
1. Re: The Statistical contours and Bathtub curves do not match in FastEye Channel Analyzer
weston_beal Jun 17, 2011 10:20 AM (in response to weiming.qian)1 of 1 people found this helpfulWeiming,
You are interpretting the two graphs correctly. I can see that the curves do not correlate, so there is a problem in the analysis. It appears that the data is correct for the case of 0V, but then it deviates for other voltage values. I have not seen this problem before. Can open a service request on supportnet.mentor.com and send the test case for our investigation?
Regards,
Weston

2. Re: The Statistical contours and Bathtub curves do not match in FastEye Channel Analyzer
weston_beal Jun 17, 2011 11:56 AM (in response to weiming.qian)1 of 1 people found this helpfulWeiming,
I looked at the curves a bit more and think they are consistent, but with a factor of 2 on the voltage value. If you take the voltage value from a curve in the bathtub curves and multiply that by 2 then the V*2 value correlates with the eye contours.
What version of HyperLynx did you use to generate these plots. We'll need to check that this problem doesn't manifest itself in the latest release.
Regards,
Weston

3. Re: The Statistical contours and Bathtub curves do not match in FastEye Channel Analyzer
weiming.qian Jun 20, 2011 1:37 AM (in response to weston_beal)I am using the following version to generate those two plots:
HyperLynx v8.1.1
[v8.1 build 509 (509, 0)]
Oct 5 2010
Operating system in use
Windows 7 v6.1, Build: 7600,
Compatible with Windows 7, Vista, and XP

4. Re: The Statistical contours and Bathtub curves do not match in FastEye Channel Analyzer
weston_beal Jun 20, 2011 10:46 AM (in response to weiming.qian)This problem with the factor of 2 in the voltage will be fixed in the next release of HyperLynx.

5. Re: The Statistical contours and Bathtub curves do not match in FastEye Channel Analyzer
weiming.qian Jun 21, 2011 1:39 AM (in response to weston_beal)Thanks very much for confirm this problem. I also tried IBISAMI simulation flow on Xilinx Virtex6 GTX IBISAMI model, there is the same problem there between statistical contours and bathtub curves.
So for the moment, I should trust the statistical contours and ignore bathtub curves, right?

6. Re: The Statistical contours and Bathtub curves do not match in FastEye Channel Analyzer
weston_beal Jun 21, 2011 1:54 PM (in response to weiming.qian)The bathtub curves are still useful. You just have to adjust the voltage so that it correlates with the eye contours. Start with the center voltage in the eye diagram. In clean differential signals this is 0V. Then calculate the difference between the voltage labeled on a BER curve and the center voltage. Double that voltage and it should correlate with the voltage in the eye contour plot.