CD player's eyepattern

[Steven505]

Is there anybody who can explain how the eyepattern in a cd player is formed?
In my opinion it represents a serial datastream, however, it looks like sinewaves written throug eachother.
What does this eyepattern represent?

[Javin5]

It represents the scope picture of a datastream, triggered by the clock of that data. The transitions are rounded because of bandwith limitation, so it may look like sine wave shaped transitions (sort of). In the ideal case, the eye pattern is wide open, i.e. the transition crossings of all superimposed tracks are close together, providing a good margin for sampling the data stream with the PLL-recovered clock, even if that clock has some jitter. Yes, this an analog signal, even if it represents digital data.

[wakibaki]

It's a bitstream displayed so that the bits are overlaid.

The vertical height of the eye is the amplitude of the signal. As the eye shrinks in the vertical direction bit errors increase as the signal level is no longer reliably interpreted as '0' or '1'

If the eye shrinks in the horizontal direction bit errors increase as jitter causes the sampling to occur at the wrong time.

Both problems can be present at the same time.

w

[phofman]

IIRC, CD players do not need PLL, but have a buffer for the incoming stream from the reading head, and control level of the buffer fill by regulating rotation speed of the disk. That way the outgoing data stream should be independent of the reading jitter (though it certainly propagates through the supply lines).

[Steven505]

Thanks a lot for the information. So if I'm right, this datastream is only made visible in the form of an eyepattern. The decoder receives serial data from the read electronics? I mean, the changes from pits to lands are picked up by the photodiodes as flashes and are transmitted to the decoder as a serial stream?

[Guido Tent]

Originally posted by Steven505
Thanks a lot for the information. So if I'm right, this datastream is only made visible in the form of an eyepattern. The decoder receives serial data from the read electronics? I mean, the changes from pits to lands are picked up by the photodiodes as flashes and are transmitted to the decoder as a serial stream?

yes, it is the summation of the 4 central diodes. this signal is first sliced and then decoded

one can see and measure a lot of the drive quality looking at and measuring on this signal only

best

[wakibaki]

Yeah, obviously when you have a serial datastream you can generate an eye pattern at any point in it's travel from source to sink, so it's an analytical tool employed all over the place in digital comms for debug, but it doesn't generate numerical performance data as e.g. bit error rate.

w

[Guido Tent]

Originally posted by wakibaki
Yeah, obviously when you have a serial datastream you can generate an eye pattern at any point in it's travel from source to sink, so it's an analytical tool employed all over the place in digital comms for debug, but it doesn't generate numerical performance data as e.g. bit error rate.

w

exactly

Some decoders have a an error flag which gives a signal when an error occurs. You may connect a counter and see how many errors appear (bear in mind that many if not all can be corrected).

The more sophisticated decoders like we used in DVD recordable systems have both jitter en bit error measurements integrated, so the drive software can adapt its' laser power and strategies prior and during writing, to optimise the burning process........

best

[Cliff45]

Steven505-

Maybe this will help?

The "eye pattern" is made on the display of an oscilloscope by triggering the traces from a stable (extremely stable!) PLL reference clock, with the oscilloscope timebase set to one "unit interval" (UI) of the data stream being measured. The eye pattern itself is built up from continuous re-draws of the display by subsequent bit transitions of the data stream. Some special-purpose devices actually have a "clock recovery" circuit built-in for this use, like the "serial data analyzers" made by LeCroy, Textronix, Agilent, and others. These types of machines are used to measure high-speed serial data streams like those produced by SATA disk drives, telecom fiber optic links, and computer buses like PCIe. They all produce a "eye pattern" if you need to measure the quality of a data link between two points.

A good article on eye pattern measurements done by Mike Hertz (perfect name for a guy who's an E.E., if you asked me!) from LeCroy, Inc. is located here: IEC Newsletter, August 2007

The generated signal trace can be used to interpret the quality of the received signal stream. It reveals noise as the waveform's amplitude variations become indistinct, and jitter as the transitions shift about the time intervals of the code period. Peak shift, dc offset, and other faults also will show up in this display as well. Noise in a channel will tend to close the pattern vertically, and jitter closes it horizontally.


At what point in the chain you measure determines what the eye pattern means. "Interface" jitter is measured between two points that are transferring serial data in the digital domain, and "Sampling" jitter is measured where the data is converted from the digital domain to analog or vice-versa. A good question one might ask is, "Do I want to measure the accuracy of the mechanism that reads the storage medium, or do I want to measure the accuracy of the medium itself?" Measuring the quality of a CD disk would be done by examining the analog RF 3T-11T signal the is output from the pickup itself. The 11T signal is derived from the reading of a 11T pit/land and a 3T signal is from a 3T pit/land.

Katherine Cochrine put together an article on the subject of CD testing back in 1996, part of it is reproduced here:

CD Testing: The DIgital and Analog Sides


HTH-

[Steven505]

Thank you verry much Cliff45!!

You have been a great help in my understanding of eyepatterns.
I think I understand the basics of it now.

Thanx again!

Steven