PROCESSOR-SDK-AM437X: EtherCAT large TxPdo issue

The EtherCAT experts have been notified. They will respond here.

Hi Karl, thanks a lot for the detailed information, I will reach our EtherCAT protocol developers for guidance here, and come back to you.

thank you,
Paula

Hi Karil, we filed a JIRA for this possible bug. We will keep you informed.

thank you,
Paula

Hi Paula,

Yes, please keep me informed.

Thanks,

Karl

wireshark_trace.zipHi Paula,

Any update?

I repeated my test of sending 2K PDO successfully by using Beckhoff evaluation hardware but the TI hardware still fails. Please forward the attached wireshark_trace.zip file (Both TI and Beckhoff test wireshark traces, and Beckhoff engineer emails) to TI ethercat firmware developer. 

We choose TI chip due to its large 28KB PD RAM and we have a tight schedule to get this project done. Please help! 

Thanks,

Karl

Hi Karl, we have forecast a fix for PRU-ICSS-ETHERCAT-SLAVE_01.00.03. I will check with the developers about schedule for EtherCAT 1.0.3 release and come back to you.

thank you,

Paula

Hi Paula,

Thanks for the update.

Just a summary of the issues here for TI vs Beckhoff

1. The WC is different for large TxPdo. Please verify the fix for TxPdo as big as 6 to 8k. TI am437x EtherCAT chip has enough PD RAM to do this and this large size TxPdo is what required for our new device.

2. While testing in Free-Run mode, in each cycle, APPL_InputMapping is being called twice with TI chip vs once by Beckhoff chip for large TxPdo

Hope both issues will be fixed in the 01.00.03 release

Regards,

Karl

Hi Karl, thanks for the update. I checked schedule for 01.00.03 PRU-ICSS-ETHERCAT-SLAVE release is planned starting June. I will pass latest observations to our development team.

thank you,

Paula

Hi Paula,

We have demo scheduled for our customer in mid July. Is it possible that I could get 01.00.03 PRU-ICSS-ETHERCAT-SLAVE Beta/RC pre-release version to try out?

Thanks,

Karl

Hi Karl, let me ask and come back to you.
thank you,
Paula

Karl, release is in TI.com

Fixed issues:

PINDSW-1396

AM57x

EtherCAT Firmare upgrade with TwinCAT not working on AM572x

PRU-ICSS EtherCAT Release Notes

Please give a try and let us know how it works in your side.

thank you,

Paula

Paula,

Could you post a download link for the new version 01.00.03?

I see the release notes 01.00.03 but http://www.ti.com/tool/PRU-ICSS-INDUSTRIAL-SW 

Karl, you can download it from here:
software-dl.ti.com/.../index_FDS.html
thanks
Paula

Paula, thanks a lot. I will give a try today. Karl.

Paula,

I tested the new firmware 01.00.03. The reported issues are partially fixed.

The working counter now correct. However, when testing TxPdo size of 2K Bytes  for each EtherCAT cycle of 2ms, there are two PDI_Isr() interrupts instead of one  that triggered two calls of PDO_InputMapping(). These two interrupts are about 300μs apart. See the attached image (A1 is the first interrupt, A2 is the second interrupt, 2ms is the cycle time).  

Then I tried 6 KBytes TxPdo, for each EtherCAT cycle, there are three PDI_Isr() interrupts coming about 300μs apart. Why?

In comparison, when testing TxPdo size of 2K Bytes with Bechhoff EL9800 hardware, there is only one PDI_Isr() interrupt for each EtherCAT cycle of 2ms.

Please help.

Thank you,

Karl

Hi Karl, I will pass this info to the developers.
thank you,
Paula

Hi Karl, it was a mismatch in our testing, we wil look into this and let you know soon (probably by next week).

thank you,
Paula

Hi Paula,

Thanks for the update. We really need this resolved ASAP.

Thank you very much,
Karl

Status update?  Is there a purchase option for the source code to EtherCAT PRU firmware?  We have missed several deadlines and without a clear schedule it makes it hard to continue this path.

Hi Tim, We are planning to address this in our release v1.0.4 during July. Will share exact details by next week (07/05).
Thank you,
Paula

Hi Paula,

Thanks for the update

Karl

Paula,

Is it possible to get the test plan for validating this issue?  For example, will the test cover the following:

• 6K TxPDO (largest size PDO possible with the A437x)
  • timing validation
    
    • minimum time it takes  process 1 packet (at max packet size, 1468bytes)
    • minimum time it takes to process entire 6Kbyte PDO  (time it takes to process 5 packets)

• 6K TxPDO,  128 byte RxPDO, (2) 256 byte mailbox
  
  • same as above, validating timing.

The goal being the issue is understood and fix is also validated without waiting another test/reporting/fix cycle.

Also, is it possible to share the test configuration? i.e software for the master, type of platform used for the master

Hi Tim, I will convey your test comments to the developers and I will ask for our test configuration setup information.

thank you,

Paula

Hi Tim, a quick clarification,
Could you please clarify us what do you mean by "time taken to process 1 packet"?. Do you mean the total time taken from the PDI side writing until the Ecat master reads?

thank you,
Paula

Paula Carrillo said:

Hi Tim, a quick clarification,
Could you please clarify us what do you mean by "time taken to process 1 packet"?. Do you mean the total time taken from the PDI side writing until the Ecat master reads?

Correct, the goal is to validate the following:

• The time it takes to process the interrupt for the first packet
  
  • enter the ISR
  • copy 6K to shared memory
  • Handshake with the PRU
  • exit the ISR
• remaining packets are processed by the PRU without a PDI interrupt
• The PRU's are able to process and forward data, at maximum Ethernet packet rate.

The goal is to for EtherCAT slave is to keep up with at-least 65Mpbs from the master.

Tim, we can have an Engineering Release by the end of this week with the fix. We would like to share it with you, before our final release (planned end of July), so we can confirm everything works as requested.

Could you please send me a private message with your email so I can set a SFTP?

thank you,

Paula

Paula,

PM sent.

Tim