Part Number: AM3358
Other Parts Discussed in Thread: AM5728
Tool/software: Linux
Hi,
I'm working for time synchronization between devices using PTP.
AM3358 / AM5728 has CPTS, which timestamps the egress / ingress PTP packets using CPTS_REF_CLK to be able to make adjustment for synchronization using hardware based value.
linuxptp project supports TI CPTS, so ptp4l, which is the part of linuxptp project, works with CPTS.
When ptp4l is executed on BBB (AM3358), the CPTS counter value associated with the egress PTP packet timing, is sometimes about 20 usec later than the actual timing.
So, the PTP synchronization is unstable with the BBB included system, especially BBB is the PTP time master.
When I execute the same test using BeagleBoard x15 (AM5728). The result is stable.
I don't observe the problem of the ingress PTP packets, on both BBB and BeagleBoard x15.
I want to know why this difference occurs.
Is this caused by the version difference of CPTS module ? or caused by the peripheral devices, especially Ether PHY ?
Considering Ether PHY, BeagleBoard x15 uses KSZ9031RNX, BBB uses LAN8710A.
Are there any possibilities that has the CPTS counter timestamping problem for PTP egress packets ?
If yes, are there any workaround ?
Regards,
Tommy
