9.0.1 matched lengths , differentials, phase tolerance

Discussion created by lalexman on Aug 6, 2009
Latest reply on Jul 9, 2013 by TBD

Hi Andy,


I am evaluating version 9.0.1 and have a few questions.


I think that the new area for accordions and differential pairs might be partially from some information I provided in SR's back in 2005. In that information I provided a sample design. It looks like some of my ideas we implemented but not all of them. I do realize how hard it must be to implement what everyone wants and not just one person. Anyways perhaps you could response to a few issues I still see with my test design.



1. The autorouter cannot route differential pairs thru a connector that has staggered pins irregardless of the settings for obstacles. This causes the lengths of the differential pairs to be much longer then necessary.Are there any plans of fixing this ?


2. In order to route lots of differential pairs or single ended signals on dense boards and not have many extra layers its best(IMHO) to not have lots accordions but to have one accordion that is parallel to the trace and not perpendecular to the trace. The reason for this is perpendecular accordions use up routing areas for additional traces. The test board I provided shows how parallel accordions allow more traces to routed in a tight area. Is there a reason this was not implemented in the new software ?


3. When routing differential pairs there is is a need to match the lengths of a class of pairs and also a need to match the length of each pairs traces to each other. The matched length of the class is usually different then the matched length of each pairs traces to each other . For example a class might be matched length to 1000 mils but the matched length of each pairs traces might be 25 mils. The software currently can route the pairs traces to each other at the same value of the classes matched length or to a tolerance of zero. While zero meets the spec we do not need it match that close since it causes extra accordions that are not necessary or desired.


Thanks for your time .


Leonard ALexman