Проблемы с Tuning

Начат dmmos


Подскажите, в чем может быть проблема.
Задаю в CES matching для нескольких цепей (электрических). Цепи по структуре идентичны. Развожу их. Далее выделяю одну из цепей (логических) - жму иконку tune - ничего не происходит. Вроде, делаю все как в видео. Почему не работает автоподстройка длины?

Пользователь удален

Заметка из БД ментора: The following is a checklist for the more common causes of tuning failure or poor tuning results: 1 - ALL nets involved in a tuning "equation" (that is, both nets of a diff pair, all nets of a match set, all nets interrelated by Delay Formulas equations you’ve set up, …) MUST be 100% routed before ANY of them will tune. Please check that the nets you are trying to tune, plus any nets to which or to equations involving which you are trying to tune them, are fully routed. 2 - Fixed or locked trace segments cannot be modified. Therefore, to ensure tunability, please ensure as far as is feasible that all segments of all nets to be tuned are unfixed and unlocked. 3 - Tolerances. In the Net Properties dialog (or the equivalent area in CES), there are tolerance values for diff pair tuning, match sets, and delay formulas. If these tolerances are so tight that the tuning algorithm cannot adhere to them, it will often abandon the attempt. You need to ensure that these tolerances are realistic, taking into account the other restrictions (below) - in other words, for example, if all number of tuning patterns are allowed and there’s stacks of board space available around the traces of the net to be tuned, a fine tolerance will probably still work… While experimenting with getting tuning to work, and seeking to eliminate any other possible reason preventing successful tuning, I would suggest slackening these tolerances considerably, even if only to tighten them again once you’ve established that tuning will work on the design. And please also check that the type of units in the tolerances corresponds to that in the values - i.e. time with time, or length with length. 4 - Editor Control Tuning Tab. As above, if the restrictions are so tight that the tuning algorithm cannot tune within the specified tolerances, it will likely not tune at all. Again, for experimentation I would initially allow all patterns, and reduce minimum values to 0 and maximum values to an inproportionately large value. These can be tightened to reasonable design requirements once it is established that the tuning algorithm is working successfully (please note use of Online DRC for verifying tuning results), although again, one must be realistic in the restrictions one imposes, they must of course satisfy board design requirements, but must also satisfy feasibility for the algorithm! 5 - Space is required to tune! Ensure as far as possible that there is plenty of routable free space on the board in which to tune the appropriate signals. NB the tuning algorithm does NOT reroute, change layers, et al. for the purposes of completing the most effective tuning result possible. The design must prepare a realistic layout for tuning. Very often, bringing a signal out into open space will make all the difference, and enable the tuning algoithm to complete an effective tuning result. Effective use of space AND LAYERS by the designer is critical as an input to the tuning. If all else fails, try experimenting by bringing the signals to be tuned out into open board space, and retry. 6 - This relates also to #4 above. It is usually adviseable to route and tune the most critical signals first, and work down from there towards the least critical nets. This will ensure (see #4 above) maximum space and therefore maximum tuning capability and best results for the most critical nets. 7 - Are one or more of the nets in question DIFF PAIR nets? If ANY diff pair nets are involved, then ALL Online DRC Diff Pair violations for them MUST be cleared before the tuning algorithm with ATTEMPT to tune. 8 - If all else fails, double-check the Net Properties settings to ensure that the tuning parameters are correctly defined, there are no typos, and in particular that the syntax of t.he Delay Formulas definitions, if applicable, if correct throughout. If needs be, start with a simple test involving one or two cells in the table, and work up from there.

Эта тема закрыта модератором.