PDO mapping with SOEM 2.0.0

[liviaf99]

Hi everyone,
i am new with EtherCAT and SOEM library for CiA402 servo motor control. I am actually testing the new version (2.0.0) of SOEM which includes ENI parsing for the generation of the initial SDO for PDO configuration, SM, FMMU ecc.
I am currently using the Master against an EtherCAT Simulator which takes as input the EXI file of the network (EXI contains ENI information + EEPROM and CoE Object Dictionary of the Slave). Everything works properly with the IgH Master but with SOEM I am currently encountering the following problem:
The PDO mapping fails only when the ProcessImage I set in the ENI file (and given in input to the Master SOEM) does not match with the PDO mapping specified in the EEPROM of the Slave. On the contrary, if I set in the ENI a ProcessImage equivalent to the one I find in the EEPROM, everything works. Shouldn't the SOEM Master allow PDO remapping? The PDOs specified in the ESI file of the Slaves are read-write.

Any suggestions on which could be the issue?
Thank you

[ArthurKetels]

SOEM is a library with many functions. The configuration and mapping functions are one one way of doing it. If you have other requirements then formulate other functions to do the job.

SOEM handles dynamic PDO mapping just fine. But you have to realise the configuration logic. It first reads the slave ID via the SII (eeprom), then reads the capabilities from SII to see if CoE or SoE are supported. If they are, then the mapping follows the structure of CoE/SoE, if not then it falls back to the mapping in SII. The same is true if CoE/SoE mapping fails, then the fall back is SII.

So there must be some inconsistencies in your slave simulation. Make a Wireshark trace (.pcap file) of the discovery and mapping sequence and post it here. Then we can have a look what exactly happens.

[liviaf99]

WireShark.zip
ENI.xml

Thank you for the answer. I have attached a ZIP file containing two Wireshark traces.
Attached files: WireShark.zip (contains WireShark_SOEM_1.pcap and WireShark_SOEM_2.pcap). I have also attached the ENI file with which I build the eni_test application.

I am testing the Master (SOEM 2.0.0) against an EtherCAT Simulator. In both cases, the ENI file provided to the Master defines RxPDO (0x1600) and TxPDO (0x1A00) with 3 entries each.

Case 1: WireShark_SOEM_1.pcap (Mapping FAILS)
Slave EEPROM state: Contains default ESI mapping (RxPDO 0x1600 with 8 entries; TxPDO 0x1A00 with 7 entries).
The PDO remapping to the 3 entries defined in the ENI fails, the slaves remains in PREOP.

Case 2: WireShark_SOEM_2.pcap (Mapping SUCCESS)
Slave EEPROM state: Manually modified so that the internal mapping matches exactly the 3 entries defined in the ENI file.
The configuration completes successfully and slaves enter in OP.

The Slaves both support CoE and EoE (as specified both in the ENI file and in the EEPROM hex configuration). Despite the InitCmds are the same for both cases, in the first scenario the Write CoE command seem to fail and the slaves never enter in SAFEOP.

Thank you for your support!

[ArthurKetels]

A number of weird behaviours. I guess you do not run a vanilla SOEM example code.

Normally SOEM assigns slaves addresses starting from 0x1001, in your case it starts from 0x0001, and then later it switches to 0x1001. Recovery code then tries to find the 0x0001 slave again (without success, because it is reassigned 0x1001).

The main problem however is that there is no mapping defined in the slave. SDO 0x1a00:01 and 0x1600:01 return 0x00000000. Therefore the CoE mapping fails and then SOEM falls back to SII (eeprom) to find the PDO configuration.

I think you wanted to test the ENI parser in SOEM. But as of today there is no functioning ENI parser in SOEM. eni_test.c is just a mock-up with no real functionality.