IMXRT1052 MAC & PHY IEEE1588 signals

debug · ‎02-19-2023

Per section 4.1 of the reference manual, there are some pins defined for IEEE1588 function. Event 0~3 IN and OUT: ENET_1588_EVENTx_IN/OUT, it seems the IN/OUT defintions for all four are all the same.

I am looking into using IEE1588 for AES67. The IEEE1588 capable PHY does not explicitly provides these signals, the PHY has a few GPIO that can be configured (not checked yet if they can fulfill the IEE1588 signals a defined per IMXRT1052 reference guide). it looks like on one pair of IN/OUT may be needed

Is is mandatory to use those pins. what are pro/cons on (not) using those ? When using those, what needs to be done at the MCU side ?

jingpan · ‎02-20-2023

Hi @debug ,

I think it isn't mandatory. ENET_1588_EVENT in/out is for PPS.

"The most common way to analyze time synchronization comes from looking at the pulse per second (PPS) signal. Sending out a pulse at every second transition produces a PPS signal. For many older systems, the PPS signal is the only way to measure the success of time synchronization. The primary disadvantage to this measurement is that this effectively samples the error every second. Since there is
not necessarily a correlation between the second transition and the clock synch update, it is difficult to
achieve dependable results. Another issue with the PPS measurement is that the PPS signal is typically
generated from a digital output that will add additional error to the synchronization results. That additional error will only impact digital inputs and outputs, but not the synchronized clock itself and, thus, should not be included in the synchronization measurement.
The most accurate method to measure clock synchronization is set both the master and slave to generate a clock output at a known frequency and then compare those two clock signals. This provides the error at many more times a second, providing a more accurate view of the time synchronization. As an additional benefit, the clock output can be handled through an analog output that will not add additional synchronization error."

So, you can see that both RT1050/RT1170 evk don't use these pins, but they can support IEEE1588. Please refer to this application notes.

https://www.nxp.com/docs/en/nxp/application-notes/AN12149.pdf

Regards,

Jing

View solution in original post

jingpan · ‎02-20-2023

Hi @debug ,

I think it isn't mandatory. ENET_1588_EVENT in/out is for PPS.

"The most common way to analyze time synchronization comes from looking at the pulse per second (PPS) signal. Sending out a pulse at every second transition produces a PPS signal. For many older systems, the PPS signal is the only way to measure the success of time synchronization. The primary disadvantage to this measurement is that this effectively samples the error every second. Since there is
not necessarily a correlation between the second transition and the clock synch update, it is difficult to
achieve dependable results. Another issue with the PPS measurement is that the PPS signal is typically
generated from a digital output that will add additional error to the synchronization results. That additional error will only impact digital inputs and outputs, but not the synchronized clock itself and, thus, should not be included in the synchronization measurement.
The most accurate method to measure clock synchronization is set both the master and slave to generate a clock output at a known frequency and then compare those two clock signals. This provides the error at many more times a second, providing a more accurate view of the time synchronization. As an additional benefit, the clock output can be handled through an analog output that will not add additional synchronization error."

So, you can see that both RT1050/RT1170 evk don't use these pins, but they can support IEEE1588. Please refer to this application notes.

https://www.nxp.com/docs/en/nxp/application-notes/AN12149.pdf

Regards,

Jing