MCU-PLUS-SDK-AM243X: Endat2.2 recovery time counter for safety

Hi Alan,

Thanks for your query.

I will check on this and get back to you.

Best Regards

Achala Ram

Hi Achala

Thank you for following-up on RT counters. Heidenhain documents suggests Non reset RT counter method as a solution to detect stuck at fault. We are open to other method/solution as long as we can demonstrate stuck at fault detection.

Thank you

Alan I

Hi Achala

Hope you are doing well. Do you have any update on this enquiry?

Thanks

Alan I

Alan

Please note that this is a holiday season in India. Please expect the response by next week.

Feel free to ping the thread next week if you do not get a response within a few business days.

Regards

Dhaval

Hello Ti experts

I am just pinging the thread 

I would like to understand if this requirement can be addressed

Thank you

Alan I

Hi Alan,

Sorry for delay!!

Please refer to the documentation : AM243x Motor Control SDK: EnDat Protocol Design. for the RT measurement method supported by current EnDAT firmware. 

Method is as follows: we start counter from zero at last rising clock edge, and then read counter value on the following falling Data edge. The counter value read at the falling data edge is RT, and we store this value in shared memory as recovery time. The counter continuously increments between last rising edge of the EnDat-Clock (positive signal) and the next falling edge of the EnDat-Data (positive signal). 

Regarding "stuck-at fault detection", We are checking with Heidenhain that how we can detect & validate "stuck at" error with non-resetting counter values!

Can you share your use case of stuck at error with RT ? Share the conditions where you think that it should detect as stuck-at error. 

Thanks & Regards,

Achala Ram 

Hi Achala

Thank you for the reply. As you noticed in my original post, the requirement for non-reset RT counter is a requirement specified by Heidenhain in their document "Heidenhain document (D1128897 - 03 - A - 02 8) - Safety with EnDat 2.2 and Non-Safe EnDat Master"

Please refer to section 3.5.2 of above document.

It is good that you are in contact with Heidenhain about this topic and we will be waiting the result of your discussion.

Let me know if you need more information

Thank you

Alan I

Our use case for RT is specified by section 3.3 of "Heidenhain document (D1128897 - 03 - A - 02 8) - Safety with EnDat 2.2 and Non-Safe EnDat Master"

RT Test requirement is specified by section 4 of the same document

Where RT Test needs to be performed as part of Safe Operation test 

Let me know if you need more information

Alan I

Hello Achala

Do you have any update on non-Reset RT counter?

Thanks

Alan I 

Hello Achala

Do you have any update on this issue?

Thanks

Alan I

Hi Alan,

After reviewing the Heidenhain documentations, we found that some steps are missing to meet safe CU requirements in our current firmware. Thanks for bringing this to our attention.  We will enable all the necessary safety related requirements, including stuck at error in the firmware, which will be available in the next motor control sdk release.

Thanks & regards,

Achala Ram

Achala

Thank you for the reply. I noticed the motor control SDK 9.2.0.11 was released few weeks ago. Do you have indicative time frame for next sdk release.

I am having problem to get RT measurement going on  mode 200: Start periodic continuous mode  ( frorm endat diagnotic) 

Attached is my modified version of endat_drv.c / and endat_diagnotic.c where I tried to use pos_cmd=9  in periodic trigger mode.

endat_diagnostic.cendat_drv.cendat_drv.h

Let me know if you need more information

Thanks

Alan I

Alan Ismail said:

I am having problem to get RT measurement going on  mode 200

I am reviewing your changes meanwhile can you share, What issue are you facing with this? and what values are you configuring for IEP reset cycle & trigger point ?

Thanks & regards,

Achala Ram

Hi Achala

I tried several IEP Cycle  ie , 200000 , 400000 , trigger was set at 1000

I have configured the encoder to do short RT 

I have modified the endat_diagnotic.c such that  cmd 200 uses Endat Mode 9 with MRS code of 0x4d. In theory the RT should be performed on every endat mode 9 transaction.

Let me know if you need more info.

Thanks

Alan I

Hi Alan,

I am getting RT for CMD 9 with your changes, but these values are in PRU cycle counts, not in nanoseconds, which is expected. The endat_command_process function performs the RT conversion, and endat_get_recovery_time returns the RT in nanoseconds. However, in periodic mode, this function is called only once, so the RT conversion does not happen after the first measurement, and endat_get_recovery_time returns the RT in PRU cycle counts.

Thanks & Regards,

Achala Ram

Hi Achala

Thank you for trying out my modified version of endat_drv.c

If I understand correctly , the last column  ( value 571) is cycle count , then value in ns is  571 * 1/300e6 * 1e9 = 1903 ns which matches short RT value of Endat Encoder  ~2us.

May I know the IEP cycle and trigger points used for example above

Thank you for your hep

Thanks

Alan I

Hi Alan,

Alan Ismail said:

I tried several IEP Cycle  ie , 200000 , 400000 , trigger was set at 1000

I used, 200000 for cycle count & 1000 for trigger point. 

Alan Ismail said:

then value in ns is  571 * 1/300e6 * 1e9 = 1903

PRU clock is set to 200MHz, so exact value will be 571*(1/200e6)= 2.8us.

Thanks & Regards,

Achala Ram

Archala, thank you for the clarification.

Could you let me know estimated release date for next motor control SDK

Thanks

Alan I

Hi Alan,

Can you please review the method for RT and confirm, will this work for you?

To measure Recovery Time (RT), we will use the PRU cycle counter. The measurement begins by setting the PRU counter to zero at the last rising clock edge (CLK line). Afterward, the firmware will read the PRU counter value at the next falling Data edge (RX line). This captured value represents the Recovery Time (RT), which is stored in the lower 32 bits of a 64-bit timer.

The upper 32 bits of the 64-bit timer will function as a 32-bit counter, incrementing with each command request based on the measured RT. This counter can be used to detect “stuck-at” faults: if the counter fails to increment after an RT measurement, it indicates a potential fault.

Thanks & regards,

Achala Ram

Hello Achala

We are just a user and we can not make any recommendation on the implementation method. I urged you to work with Heidenhain to meet all implementations and testing requirements as outlined in Heidenhain document D1128897 - 03 - A - 02 Safety with EnDat 2.2 and Non-Safe EnDat Master

Thank you

Alan I

Hello Achala

Happy New Year. Do you have any update on this subject? 

Thank you

Alan I

Alan

Sorry for the delayed update. We will provide an update by end of this week.

Regards

Dhaval

Hi Alan,

Sorry for the delayed update.

We have started the implementation for RT, and once it is completed, we will share the patch with you (next week). It will take approximately one week to finish development, review, and testing. 

And regarding the motor control SDK release date, we will provide an update on that soon.

Thanks & regards,

Achala Ram

Hi Archala

Thank you for looking into this issue and we are eagerly awaiting the patch

Alan I

Hi Alan,

Files to Update/Modify:

• motor_control_sdk_am243x_09_02_00_09\examples\position_sense\endat_diagnostic\endat_diagnostic.c
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\driver\endat_drv.c
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\firmware\endat_diagnostic.cmd
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\firmware\endat_interface.h
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\firmware\endat_main.asm
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\include\endat_api.h
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\include\endat_drv.h
• motor_control_sdk_am243x_09_02_00_09\source\position_sense\endat\include\endat_interface.h

Action to Take:

• Update or copy all these files from the provided attachment 
• Additionally, make sure you use the updated firmware binary and driver libraries that are also in the attachment.

motor_control_sdk.zip

Thanks & Regards,

Achala Ram

Hi everyone,

I have a question related to this topic.

We also plan to use TI Sitara AM243x in a safety related application, for which we need a safe position information. We´d like to follow the same implementation note by Heidenhain (Safety with EnDat 2.2 and Non-Safe EnDat Master).
Is it possible to follow the guidelines provided by Heidenhain using the EnDat2.2 master from the AM243x MotorControlSDK or are there any further restrictions?

Thanks in advance & best regards
Christian Brunner

Yes Christian,

We have implemented RT as per the method and guidelines described in Heidenhain document (D1128897 - 03 - A - 02 8) - Safety with EnDat 2.2 and Non-Safe EnDat Master, This will be available on ti.com with next AM243x Motor Control SDK release.  

According to 

Hi Achala

Thank you for Patch. I have run the patch using endat_diagnostic.c and RT looks much better than before.  I have one additional question regarding RT.

Figure 6 from (D1128897 - 03 - A - 02 8) , shows that RT register is only updated once per safety cycle. 

 Question : In order to match RT counter update rate as per figure 6,  Can I enable RT measurement only to specific MRS code during safety cycle

Thank you

Alan I

Hi Alan,

Alan Ismail said:

Can I enable RT measurement only to specific MRS code during safety cycle

can you share more details on this ? to detect "stuck-at' RT -register should change in normal operation also. 

Thanks & regards,

Achala Ram