• 状态 TI Thinks Resolved
  • 回复 2 回复
  • 答案 1 答案
  • 订户 16 订户
  • 视图 58 视图
  • 用户 0 会员在此处
选项
选项
相关

AM2432: The question about the long processing time of EtherCAT mailbox data frames

yijin zhao
Prodigy 50 points
Part Number: AM2432


hello,

The SDK version is 09.02.00.15 and the example is ethercat_slave_beckhoff_ssc_demo.

When testing email communication, we found that the interval between the frame of the email request sent by the master and the frame returned by the slave after processing was about 2ms. As shown in the following picture, the time it takes for cnt to change from 0 to 1.

Why does the processing time take so long?

We tested other chips and found that the processing time was probably within 200us.

b8badb3d8e4e476e81bf59ee6992396e.png

  • 0
    TI__Mastermind 31235 points

    Hello, we have received your case and the investigation will take some time. Thank you for your patience.

  • 0 回复 Eirwen
    TI__Mastermind 41810 points

    Are you seeing this long time only for Mailbox packets?

    Can you share the topology of the test environment? Is there only 1 device in the network?

    Which board are you using? Is it a custom board?

    Can you share the complete ICSS Memory dump:

    • 0x30000000 to 0x30040000 (if you're using ICSSG0)
    • 0x30080000 to 0x300C0000 (if you're using ICSSG1)

    Can you also test with the latest firmware available in INDUSTRIAL-COMMUNICATIONS-SDK-AM243X 2025.00.00.08. Steps to migrate to the new EtherCAT firmware without affecting your development SDK can be found here: EtherCAT Firmware Migration guide

    Adding to this can you share the topology of the test and what's the timestamp unit of wireshark time base and how they are capturing this - via TAP (which one)? Can they share wireshark logs. Packets they have marked in green is from PLC. Response from DUT arrived in < 2us if time base unit is in nano seconds which I believe is the case here. Since EtherCAT is on the fly protocol, it will never take 2ms for returning a frame !