-2 nanoseconds of jitter (yes this is huge) :

The disc contains a recording of silence, so this is not data related jitter.
Tapping the CD player with the hand or seeking between tracks makes the curves dance on the screen.

The causes are simple.

In the CD player, there is a spindle motor, another motor to move the lens along as the CD is read, and a tracking servo to align the lens and keep it on-track as the CD rotates. This tracking servo is constructed with moving coils, driven by the electronics in a feedback loop to keep the lens well positioned.

This is inevitable, but it is particularly nasty, as vibrations of the case (caused by music playing) will make the CD and pickup vibrate, and this will have to be compensated by the tracking servo. The player must move its lens to compensate the movement of the disc, which means the currents in the moving coils will vary according to the vibrations.

All these coils and motors create stray electromagnetic fields which are picked up by the CDP electronics. They are probably driven by PWM which causes even more interference.

Also, the current is, or course, drawn from the power supply, which is severely underpowered. It causes up to 0.5 volts of ripple on the unregulated +12V opamp supply rail, and a few tens of mv of ripple on the 7805-regulated 5V power supply which feeds the digital circuits generating the SPDIF signal. It also modulates the switching noise from the rectifier diodes, which will contaminate other gear plugged nearby.

Essentially, the powersupply of this player is about as clean as the unregulated HV rails from a class-AB amplifier operating near full power (the analogy is relevant, because the current draw is a distorted version of the box vibrations, which means, the music currently playing).

Update : videos

The player is now dead, ie. it can't play discs, but it can still try !
So, I made a video with the scope at 5 ns/div synced to the DAC master clock (steady trace) and showing the CDP SPDIF output (the other trace). The master clock moves a bit on screen too, because the horizontal centering pot of my scope is a bit old. This video shows what happens to the SPDIF when the tray is closed and the servo moves to try to read the disk. Nice power supply modulation eh ?

Download the video.

Conclusion

So, a CD player should at least have the following :

• separate power supplies (including transformers) for :
  • the analog parts (this includes the clock and DAC)
  • the digital parts
  • and a dedicated powersupply for the motors, servos, displays, etc.
• Adequate internal shielding
• Layout and circuit design insuring that the stray EM fields from the motors and servos don't disturb the sensitive parts (clock, DAC)

For instance, my now dead Marantz CD63SE had a internal brace to counter vibrations, a purpose for which proper power supply design would have been much more efficient.