Haven't used RT-Preempt, but I think you use:
#define CLOCK_TO_USE CLOCK_MONOTONIC clock_gettime(CLOCK_TO_USE, &currTime); clock_nanosleep(CLOCK_TO_USE, TIMER_ABSTIME, &wakeupTime, NULL); Graeme. ________________________________ From: Jiarui Lian <[email protected]> Sent: Friday, 13 January 2017 11:22 To: Graeme Foot Subject: Re: Patch for Distributed Clock? Hi, Graeme: Thanks a lot: I have downloaded and check new version now. I am using RT-Preempt, and your sample code is using RTAI: * Sample code uses functions from RTAI: rt_get_time_ns(), rt_get_time(), rt_.... * RT-Preempt seems not having equivalent functions. Can we just use regular linux/time.h functions? I am afraid that regular time.h functions can't fulfill real-time requirements, what do you think? Anyway I should try. Thanks! Jerry ________________________________ From: "Graeme Foot" <[email protected]> To: "Jiarui Lian" <[email protected]> Cc: "etherlab-users" <[email protected]> Sent: Thursday, January 12, 2017 5:03:04 PM Subject: Re: Patch for Distributed Clock? Hi, Thats one of the older forum postings and is code for an older version of the master (revision 2266). The changeset I sent is for master revision 2526 (still old but I haven't needed to move on yet). Have a look at this one: http://lists.etherlab.org/pipermail/etherlab-users/2016/003026.html Note: there is a bug in the ecMod_syncDistClock() funciton from that post. It should probably instead be (changed line in bold): int32_t ecMod_syncDistClock( void *this /**< pointer to module etherCATModule_s */ ) { etherCATModule_s *ecMod = this; uint32_t masterTime; // cache lower 32 bits of prev master time and get now masterTime = (uint32_t)ecMod->m_dcTime; ecMod->m_dcTime = app_getTimeNS(); // use the dc ref slave to adjust the masters time base // get lower 32 bit of clock time from reference slave (after first scan) if (ecMod->m_getDCDiff) { int res; uint32_t slaveTime; res = ecrt_master_reference_clock_time(ecMod->master, &slaveTime); switch (res) { case 0 : { // calc time diff ecMod->m_dcDiff = masterTime - slaveTime; } break; default : { // no ref clock found or datagram failure ecMod->m_dcDiff = 0; } } } else { ecMod->m_dcDiff = 0; ecMod->m_getDCDiff = true; } // call to sync slaves to ref slave // (which is used for ecrt_master_reference_clock_time) ecrt_master_sync_slave_clocks(ecMod->master); // set master time in nano-seconds ecrt_master_application_time(ecMod->master, ecMod->m_dcTime); return 0; } Graeme. ________________________________ From: Jiarui Lian <[email protected]> Sent: Friday, 13 January 2017 04:57 To: Graeme Foot Subject: not found Re: Patch for Distributed Clock? Hi, Graeme: I couldn't found the following functions in your patch: (as you mentioned in: http://lists.etherlab.org/pipermail/etherlab-users/2012/001642.html) 1) ecrt_master_setup_distributed_clock() 2) ecrt_master_sync_slave_clocks_diff() 3) I have changed the ref_sync_datagram to 4 bytes instead of 8. Did you send the wrong patch, or you change the function names? Thanks! Jerry ________________________________ From: "Graeme Foot" <[email protected]> To: "Jiarui Lian" <[email protected]> Cc: "etherlab-users" <[email protected]> Sent: Wednesday, January 11, 2017 5:05:53 PM Subject: Re: Patch for Distributed Clock? Hi, I've attached an hg changeset of the patches I use. You can get the DC related ones from it. If you look in the etherlab-dev forum you will also find Gavin's patchsets which will be more up to date. Do a search for "etherlab-users dc" (or similar) to find posts related to how to set up distributed clocks. eg: http://lists.etherlab.org/pipermail/etherlab-users/2016/003014.html But to answer you question: A system requires one master clock that all other clocks in the system sync to. That clock can either be the clock in the EtherCAT master or it can be a clock on one of the slaves (the dc reference slave). Separate to that, all dc slaves on the bus need to be synced to a reference slave clock. This should be the first slave on the bus that supports dc. The EtherLAB master will automatically select this for you, or with the patch you can select it yourself. Generally the EtherCAT master clock has too much jitter to provide a nice stable system, so instead I use option two where my dc reference slave is the master clock and I adjust the EtherCAT master clock to it (this is the default option used by TwinCAT). Also note (as described in the post I linked to above), there are two levels of dc support in slaves. DC clock level support (alot of simple IO slaves have this) and the ability to sync the IO to the dc clock (generally only more advanced slaves support this). The DC reference slave only requires the DC clock level of support. In your case, if your AX5206 slave is the first slave it can be both the reference slave and the DC clock master. Regards, Graeme. PS: Our system uses various IO modules and amps. Our dc reference slave is often our first IO module (an EL1008). ________________________________ From: Jiarui Lian <[email protected]> Sent: Thursday, 12 January 2017 06:39 To: Graeme Foot Subject: Patch for Distributed Clock? Hi, Dear Mr. Graeme Foot: I am studying IgH-EtherCAT-Master to control AX5206, and I saw your post in 2012: http://lists.etherlab.org/pipermail/etherlab-users/2012/001642.html * Would you mind to send a copy of the patch to me? * And a question, when you said: "pick a ref slave and update my master time based on the ref slave time." Does your system consist of at least two slaves with DC capability? Master(Jitter) ---- Slave.1 (DC, Ref-to-Master) ---- Slave.2(DC, Ref-to-Slaves) ---- MoreSlaves In my case, I have only one slave with DC: Master(Jitter) ---- Slave.1(AX5206, DC) --- MoreSlaves Any suggestion to me? Thanks, your help is appreciated! Sincerely, Jerry Bertec Corp.
_______________________________________________ etherlab-users mailing list [email protected] http://lists.etherlab.org/mailman/listinfo/etherlab-users
