Advanced Search

"-not" works for inputs, but not outputs
(without some fairly micky-mouse logic)

Hello all, I am rebuilding an old gantry type machine which I plan to use for cutting mostly with oxyfuel and eventually for thinner plates with the plasma source.  The machine is 10m long on Y (longitudinal axis) and about 5m wide on X (transversal axis).
I would like to use linuxcnc and I have bought many years ago the servo motors and drives from DMM Tech, more precisely the DYN4 
series. I have 2 motors for the Y axis which I plan to use in parallel (1kW each motor) and a smaller motor to move the torches on the X
axis (a 750W one). All motors have a single turn absolute encoder.

Now recently I have visited some exhibition and I saw today most of industrial things are on EtherCAT. I saw that LinuxCNC supports it 
in the latest versions, but since I have no experience with EtherCAT I would like to ask if I have to pay some special attention to certain things or there are no troubles at all with the compatibility ?

My plan is to make linuxcnc act as an master node, and send signals to the motion controller which will then through some adapters transform the signal to pulse/dir and send to the drives.

This is the general idea on how this should be working. 

Can I hear some comments from you guys , anybody have a similar setup , should I go some other way around the thing ?
Thanks for the time.

Make sure you test your kinematics for A=0 , C=0 first so you can confirm that the kinematic model returns proper results. Also make sure that the coordinate values switch between the same values when you change between trivial and RTCP kinematics.

have a look at the millturn simulation configuration:
github.com/LinuxCNC/linuxcnc/tree/master...xis/vismach/millturn

The limits are switched between two sets configured in the ini file.
See 'mcodes' for the switching scripts and the files in 'remap_subs' that actually call them.

Hello, I've encountered a similar problem. Where can I find an example of a milling machine with switchable soft limits?

Hello everyone, I'm a DIY enthusiast. My English isn't great, but I'll try my best to explain clearly. I'm encountering a problem while using LinuxCNC as the operating system for my machine tool. When I replace the existing kinematics system with my own custom-built five-axis kinematics system, the machine tool coordinate values ​​change when I switch to RTCP mode at the current position, causing a system alarm because the machine tool coordinate system has exceeded its limits. My understanding is that enabling RTCP changes the workpiece coordinate system values, not the machine tool coordinate system values. I haven't been able to solve this problem yet.

Additionally, my machine tool structure is A (head), C (table), with the distance from the C rotation center to the machine tool zero point being (Xrc, Yrc, Zrc). The tool length is set to dt, and there is an offset between the two rotary axes with offset values ​​of (Xos, Yos, Zos).

Please share zipped configuration folder, or whatever you are willing to share to shed light on the issue.

I'm not sure on this but I thought there was a change to support 64 bit floats for instanceslike this that managed the wrap but I have no experience. There is a big body of work pending in master branch around 64 bit pins. Message Grandixximo on Discord.

Alright, the 4th axis joint error inverting is 32 bit count roll over.

I am getting confusing info in my searches, but there seems like there is a hall function to manage this? Anyone have some insight as how you implement it?

thanks!

ok, the joint error is something else. it happens at 912.45 ALWAYS. it must and over run of something. we have like 800 million counts per turn of the axis, so its inverting at 2 billion encoder counts.

Ill put this in my other thread as it i dont think has to do with the ethercat jitter problem.

rodw wrote:

can try for now to do

loadrt lcec
...
initf lcec.activate servo-thread
addf lcec.read-all servo-thread

thats already in my hal

I did find something interesting. at least specifically on the 4th. when it give a joint error, the encoder position INVERTS, as does the display. And it STAYS inverted. so ill be jogging at 900 degrees on the A... and then joint error and it become -900 degrees.

I dont know if the other axis joint errors are also inverting, i cant "see" them yet. If they are they are returning back to position the next frame. The rotary axis for some reason doesn't return, it just flips sign.

thanks, ill look it all over in the morning.

abbasezzi wrote:

i am controlling drive by using modbus 485, my controller is teensy 4.1 with wavetec RS485 module between drive and controller. the drive is being controlled in PR/velocity mode.. the PR mode is being used for homing and velocity for normal 360 degree rotations. i am facing issue in the end when the arm reaches to zero/home position.
 

Update your machine to use LinuxCNC and we might be able to help.

ihavenofish wrote:

4th axis!

So, in redoing linuxcnc and adding this new leadshine L6 servo, ethercat is not totally ferked. I am getting random limit and joint errors during program runs, during jogging, sitting idle. I managed to see one of these errors with the scope and it is a one frame spike of ferror that's seemingly infinite value.

What the heck is going on. Ill upload my config files shortly. Every camera cut in the video is a random joint error on a random axis.

I have the jitter fix build, and not a whooole lot else really changed from the last build. HALP!

there have been big changes.
forum.linuxcnc.org/ethercat/45336-etherc...-how-to-step-by-step
follow the updated process and also the added Hal file command mentioned in response to your other post
 

Also the Ethercat repos have changed to point to linuxcnc-ethercat
The instructions in my Ethercat sticky have been updated so also follow them
2.9.10 ISO should install the new repos mentioned but follow the  instructions in step 2
forum.linuxcnc.org/ethercat/45336-etherc...-how-to-step-by-step