Repository navigation
Releases
ecat ecat-v0.23.1
A bug-fix release for LinuxCNC 2.9 and 2.10: CiA 402 drives whose process data has no operation mode (such as the StepperOnline A6 with its default PDOs), and help for drives that will not stay enabled because frames are lost.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.5 from MetalMusings/ethercat (unchanged since 0.21): sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb. Then the file for your LinuxCNC:
- LinuxCNC 2.9 (2.9.4 or newer, as packages):
linuxcnc-ecat_*+lcnc2.9.debian13_amd64.deb(or.debian12). - LinuxCNC 2.10 run-in-place: unpack
ecat-rip-0.23.1-debian13.tar.gz(or-debian12) and run./install.ecat.sois built against LinuxCNC commit b0e7b191b9 (2.10.0~pre2-436); the installer refuses if your tree's HAL headers differ and says how to build from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.debwithout+lcnc2.9.
Moving from the IgH kernel master (EtherLab)? Stop it and keep it from starting before you use the userspace master: sudo systemctl stop ethercat; sudo systemctl disable ethercat, and restart the computer if lsmod | grep ec_ still lists modules. Two masters on one network card take each other's frames.
The window, ecatsetup --gui, needs python3-pyqt5. See the README and getting started.
What is new in 0.23.1
Fixed: CiA 402 drives without the operation mode in their process data. The StepperOnline A6's default PDOs (0x1701/0x1B01) have neither 0x6060 (mode) nor 0x6061 (mode display). The cia402 profile then enabled the drive in whatever mode it had (0, no mode), and the drive dropped out of "Operation enabled" at once. Now the profile writes the configured mode by SDO when 0x6060 is not mapped, reads it back, and the master writes it again at each configuration; a drive that refuses the mode stops the load with a message. (cia402)
Fixed: EtherCAT Setup proposed generic for the A6. Its ESI gives the CiA 402 profile as Profile/ChannelInfo/ProfileNo, and the drive has no SDO information, so 0x1000 was never read. 0x1000 is now read from every CoE slave, and the ESI's dictionary decides what else is read. Propose again on the slave page turns a generic entry from an older version into cia402.
New: "What this means" for the event log. When the event log shows working counter zero while running, or the master's skipped, unmatched or timed-out datagrams, the Event log card on the Troubleshoot page starts with what it means (in such a cycle ecat shuts every drive down, which is why a drive reaches "Operation enabled" and falls back) and the usual causes, each judged on this computer: likely, possible, not checked (with how to find out) or ruled out. The causes looked at: the IgH kernel master still loaded or enabled for the next boot, the servo thread late, no PREEMPT_RT kernel, no isolated CPU, Energy Efficient Ethernet on the card, the card also carrying the LAN, the card's interrupt, a Realtek (r8169) card, and a cable or a slave's power. The same text is at the top of the report and after ecatsetup --events. (docs)
The report holds more of what decides whether frames get through (Save report... or ecatsetup --report): the EtherCAT masters on the computer (the IgH kernel master's device, modules and ethercat.service, free userspace masters, LinuxCNC), the running master's own status with its lost-frame counter, the transport, the computer as the real-time check reads it (kernel command line, isolated CPUs, network cards with driver, EEE and coalescing, ...), and where EtherCAT Setup runs from (a run-in-place archive's version). Disk UUIDs on the kernel command line, MAC addresses and your home folder's name are left out of the whole report.
Tried before the release: the A6 fix against a simulated A6 (an EaserCAT 3000 board acting as the drive, with the A6's identity, PDOs and missing SDO information) on a real bus; "What this means" and the report on this machine's own event log and bus. Not tried: a real A6 (a user is about to).
ecat 0.23
ecat for LinuxCNC 2.9 and 2.10, with a new part: the real-time check (the Timing tab), for any LinuxCNC computer, EtherCAT or not.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.5 from MetalMusings/ethercat (unchanged since 0.21): sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb. Then the file for your LinuxCNC:
- LinuxCNC 2.9 (2.9.4 or newer, as packages):
linuxcnc-ecat_*+lcnc2.9.debian13_amd64.deb(or.debian12). - LinuxCNC 2.10 run-in-place: unpack
ecat-rip-0.23-debian13.tar.gz(or-debian12) and run./install.ecat.sois built against LinuxCNC commit b0e7b191b9 (2.10.0~pre2-436); the installer refuses if your tree's HAL headers differ and says how to build from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.debwithout+lcnc2.9.
The packages now recommend stress-ng and ethtool, which the real-time check uses; apt installs them with ecat. The window, ecatsetup --gui, needs python3-pyqt5. See the README and getting started.
What is new in 0.23
The real-time check: where jitter comes from, and what to change. A new Timing tab in EtherCAT Setup, and ecatsetup --rt-check on the command line (docs):
- Measures how late every real-time thread starts, cycle by cycle: beside a running LinuxCNC (any I/O: Mesa, EtherCAT, parallel port; nothing that talks to the I/O is loaded) or the INI's threads alone in halrun; at rest, as the computer is used, or under
stress-ng. A small probe component,ecat_rtprobe, is shipped besideecat.sofor this. Twice for a baseline, so that a later change counts only when it moves the numbers by more than chance does. - Reads the computer: kernel and its parameters, isolated CPUs, hyperthread siblings, frequency and idle states, every interrupt and where it is handled, network cards (coalescing, EEE), graphics, ACPI GPEs, and with As root SMIs and
rtla timerlat. Late cycles are matched against what happened at the same time. - Findings from a rule catalogue (YAML, 41 rules in five areas; your own rules in
~/.config/ecatsetup/rtrules/), each with what was seen, why it matters, what to do, risk, cost, how to check it, how to undo it and its source. - Look or Improve. In Improve, a run-time change is tried with one button (it asks first, runs with sudo, never needs a restart), measured again, and kept or undone. Every change is logged, with an undo script.
- To do yourself: the changes to make lasting, written out ready to use: the new
GRUB_CMDLINE_LINUX_DEFAULTline, sysctl and modprobe files, a start-up service, the BIOS. - Save report for asking for help.
Tested on two of our computers: settings that were known to be right were taken away one at a time, and the check found them again. A user then tried it alone on a third computer, using only the window: a Mesa Ethernet machine with LinuxCNC 2.9.4. Fixed from that test:
- With load now says when
stress-ngis missing, and does not start. - A finding says when
ethtoolis missing. - Two new rules: an isolated CPU that no real-time thread uses (for example
isolcpus=1,2,3with the threads on 3 →isolcpus=1,3), and the Mesa card's interrupt on the real-time CPU. The Mesa interrupt rule has not been tried on a Mesa card yet.
Choose the machine's INI file before measuring, even when LinuxCNC is not running. The check finds the I/O's network card from it. Without it, the rules about that card cannot see it.
The EtherCAT component (ecat.so) is unchanged since 0.22.1; only the probe ecat_rtprobe.so is new.
ecat ecat-v0.22.1
ecat for LinuxCNC 2.9 and 2.10: from this release the same ecat is built for both series. LinuxCNC 2.9 (2.9.4 or newer, Debian's or linuxcnc.org's packages) gets a package; 2.10, which has no Debian packages yet, gets the run-in-place archive and a package as before.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.5 from MetalMusings/ethercat (unchanged since 0.21): sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb. Then the file for your LinuxCNC:
- LinuxCNC 2.9 (2.9.4 or newer, as packages):
linuxcnc-ecat_*+lcnc2.9.debian13_amd64.deb(or.debian12). It replaces ecat 0.11. - LinuxCNC 2.10 run-in-place: unpack
ecat-rip-0.22.1-debian13.tar.gz(or-debian12) and run./install.ecat.sois built against LinuxCNC commit b0e7b191b9 (2.10.0~pre2-436); the installer refuses if your tree's HAL headers differ and says how to build from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.debwithout+lcnc2.9.
The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.22.1
ecatsetup --check names HAL names that are too long for your LinuxCNC. LinuxCNC 2.9 takes at most 47 characters in a pin or parameter name, 2.10 takes 55. A longer one (often a pin named after a long ESI entry) made the component refuse to load, with only "length 48 too long for name" in LinuxCNC's error window. --check and the window's check now say which pin and how long it may be, so you can give it a shorter name before you start LinuxCNC. Found when 0.22 was first run in LinuxCNC 2.9.
Tried in LinuxCNC 2.9.4 with motion (0.22 + this check): a plasma cutter with three Rtelligent ECT60 drives in CSP and an I/O board, 500 us servo period: homing, jogging, all slaves in OP, two cycles from output to status, no events while it ran.
What is new in 0.22
LinuxCNC 2.9. Everything up to 0.21 (tuning, feedforward, EoE, serial channels, unused slaves, the DC reference clock, From lcec, the window) now works on 2.9 too. The component is written against 2.10's HAL API; on 2.9 one header gives that API on 2.9's pins, so there is no separate 2.9 code. Where the series differ:
- Integer pins have 32 bits on 2.9 (64 on 2.10). Commands are held within the PDO entry's range anyway, so entries up to 32 bits behave the same.
- LINT/ULINT entries and the encoder's 64-bit count are
floatpins on 2.9 (exact up to 2^53).ecatsetup --checkrefuseshal-type: sint/uinton a LINT/ULINT entry on 2.9 and proposesreal;extendis for 8- and 16-bit entries only. - The same
ethercat.yamlworks on both, when its HAL names have at most 47 characters. - EtherCAT Setup finds the series by itself (HAL types, INI keys such as
HOME_SEARCH_DISTonly on 2.10, component names, LinuxCNC's pins).
ecat 0.11.x is closed. 2.9 users move to this package; 0.11 gets no more fixes.
Fixed: the 2.10 package lacked the shipped SYNC0 measurements (ecatsetup/measured/) since 0.20; the run-in-place archive had them.
Tried on 2.9.4 before 0.22.1: ecatsetup --check and the I/O test in the window on a plasma cutter's bus (three ECT60 drives and an I/O board): all slaves in OP, DC phased, two cycles from output to status, the drives without fault. Not tried on 2.9 yet: 2.9.5 to 2.9.10 (same HAL API), Debian 12.
ecat ecat-v0.22
ecat for LinuxCNC 2.9 and 2.10: from this release the same ecat is built for both series. LinuxCNC 2.9 (2.9.4 or newer, Debian's or linuxcnc.org's packages) gets a package; 2.10, which has no Debian packages yet, gets the run-in-place archive and a package as before.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.5 from MetalMusings/ethercat (unchanged since 0.21): sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb. Then the file for your LinuxCNC:
- LinuxCNC 2.9 (2.9.4 or newer, as packages):
linuxcnc-ecat_*+lcnc2.9.debian13_amd64.deb(or.debian12). It replaces ecat 0.11. - LinuxCNC 2.10 run-in-place: unpack
ecat-rip-0.22-debian13.tar.gz(or-debian12) and run./install.ecat.sois built against LinuxCNC commit b0e7b191b9 (2.10.0~pre2-436); the installer refuses if your tree's HAL headers differ and says how to build from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.debwithout+lcnc2.9.
The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.22
LinuxCNC 2.9. Everything up to 0.21 (tuning, feedforward, EoE, serial channels, unused slaves, the DC reference clock, From lcec, the window) now works on 2.9 too. The component is written against 2.10's HAL API; on 2.9 one header gives that API on 2.9's pins, so there is no separate 2.9 code. Where the series differ:
- Integer pins have 32 bits on 2.9 (64 on 2.10). Commands are held within the PDO entry's range anyway, so entries up to 32 bits behave the same.
- LINT/ULINT entries and the encoder's 64-bit count are
floatpins on 2.9 (exact up to 2^53).ecatsetup --checkrefuseshal-type: sint/uinton a LINT/ULINT entry on 2.9 and proposesreal;extendis for 8- and 16-bit entries only. - The same
ethercat.yamlworks on both. - EtherCAT Setup finds the series by itself (HAL types, INI keys such as
HOME_SEARCH_DISTonly on 2.10, component names, LinuxCNC's pins).
ecat 0.11.x is closed. 2.9 users move to this package; 0.11 gets no more fixes.
Fixed: the 2.10 package lacked the shipped SYNC0 measurements (ecatsetup/measured/) since 0.20; the run-in-place archive had them.
Tried on 2.9.4: ecatsetup --check and the I/O test in the window on a plasma cutter's bus (three ECT60 drives and an I/O board): all slaves in OP, DC phased, two cycles from output to status, the drives without fault. Not tried on 2.9 yet: LinuxCNC itself with motion (next), 2.9.5 to 2.9.10 (same HAL API), Debian 12.
ecat ecat-v0.21
ecat for LinuxCNC 2.10. There are no 2.10 packages for Debian yet, so the main way to install is for a run-in-place LinuxCNC; there is a package too.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.5 from MetalMusings/ethercat (0.21 needs it for the DC reference clock, below): sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb.
- Run-in-place LinuxCNC: unpack
ecat-rip-0.21-debian13.tar.gz(or-debian12) and run./install. It putsecat.soin your tree'srtlibandecatsetupin~/.local/bin;./install --uninstalltakes it away.ecat.sois built against one LinuxCNC, commit b0e7b191b9 (2.10.0~pre2-436): the installer compares your tree's HAL headers with that build's and refuses on any difference (a different HAL API could crash a machine), and then says how to build ecat for your LinuxCNC from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.deb(debian12, debian13).
LinuxCNC 2.9 users take ecat-v0.11. The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.21
Tuning a servo drive in CSP, without LinuxCNC. In CSP the drive does the control; what LinuxCNC sees is the following error, and FERROR has to allow for it. EtherCAT Setup's new Response card (Test page) moves a drive out and back, records every servo cycle, and says what the error is made of: the bus's two cycles, a command filter in the drive, its position loop, the part that follows the acceleration, overshoot, an offset. Look for settings in the drive lists the drive's settings that may be the cause (from its quirk file, then by name in its own dictionary); Try it writes one to the drive's RAM with its own value kept, the next measurement is compared with the last, and Keep as setting / Undo end it. Last, FERROR and MIN_FERROR from the result. A guide: Tuning a servo drive, step by step; the start page has the task. On a plasma cutter's Rtelligent ECT60 drives this found 0x2009 "Filter Time" at its maximum (25.6 ms): with 2 ms the lag went from 17 to 4.6 ms and the error at 100 mm/s from 1.65 to 0.49 mm.
Feedforward from LinuxCNC (cia402.feedforward). For drives with 0x60B1 (velocity offset) and 0x60B2 (torque offset): the cia402 profile works them out from pos-cmd every cycle, in step with the target, 0 whenever the drive is not following in CSP and without a kick at enable. The Response card measures the torque per acceleration, tries the feedforward in the test and keeps it in the file; Apply writes the joint's velo-scale for it. The drive must be told to use them (a vendor setting). Tried against a simulated drive on a real bus (an EaserCAT 3000 acting as a CiA 402 drive) and in LinuxCNC with it: 40 mm moves at 100 mm/s, joint.0.f-error under 0.09 mm with feedforward, "joint 0 following error" without. Not tried on a real drive with these objects (the ECT60 has neither).
Behaviour change: the DC reference clock is the first slave with dc: (an assign-activate other than 0x0000), after ETG.1020: "the first device that should be synchronized is used as the reference clock". The master took the first DC-capable slave on the bus, also one the file leaves unused or does not have, and requested OP of it, which an unconfigured drive refuses (PREOP + ERROR in the log). Where the first DC-capable slave has no dc: (an EK1100, an I/O terminal), the clock moves to the first slave with dc:. Tried on two buses: three unused ECT60 with a DC slave after them (now the clock, no error), and an EK1100 + terminals + two ECT60 (the clock moved from the coupler to the first ECT60: all OP, two cycles, PLL error within 3 us). ecatsetup --check names the reference clock. Needs the master 1.5; with an older master the choice has no effect.
EtherCAT Setup, also
- Master page: Interrupts on CPU and Master threads on CPU (
irq-cpu,cpu), with the isolated CPUs named. - Measured SYNC0 results carry the slave's revision;
ecatsetup --device-datalists the quirk files and results, yours and the shipped ones, and what looks stale or wrong. - ESI files are kept parsed between runs (
~/.cache/ecatsetup): with Beckhoff's whole ESI folder the window opened in 53 s, now in under 2 s. - The drive test sets
velo-scalein CSP too, and writes its limits to a drive whose dictionary was not read (objects of 16 and 32 bits were taken as missing).
Not tried: the Response card's Try it on a drive other than the ECT60; feedforward on a real drive; homing and mode changes with feedforward on a bus; the DC reference clock change in LinuxCNC on turner's bus (halrun only, no motion).
ecat ecat-v0.20
ecat for LinuxCNC 2.10. There are no 2.10 packages for Debian yet, so the main way to install is for a run-in-place LinuxCNC; there is a package too.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.4 from MetalMusings/ethercat (it is what 0.19 and 0.20 were tried with; EoE and eoe.host-ip need it): sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb.
- Run-in-place LinuxCNC: unpack
ecat-rip-0.20-debian13.tar.gz(or-debian12) and run./install. It putsecat.soin your tree'srtlibandecatsetupin~/.local/bin;./install --uninstalltakes it away.ecat.sois built against one LinuxCNC, commit b0e7b191b9 (2.10.0~pre2-436): the installer compares your tree's HAL headers with that build's and refuses on any difference (a different HAL API could crash a machine), and then says how to build ecat for your LinuxCNC from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.deb(debian12, debian13).
LinuxCNC 2.9 users take ecat-v0.11. The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.20
Shorter servo periods, and what to expect. On the plasma cutter (three Rtelligent ECT60 drives, an EaserCAT 7000, an Intel N100 with an igc card on xdp-native) LinuxCNC ran at 250, 200, 150 and 125 us with the machine moving and a heavy CPU load (stress-ng, 10 minutes each at 150 and 125 us). 125 us is the shortest these drives take: at 100 us an ECT60 refuses SYNC0 (AL 0x0036) and the bus never reaches OP. What was seen, the limits and how to measure your own: Servo period and performance. One machine, one kind of drive: not a guarantee for yours. 1 ms stays the usual setting.
The component
master: irq-cpuandcpu. Pins the network card's interrupts (and the master's threads) to a CPU, normally the isolated one the servo thread runs on (irq-cpu: 3withisolcpus=3). By default they are handled on the CPUs that also run the GUI and the mouse, and at 250 us and shorter a lost cycle came now and then (it stops the drives, by design). With the interrupts pinned none was seen in the later runs, three runs of 3 to 10 minutes, so this is a strong hint, not proof.- The automatic SYNC0 shift's margins are a sixth of the servo period when that is under 50 us (from 300 us up as before). With 50 us each, a 125 us period left 25 us for the spread of the frames' send time and one late frame sent SYNC0 to the three-cycle position for the rest of the run.
unused: trueper slave. A slave that is on the bus but not used by LinuxCNC is compared with the bus at start and then left alone: no configuration, no pins, it stays in PREOP. (A slave that is not in the file at all was always treated so.) A slave that is in the file and does not reach OP still stops every drive: the working counter is for the whole domain.- Behaviour change:
ecat.0.all-opmeans that every slave of the file is in OP. It used the master's bus-wide state, so a slave that was only on the bus kept it false. Thestate-*pins still tell the whole bus. The event log's AL states follow the slaves of the file too.
EtherCAT Setup
- Test page, Synchronisation card: the
pll-errspread, the room for two cycles (what decided three cycles at 125 us), where the network card's interrupts are against the isolated CPU, and what was measured on the drives. Remember this period for the drives... keeps a good (or refused) run; the results shown on the slave's page come with the program (ECT60: 100 to 250 us) and your own go to~/.config/ecatsetup/measured/sync0.yaml. - Slave page: the SYNC0 cycle the slave gets (servo period times factor), Measured on the bus, a warning at or below a refused period, and Used by LinuxCNC.
- Docs: Measured SYNC0 cycles.
Not tried: a start-up burst of lost cycles (about 1.4 s, 3 s after the bus is activated) remains at 150 us and shorter; the machine is in E-stop then, but do not switch it on in the first seconds. Serial channels (EL6021) are still only tried over a real wire against a slave of our own, not on a terminal; EoE only on one slave of our own. Two EoE slaves on one bus, and the Remember this period button on a long real run, were not tried.
ecat ecat-v0.19
ecat for LinuxCNC 2.10. There are no 2.10 packages for Debian yet, so the main way to install is for a run-in-place LinuxCNC; there is a package too.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.4 (needed for EoE and eoe: {host-ip}; ecat 0.19 also runs on 1.1 to 1.3 without them) from MetalMusings/ethercat: sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb.
- Run-in-place LinuxCNC: unpack
ecat-rip-0.19-debian13.tar.gz(or-debian12) and run./install. It putsecat.soin your tree'srtlibandecatsetupin~/.local/bin;./install --uninstalltakes it away.ecat.sois built against one LinuxCNC, commit b0e7b191b9 (2.10.0~pre2-436): the installer compares your tree's HAL headers with that build's and refuses on any difference (a different HAL API could crash a machine), and then says how to build ecat for your LinuxCNC from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.deb(debian12, debian13).
LinuxCNC 2.9 users take ecat-v0.11. The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.19
Serial channels (EL6001, EL6002, EL6021, EL6022). A serial terminal's channel becomes a serial port on the computer (/dev/ttyECAT4) that mb2hal and Modbus VFDs use like /dev/ttyUSB0: serial: per slave, baud, frame, send continuous, the settings changed by the program on the port while running, status pins; EtherCAT Setup proposes the channel, has a Serial ports card with mb2hal's lines, and the HAL check warns when mb2hal's parity or data bits differ from the channel's. How-to: A Modbus VFD on a serial terminal. Not tried on a real terminal (we have none): built from Beckhoff's manual and tried over a real wire, mb2hal against a Modbus slave, 2400 to 115200 baud, with a slave of our own (an EaserCAT 3000) made to behave as an EL6021, in halrun. A capture from a real terminal (ecatsetup --capture) is welcome.
Ethernet over EtherCAT (EoE). eoe: per slave gives a slave that speaks EoE an IP address, and host-ip gives the computer an address on the slave's network device eoe0s<position>, so that a device's web page opens in a browser while LinuxCNC runs. EtherCAT Setup has a Network (EoE) card that proposes a free subnet. Needs master 1.4. How-to: A device's web page over EoE. Tried on one slave of our own (an EaserCAT 3000 with a web server): in halrun, and in LinuxCNC with traffic while a plasma cutter homed and jogged (7458 of 7458 pings answered, no events, no following errors); not on a bought device, nor with two EoE slaves on one bus.
EtherCAT Setup: the window's new cards above; --capture / Save capture... also records what a serial terminal does.
Everything else in 0.18 is unchanged; the component's cia402, encoder and analog functions are as they were.
ecat ecat-v0.18
ecat for LinuxCNC 2.10. There are no 2.10 packages for Debian yet, so the main way to install is for a run-in-place LinuxCNC; there is a package too.
Update from 0.17 or earlier if you use opmode-cmd, homing by the drive, or PP: see the first two points under The component.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.1 or newer (1.3 recommended) from MetalMusings/ethercat: sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb.
- Run-in-place LinuxCNC: unpack
ecat-rip-0.18-debian13.tar.gz(or-debian12) and run./install. It putsecat.soin your tree'srtlibandecatsetupin~/.local/bin;./install --uninstalltakes it away.ecat.sois built against one LinuxCNC, commit b0e7b191b9 (2.10.0~pre2-436): the installer compares your tree's HAL headers with that build's and refuses on any difference (a different HAL API could crash a machine), and then says how to build ecat for your LinuxCNC from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.deb(debian12, debian13).
LinuxCNC 2.9 users take ecat-v0.11. The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.18
The component
- CSP holds a fixed position while it waits for
pos-cmd. After a change into CSP (opmode-cmd, or homing by the drive), the cia402 profile wrote the drive's actual position as the target every cycle untilpos-cmdcame within 10 counts. A drive that still moved a little, or feeds the target's velocity forward, then ran on at that speed: an ECT60 crept on at 1.5 to 3 mm/s after a change from PV to CSP, without a fault. The position is now taken once and held. - PP needs
0x607Fonly where the drive has it. The file had to set0x607D:01,0x607D:02and0x607F:00formode: pp; the Rtelligent ECT60 has no0x607F, so PP could not be configured on it.0x607D:01and:02stay required, and the error now says what to add (type DINT, the lowest and highest position in counts). EtherCAT Setup warns, with the bus, when the drive has0x607Fand the file does not set it.
EtherCAT Setup
- HAL & LinuxCNC: the Joints card always shows the INI file in use, with a button to choose another one or read the same one again (also File > Machine INI file...). Before, it could only be changed on the Master page.
- Without an INI file, or one without
SERVO_PERIOD, the servo period is typed on the Master page, for trying slaves with the I/O test without a machine configuration. - The question before Write lists each file with what is done to it, and the plan's notes, such as a
.tclHALFILE turned into a.halone. - With
servo-thread-follows-ref-clock: falsethe window setsref-clock-sync-cyclesto 100 and warns above 100 (also--check): at 1000 the slaves' clock wandered against the servo thread (measured, see below). - The window is quicker while editing: the HAL & LinuxCNC page is built only while it is shown.
Tried at a machine (a plasma cutter with three ECT60 drives, LinuxCNC 2.10 run-in-place, QtPlasmaC; details in docs/development.md): cutting with the arc; sync0-shift-ns: auto while moving, also with a CPU load (1 ms); zero-at-start with homing by switches; halcmd stop and start with DC slaves; opmode-cmd between PV and CSP; PP from the window; PV as spindle.0 (M3, M4, S changes, a program, E-stop); DC clock variants (500 us, another reference slave, false with 1, 100 and 1000 sync cycles); the master's messages in the window; and the window's newer cards on a real screen. Not tried yet: the spindle encoder with a spindle running, modular devices, analog terminals, the EL7041, drives with several axes, sync0-shift-ns: auto at 500 us while moving.
Found on the way: the ECT60's velocity actual value (0x606C) is noise; for a PV spindle on it, derive spindle.N.speed-in from pos-fb.
ecat ecat-v0.17
ecat for LinuxCNC 2.10. There are no 2.10 packages for Debian yet, so the main way to install is for a run-in-place LinuxCNC; there is a package too.
Update from 0.12-0.16 if your computer has no LinuxCNC 2.9 installed: ecat.so of those versions loads only where LinuxCNC 2.9's /usr/lib/liblinuxcnchal.so.0 exists. Without it, loadrt ecat stops with dlopen: liblinuxcnchal.so.0: cannot open shared object file. 0.17 fixes this.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.1 or newer (1.2 recommended) from MetalMusings/ethercat: sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb.
- Run-in-place LinuxCNC: unpack
ecat-rip-0.17-debian13.tar.gz(or-debian12) and run./install. It putsecat.soin your tree'srtlibandecatsetupin~/.local/bin;./install --uninstalltakes it away.ecat.sois built against one LinuxCNC, commit b0e7b191b9 (2.10.0~pre2-436): the installer compares your tree's HAL headers with that build's and refuses on any difference (a different HAL API could crash a machine), and then says how to build ecat for your LinuxCNC from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.deb(debian12, debian13).
LinuxCNC 2.9 users take ecat-v0.11. The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.17
The component
ecat.sois no longer linked toliblinuxcnchal. That link, kept from lcec_uspace, pointed at the build machine's LinuxCNC tree, soecat.soloaded only where LinuxCNC 2.9 was installed too. It now takes HAL and RTAPI fromhal_lib.soandrtapi_app, as LinuxCNC's own modules do. Nothing else in the component changed.
EtherCAT Setup
- PDO assignment (a slave's Operation and sync card) shows the ESI file's predefined assignment the configuration uses, such as Standard 32 Bit (MDP 511) on an EL5101. Before, it always showed the default. Choosing another one proposes the slave again with it, after asking. The PDO and Sync boxes no longer take the width of their longest ESI name, which could push Propose again off a small screen.
- Inputs and outputs (HAL & LinuxCNC page): a pin that your HAL files connect, by hand or through the From lcec card, shows its signal and the other pins in grey, with file and line in the tooltip. Before, it looked unconnected. A pin chosen on From lcec is shown the same way before Apply writes it.
Tried: on turner (EK1100, EL1008, EL2008, EL5101, two ECT60; LinuxCNC 2.10 run-in-place, 2.9 removed), the machine's own lcec configuration was moved to ecat with From lcec and starts in LinuxCNC with the new ecat.so. Not tried yet on LinuxCNC 2.10: the spindle encoder with the spindle running, cutting with the arc, PV, zero-at-start, opmode-cmd, sync0-shift-ns: auto while moving, modular devices, analog terminals, the EL7041, drives with several axes.
ecat ecat-v0.16
ecat for LinuxCNC 2.10. There are no 2.10 packages for Debian yet, so the main way to install is for a run-in-place LinuxCNC; there is a package too.
Install
First the userspace master, linuxcnc-ethercat-uspace 1.1 or newer (1.2 recommended) from MetalMusings/ethercat: sudo apt install ./linuxcnc-ethercat-uspace_*_amd64.deb.
- Run-in-place LinuxCNC: unpack
ecat-rip-0.16-debian13.tar.gz(or-debian12) and run./install. It putsecat.soin your tree'srtlibandecatsetupin~/.local/bin;./install --uninstalltakes it away.ecat.sois built against one LinuxCNC, commit b0e7b191b9 (2.10.0~pre2-436): the installer compares your tree's HAL headers with that build's and refuses on any difference (a different HAL API could crash a machine), and then says how to build ecat for your LinuxCNC from source. - LinuxCNC 2.10 package:
linuxcnc-ecat_*.deb(debian12, debian13).
LinuxCNC 2.9 users take ecat-v0.11. The window, ecatsetup --gui, needs python3-pyqt5. To try ecat without disturbing your setup: getting started. See the README.
What is new in 0.16
The component (first change since 0.12)
- The master's own messages (its log, up to level info) go into the event log beside the events, such as "Slave seems not to support distributed clocks!" or an SDO abort while a slave is configured. The event log file is version 2; EtherCAT Setup reads both.
- The master's messages also go to
ecat's output: the terminal ofhalrun, the I/O test's log, or LinuxCNC's print file.
Distributed clocks: what a slave can do (docs)
- From four sources: the slave's EtherCAT chip, its SII, the ESI file, and CoE
0x1C32/0x1C33. The slave's page shows them under the Sync choice, which offers free run, the slave's own DC modes, and any AssignActivate. - Warnings, never prohibitions: DC on a slave whose chip has no DC clock (it stays in PREOP with no status code), an AssignActivate the slave does not describe, DC off on a slave that only lists DC modes, a servo period below the slave's minimum, and more. Also from
ecatsetup --check FILEwith a master running. - Each slave's messages from the master on the Slave status page and in
--diagnose; no CoE questions to slaves without CoE.
A capture of the bus, for a device the developer does not have (docs)
- Save capture... on the Slave status page, or
ecatsetup --capture: one file with every slave's identity, SII, object dictionary and values, PDO assignment, registers, the report, the ESI descriptions and, with tcpdump, a few seconds of frames. Take it while the I/O test runs to have the bus in OP. It only reads. ecatsetup --from-capture FILEruns the proposal,--diagnose,--checkand the window against a capture instead of the bus.--capture-cturns a capture into fake slaves for the component's C tests, with the process data of each cycle.
Documentation
- Troubleshooting a slave: five questions and the steps to answer them.
- New screenshots of the window, and
scripts/docs-screenshotsto take them again.
Also
- The I/O test removes the folders of earlier tests from
/tmp.
Tried: the DC checks on turner's bus (EK1100, EL1008, EL2008, EL5101, two ECT60); the master's messages there; captures of the plasma cutter's bus in PREOP and in OP (ecat in halrun, no enable), read back without the bus. Not tried yet on LinuxCNC 2.10: the new component in LinuxCNC itself (only in halrun), cutting with the arc, PV, zero-at-start, opmode-cmd, sync0-shift-ns: auto while moving, modular devices, analog terminals, the EL7041, drives with several axes.