PEAK-System Forum

Understood.

Probably last question for now: But w/the GW FD DR feature mentioned earlier, the 3-in-1 approach should be feasible?

Thanks!

Also here - it is depending on how fast do the CAN Frames arrive in a row. You could have luck that the next IRQ Trigger include all Frames of you Burst - but it could not be guaranty.

Hi,

we got hold on FD DR gateways and could experiment with irq_time_limit and irq_count_limit.

When does the relevant timer start?

Assume, the irq triggers and a UDP message is send. Thus, we're having a 'clean' state.
Does the timer start already now, or only after the 'first' CAN frame arrives?

I would expect the timer to start with the first frame, but have some indication that it might already start with the last irq.

Thanks
Thomas

Hello,

When the CAN FD controller receives a 1st frame, then it starts the "irq_time_limit" timer so that it will notify the driver at last "irq_time_limit" x 100 µs later, if no other CAN frame has been read from the CAN bus.

Regards,

Thanks!

From the developer documentation I would assume irq_time_limit is in 1ms instead of 100us.

Would ypu mind double checking if its 1000 or 100 us?

Thanks again
Thomas

Hello,

Thank you for sharing this with us. It is certainly a typo in the documentation, because the driver itself manages values in x100 µs.

Nevertheless, we will check the chain of software components involved in taking this parameter into account to confirm it here very soon.

Thank you again and regards,

Hi Thomas,

As said, we just have checked the flow of irq_time_limit parameter and we confirm that the unit is x100 µs instead of ms. Therefore, the documentation will be fixed accordingly.

Thank you again for your help,

Best regards.

Thanks!
Thus, the valid range for this parameter is 0..150 translatingt into 0..15 ms?

Hi Thomas,

The value range still is [1..15] therefore the maximum time the IP core will wait before triggering an IRQ is 1,5 ms.

Regards,

Hi Stephan,

thanks!
It seems I would need a higher wait time than 1,5 ms.
Can we discuss a way to extend this period?

Regards
Thomas