New experimental ethercat layer
- Hakan
- Away
- Platinum Member
-
Less
More
- Posts: 1334
- Thank you received: 482
15 Nov 2026 21:54 - 15 Nov 2026 22:00 #350735
by Hakan
Replied by Hakan on topic New experimental ethercat layer
New release.
- EoE works
- Serial-over-EtherCAT for EL6xxx-series works
- XDP program now stays attached to the interface
- Numerous fine-tunings and some issues fixed during testing.
EoE was tested on one of my EaserCat cards. EoE slave functionality (from SOES examples) firmware was built and flashed. Performance is around 0.5 Mbit/sec on a 0.5 ms servo loop, even under heavy load test. Not blazingly fast but enough for configuration, web page with live update of numbers, etc. EoE has not been tested on a commercial unit.
EL6xxx functionality was programmed onto the same EaserCat card from EL6021 documentation. Serial functionality tests cleanly up to 115200 baud and all variants of data length, parity and stop bits. Not tested on a commercial unit.
By letting the XDP program stay attached to the interface after shutdown of the ethercat master it takes that master just a few seconds from start to OP of all slaves.
EoE and serial are in the docs.
- EoE works
- Serial-over-EtherCAT for EL6xxx-series works
- XDP program now stays attached to the interface
- Numerous fine-tunings and some issues fixed during testing.
EoE was tested on one of my EaserCat cards. EoE slave functionality (from SOES examples) firmware was built and flashed. Performance is around 0.5 Mbit/sec on a 0.5 ms servo loop, even under heavy load test. Not blazingly fast but enough for configuration, web page with live update of numbers, etc. EoE has not been tested on a commercial unit.
EL6xxx functionality was programmed onto the same EaserCat card from EL6021 documentation. Serial functionality tests cleanly up to 115200 baud and all variants of data length, parity and stop bits. Not tested on a commercial unit.
By letting the XDP program stay attached to the interface after shutdown of the ethercat master it takes that master just a few seconds from start to OP of all slaves.
EoE and serial are in the docs.
Last edit: 15 Nov 2026 22:00 by Hakan.
Please Log in or Create an account to join the conversation.
- grandixximo
-
- Offline
- Platinum Member
-
Less
More
- Posts: 360
- Thank you received: 412
16 Nov 2026 00:37 #350737
by grandixximo
Replied by grandixximo on topic New experimental ethercat layer
Great work, keep it up!
Please Log in or Create an account to join the conversation.
- Hakan
- Away
- Platinum Member
-
Less
More
- Posts: 1334
- Thank you received: 482
16 Nov 2026 00:49 - 16 Nov 2026 00:52 #350738
by Hakan
Replied by Hakan on topic New experimental ethercat layer
0.20 Performance release
- Reduced stable cycle time
- Slave sync and timing values are in the gui, both limits from the slave info and measured during I/O test.
Several improvements bring down the servo-loop cycle-time to 125 μsecs for 4 slaves, 3 fully DC synced, during load stress-test (load average above 10). Tests used QtPlasmaC during machine operation and cutting pieces (in the air). Tests with shorter cycle times show that the drives refuse to go below 125 μsecs. Based on the tests a lower limit on servo-loop cycle-time would be around 80-100 μsecs if the drives allow it.
Useful synch and timing information is presented per slave in the gui. The drives I have don't show anything useful from the usual registers; findings from measurements will be presented instead if available. A number of measured, actual timing parameters such as turn-around cycles are shown live in the gui.
As is the custom, the document contains more information.
- Reduced stable cycle time
- Slave sync and timing values are in the gui, both limits from the slave info and measured during I/O test.
Several improvements bring down the servo-loop cycle-time to 125 μsecs for 4 slaves, 3 fully DC synced, during load stress-test (load average above 10). Tests used QtPlasmaC during machine operation and cutting pieces (in the air). Tests with shorter cycle times show that the drives refuse to go below 125 μsecs. Based on the tests a lower limit on servo-loop cycle-time would be around 80-100 μsecs if the drives allow it.
Useful synch and timing information is presented per slave in the gui. The drives I have don't show anything useful from the usual registers; findings from measurements will be presented instead if available. A number of measured, actual timing parameters such as turn-around cycles are shown live in the gui.
As is the custom, the document contains more information.
Last edit: 16 Nov 2026 00:52 by Hakan.
Please Log in or Create an account to join the conversation.
Time to create page: 0.149 seconds