Hello, A lot more experiments further now. My findings:
Tcpdump and wireshark based observations with as well the 1.4 as the 1.5.2 version of the IGH EtherCAT master: - when a frame exceeds the 128 byte threshold then increasingly often the latency of the response frame increases (seems a 100microsecond offset) but our time budget (time between send() and receive()) is 100 microseconds. This is the cause for working count 0 errors. - as long as the process data is such that frame size stays below ~128 bytes there is no problem, the working counts stay stable and response latency is more or less constant Kernel observations: - clock_nanosleep wake up latency (measured with cyclic test) is similar for the two kernels - the 1.5.2 version of the master on the 2.6.29 kernel with rt-preempt running at 5Khz with 97 bytes of process data runs just fine ;-) - the 1.4 version with the the r8169 driver from 1.5.2 running on 3.4.97 rt-preempt has the same problems as the 1.5.2 version of the master on the same kernel running at 5Khz with 97 bytes of process data - use the generic module when operating at 5Khz (only tested with 1.5.2 with frame size less than ~128 bytes) gives the same working count 0 symptoms, for our application we really seem to need the patched drivers... Use of the IGH EtherCAT Master user library: - ecrt_master_set_send_interval was not present in our application software so the EtherCAT-OP thread was running at 1Khz instead of the 5Khz of the real-time thread. This is not critical right? - in some experiments the extra datagrams injected by the EtherCAT-OP thread were driving the frame size higher than the ~128 bytes boundary, causing problems at 1Khz... In summary, I now think that we have a problem with the r8169 Realtek driver or modified Linux behaviour which increases the latency of delivery of ethernet frames. Best, -- Jeroen van der Wulp Software Designer ___________________________________________________________ Sioux. Source of your Technology Technical Software | Electronics | Industrial Mathematics | Remote Solutions Sioux Embedded Systems B.V. | Esp 405 | 5633 AJ Eindhoven | Nederland T +31 (0)40 2677100 | jeroen.van.der.w...@sioux.eu | www.sioux.eu _______________________________________________ etherlab-users mailing list etherlab-users@etherlab.org http://lists.etherlab.org/mailman/listinfo/etherlab-users
