Отправлено Михаил Евстафьев 26 декабря 2001 г. 19:36 В ответ на: Да, безглючен, в отличии от MCP2510 (+) отправлено Щ.С. 26 декабря 2001 г. 14:11

И это не единственный случай, так даже в сам стандарт CAN от BOSH попали ограничения связанные с обходом ошибок в первом чипе CAN контроллера Intel82526.
------------------------

Anomaly of the SJA1000
Stand-alone can controller

During enhanced system evaluation of the SjA1000 an anomaly has been found
concerning Acceptance Filter

When does the anomaly occur and what happens?

The anomaly occurs only within transmitting nodes which loose arbitration after
the 7th identifier bit.
It occurs in BasicCAN Mode as well as in PELICAN Mode only if the acceptance
filter is actively used (Acceptance Mask not equal toFFh)
In case of the above defined arbitration lost even the acceptance filter
algorithm may be influenced in that way, that the winning message
- is not received, even if it should pass the acceptance filter or
- is received, even if it should not pass the acceptance filter
So a message may be missed within the receive buffer or an additional not
expected message may be stored within the receive buffer.

What are possible work-arounds ?
the following software work-arounds can be considered :
- do not use the Acceptance filter at all, -> set the acceptance Mask
register(s) to don't care (value=FFh)
- select identifiers in that way, that arbitration is finished within the first
7 bits of the idedentifier area
- prevent arbitration events by software control (master- salve concept)