AM335x EtherCAT sync0 time

Hi Louis,

what tests did you do and what was the exact result? What software did you test and what master was used?

Of course we can only speak for TI examples and your own application code may have other limits.

As far as I can remember there is an ESC reg for sync pulse length. Default was set to 10us.

Regards.

Hi Frank,

I used the ACS motion as my EtherCAT master to test the ACS slaves and AM335x_ICE v1.0A( use SDK v1.1.0.3) and our product( use SDK v1.1.0.3).

Because the cyclic time which is 500us of master is fixed, but the sync0 cycle time is can be modify in the DC mode section of the ESI file.

1.Sync0 cycle time setting is 50us of ACS slaves, and it can run to OP mode with ACS master.

2.I tested the AM335x_ICE which used the AM3359ZCZ chip, I set the sync0 cycle time to 400us and it can run to OP mode with ACS master, but communication failed to modify the sync0 cycle time down to 250us.

3.We have a product which used the AM3359BZCZ chip, and 250us sync0 cycle time is work(OP mode), but 200us failed(can not run to OP mode).

I have find how to modify the sync0 pulse length, thanks for your reply.

Louis,

not sure if I can help here...

Who or what is 'ACS'? Is this related to Acontis software? Do you have a link?

I don't undertsand why you want to set the Sync time to a lower value than the master cycle time. Also I assume the DC sync time in the ESI file just defines a limit value for the master. It does not have any real effect on a device. So I assume you would need to set the sync time on your master so that it sends packets accordingly.

I assume you need to take wirehshark logs and analyze the details if your master does not provide any debug info.

I am testing with Beckhoff PLCs and there I also need to set the master/PLC task times accordingly. We also test ICE boards with Beckhoff PLCs down to 250us for every SDK release. So I know that works. I have tested lower values in the lab before but that was not a production test.

Regards.

Hi Frank,

ACS Motion Control have a SPiiPlus MMI application studio which is a development tool and used RTOS32 as real-time system which bound with the Acontis EtherCAT master stack.

I think the frequency of cyclic frames can be different with sync0 cycle time that was verified by ACS motion control master, but the sync1 cycle time (input data)  need set to as same as the frequency of cyclic frames.

ACS Motion Control Master and TwinCAT both used the <CycleTimeSync0 Factor="-16">0</CycleTimeSync0> in <OpMode> section to configuration slave. I think the value does not mean the limit value.

Among your own test, does you disable some define feature for simplification the EtherCAT stack? e.g. EoE,FoE?

I have used wireshark to get logs and I find some frame did not back from slave side, some frame disappeared after the master send out.

Do you known the Sync0 cycle time of demo video, is 100us?
http://www.youtube.com/watch?v=xw0bfjsk7Og

Hi,

2.I tested the AM335x_ICE which used the AM3359ZCZ chip, I set the sync0 cycle time to 400us and it can run to OP mode with ACS master, but communication failed to modify the sync0 cycle time down to 250us.

3.We have a product which used the AM3359BZCZ chip, and 250us sync0 cycle time is work(OP mode), but 200us failed(can not run to OP mode).

What frequency your product is running?

Did you profile how much time it takes for your application to execute SYNC task and PDI task ?

Is there any more description on the error (other than can not run to OP mode) ?

Is this possible to share wireshark logs?

From ETG2000, table49, Element Factor

Used to define SYNC0 signal cycle based on the sync unit cycle time (= EtherCAT bus cycle time = cycle time in which the process data of this device is updated).Sign of value indicates mathematical operation:
Positive value:
SYNC0 cycle time = sync unit cycle time * |Factor|
Negative value:
SYNC0 cycle time = sync unit cycle time / |Factor|
NOTE: If CycleTimeSync0 and CycleTimeSync0@Factor are “0” this  results in a total cycle time = 0. This may activate special features  (single shot).

So you can run SYNC1 at bus cycle time by setting positive Factor value.

We do not disable features during our tests. But we are not running EoE or FoE during SYNC tests either.

Hi PratheeshGangadhar,

I used another EtherCAT Master (IgH EtherCAT Master) and it can communication with our product (AM3359BZCZ) and the cycle time can be down to 100us. So I think there have some special synchronization method perhaps in their ACS Motion Slaves.

But there have another situation about that IgH EtherCAT Master can not configuration to send LRW cmd for our product (AM3359BZCZ), but configure as LRD cmd add LWR cmd for the product will be communication to OP mode.
And the ACS Motion Slaves which use the ET1100 can be communication with IgH with LRW cmd.
2555.IGH_DC_1Domain_1ms_PREOP_mode.rar

Please see No. 1584 frame, after this frame a lot of frames send out from Master but did not receive the come back frames.

Have any suggestion?

Hi,

It is work.

Thanks for your reply very much.