### Author Topic: Question on eye pattern test  (Read 3776 times)

0 Members and 1 Guest are viewing this topic.

#### fubar.gr

• Supporter
• Posts: 362
• Country:
##### Question on eye pattern test
« on: March 15, 2014, 07:02:47 pm »
This is a sample eye pattern from a CD player

Comparing it to the samples found here, the one from the CD player looks like it is composed of several overlapping signals.

Is this how CD players work or is there something wrong in the sample?

#### w2aew

• Super Contributor
• Posts: 1753
• Country:
• I usTa cuDnt speL enjinere, noW I aR wuN
##### Re: Question on eye pattern test
« Reply #1 on: March 15, 2014, 08:26:13 pm »
All eye patterns consist of overlapping waveforms. The difference between the CD player waveforms you show, and the eyes in the article, is the way the scope is triggered. In the CD example, the scope is triggered on the signal itself, which is why every waveform is starting from the same rising edge. In the case of the classic XXX looking eye, is it typically triggered by a bit clock (either explicit or recovered).
FAE for Tektronix
Technical Coordinator for the ARRL Northern NJ Section

#### Nerull

• Frequent Contributor
• Posts: 694
##### Re: Question on eye pattern test
« Reply #2 on: March 15, 2014, 08:54:52 pm »
All eye patterns show overlapping signals, the appearance of the pattern depends on how the signal is encoded. Different methods produce different patterns.

CD players use something called run length limited eight-to-fourteen modulation, which specifies a minimum and maximum number of contiguous bits. There will always be a minimum of three same-value bits in a row, but never more than 11. The rising or falling parts of the plot are the transitions between different bit values.

It doesn't really look like it, but you can think of the signal as square waves. A high signal is for a high bit, a low signal for a low bit. The time the signal spends high or low is determined by the number of bits multiplied by the clock rate. The encoding limitations ensure that there are exactly nine different lengths each wave can have. When you overlay many of these waves, of varying lengths, it creates a pattern like that.

#### babysitter

• Frequent Contributor
• Posts: 857
• Country:
• pushing silicon at work
##### Re: Question on eye pattern test
« Reply #3 on: March 15, 2014, 09:16:13 pm »
I consider putting your right hand on the timebase and starting to turn it thru the range very educative.
As a starting point trigger to a clock if available or a edge, as Alan explained.
(Edge triggering is the starting point of clock recovery for the circuit itself if there is no clock line, too.
So if you are able to get a nice eye pattern when edge triggering but not when triggering from clock signal, your clock recovery section might have trouble.)

Basically, look for good separation of the upper "cloud" of hig- and the lower cloud of low-levels,
and additional levels if present and required. the more confined the bunches of related traces are, the better.
This helps the decision if it is a high or low bit.
Usually the horizontal separation of the nodes should be a fixed time or multiples thereof, giving your symbol length.
While the CD-Player is ok when the horizontal separation wobbles a bit due to the mechanical source, this would be a certain bad sign when coming from a rigid-clocked transmitter.
Might happen possibly also due to phase shifting in a "multi-path" signal channel too.
In most cases, you also want to see the transfer from high to low or back going with a certain slew rate, shown as more or less steep transitions. If this doesnt match your defined pattern, you have too little bandwidth somewhere usually.
vertical oscillation of the traces is a sign for reflexion or similar problems, too.

I'm not a feature, I'm a bug! ARC DG3HDA

#### SoundTech-LG

• Frequent Contributor
• Posts: 572
• Country:
##### Re: Question on eye pattern test
« Reply #4 on: March 17, 2014, 08:59:48 pm »
Typically an out of focus, weak in amplitude, not clearly defined eye, is indicative of poor output from the laser (or as simple as a dirty lens).  I've seen plenty of both.

Smf