Question about SAS3_EYEOPENING

Mathieu Gagnon Mathieu.Gagnon at pmcs.com
Tue Feb 26 10:02:03 PST 2013


* From the T10 Reflector (t10 at t10.org), posted by:
* Mathieu Gagnon <Mathieu.Gagnon at pmcs.com>
*
Dear Justin,
The previous version of the script (up to 1.2) was computing only
worst-case eye closure.
This means eye closure that could happen only if thousands of bits had
the exact worst-case condition - i.e., at a very low BER, much below the
target of 1E-15.
These results were usually pessimistic.
The latest version computes an estimation of the closure at a BER 1E-15.
However, it still reports the worst-case closure earlier in the output.
This is why the final Opening - Xtalk is not equal to the above
"Worst-case opening" - "worst-case Xtalk".
I will add a mention of the target BER in the final report to reduce
confusion.
While verifying the code, I noticed that sometimes the worst-case
computed in v1.2 was however optimistic. I updated the code to compute
this, so in some cases the worst-case is lower. You seem to have a
pretty extreme case, going from 80.7% to 54.1%. Noise correction has
also been improved, and some type of responses could give different
results. In v1.2, noise correction was sometimes too aggressive.
You can look at the cumulative closure plots - any long tail not
flattening out is an indication that the noise correction was not strong
enough, while a curve that flattens out very quickly, together with the
pulse response (also plotted) showing some significant cursors, is an
indication that noise correction was too strong.
I've devoted significant efforts lately to improve these computations,
so hopefully the results will be more consistent.
The closure at a BER of 1E-15 is also less sensitive to these kind of
errors, as the effect of any remaining error does not influence so much
the result when their amplitude is not significant - whereas worst-case
closures integrate these errors without limit.
Regards,
Mathieu 
-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of justin
Sent: February-26-13 4:06 AM
To: t10 at t10.org
Subject: Question about SAS3_EYEOPENING
* From the T10 Reflector (t10 at t10.org), posted by:
* justin <power at uniconn.com.tw>
*
Dear sir,
I have a question about the simulated result.
For example in case 0;
The previous script got result as below:
Noiseless Eye Opening from fixed CDR = 80.7 %
Crosstalk Eye Closure = 19.3 % (unconstrained pattern)
Min Eye Opening = 61.4 %
And 
Opening-Xtalk = 61.4 %
The newest script got results as below:
Noiseless Eye Opening from fixed CDR = 54.1 %
Crosstalk Eye Closure = 16.4 % (unconstrained pattern)
Worst-Case Eye Opening = 37.7 %
And
Opening-Xtalk = 48.9 %
How the 48.9 % be calculated?
Thank you~
Justin
Uniconn Corp.
127, Jing An Road, Chung Ho City,
Taipei County, Taiwan, R.O.C.
Tel: (02)2948-1518 #26
Fax: (02)2940-5688
Web: www.uniconn.com.tw
Email: power at uniconn.com.tw
T
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org



More information about the T10 mailing list