Advanced Search

Have you tried adding '(AXIS,stop) ' to the top of your gcode?
This should prevent the preview from loading and might help with very large programs.

I would think that leaving it step/dir would be the easiest path
and would require few wiring changes. It also would not require
changing the drive setup, tuning the drives or tuning LinuxCNC's
feedback loop.

The only real advantage to an analog setup is that you can disable
the drives without losing position.

I have a gcode program much larger than anything I've run before. It has about 800k lines.
I'm on Linux Mint 20, using linuxcnc 2.8.4-76-g6082f1df7 and axis.
When I run the gcode, I have several issues:
1. The preview window in axis is garbage (see axis_screen.png).
2. When I start the program, I get a popup error (see first_error.png)
3. I tell it to 'run anyway' and get joint errors (see second_error.png), specifically the X axis on line 12. In the first image, you can see line 12; nothing odd there.

I searched for any x or y command in the entire file that has more than one digit in front of the decimal place (X19.0 would be found, X9.0 would not). Nada.

I loaded a 50 line gcode program I use for surfacing. It is configured to surface the same size piece of material as the very large program. The surfacing program runs with no problems.

Please advise,
Steve

I've got a 5' x 10' CNC router with a failed Mach 3 control. It has Yaskawa Servos (SGMGH-09ACA61) and was set up for step/dir with an ethernet smooth stepper. It was "open loop" with that control. It has four motors for the 3 main axis and a 5th motor for a rotary axis. I might not use the 5th axis but I do have a crazy idea for an automated dust collection "axis" that would probably use stepper motors. 

I have a Mesa 7I76e that I bought several years ago for another machine that has yet to be completed and I think this would be my fastest/easiest control to drop in but would still leave me open loop to the control. Instead I've been thinking of running with analog servo signals and the encoder feedback into LinuxCNC using a 7I92TF with a 7I77 or a single 7I97T. It seems the 7I97T might be a bit light on general I/O since the machine has an ATC spindle and several other pneumatics that need to be controlled and sensed so I'm leaning towards the former combo. 

Any suggestions wrt to Mesa card availability and the advantage of using analog vs step/dir in this application? 

 

1. I'm using Tormach's firmware for the 770 configuration which has the following for the spindle:

#    Mill Function   Pin#    I/O     Pri. func    Sec. func    Chan      Pin func        Pin Dir
#
#    Spindle Speed    1        0     IOPort       StepGen       4        Step/Table1     (Out)

2. By default the PathPilot uses stepgen mode 2. I think I'm confused as I read they use PWM and I need a step/dir which made me think I needed to use stepgen mode 0. I'll switch back to mode 2. These are the other default values:

position-scale 1.0
maxaccel 0
stepspace 0
steplen 10000

I'll change the steplength and step space to 1000 and position-scale to 7.163. Does maxaccel matter? The PathPilot hal also appears to be missing dirsetup, dirhold and maxvel which were included in the hal generated by Pncconf.
 

Chris thank you, you've answered all of my questions and suspicions.


 

1. What firmware are you using on the 7I92?
(suggest C11, not C11G) as it has a step only output
on pin 14.

2. Typically you would use stepgen mode 2 for this type of application
(since it generates a square wave). In addition you need to set the
steplength and step space to small enough values that they do not
interfere with the maximum frequency (say 1000 each)

If the full scale RPM is 3490 RPM, the step scale would be set to 25000/3490
= ~7.163

Hi @denhen89
did you manage to close the topic to the end?

Could you share the final hal file?

I started my lathe and a great option would be to add 2 mpg

received shipment of one of these: www.ebay.com/itm/235593572193
it is a Modbus K-type thermo-couple device, with read-type 04, address 1, for temp C in format "270" for value "27.0C".
my plan is to attach the thermo-couple to the spindle to monitor its temperature.

i also plan on getting some spools of waxed string so i can do some old-fashioned wire-lacing on the machine to bundle wires and air-lines. it just looks neat, and i rarely change the on-machine wires now.

5 should mean that clock is synced every 5 cycles. But its better to have first slave with refclock and -1 will let the first slave manage the clock referencing.

Grotius wrote:

@Endian,

Can you implement a cycle counting component that counts up  4.
Then read in the canopen values, in other words, open the gate to the canopen every 4 cycles?

@ Rod,
I installed the linuxcnc iso. But udev rules are not working when set, also reload rules.
So bus is not starting after reboot.
I now have to start the bus invoking sudo after reboot.
I don't get it, where it's goes wrong.






 

Yes for sure it is solution and typing custom component should be necessary.. but in thia sitution you have to know a actual timing setup of el6751 canopen master and mirror it in to to component input aetup by the user... But no every user are able to know everything and for this reason I want to did something poka oke stuff for elimination timing setup from user directly...

I have to dig lower in this rabbit hole ... It has to be able to cutomize it over SDOs because there are mapping and custom timing setup in the twincat3

If there will be 1ms period synchro at canopen master as servo thread of the lcec... complication with custom timing is done by hardware itself i think

In the future if you do need to use lspci, also use sudo in front as it also shows the loaded kernel modules , in your case it says access denied where the module name should be.

Hi Rodw, Thanks for all the incredible work so far on EtherCAT. I'm diving into both LinuxCNC and EtherCAT right now and excited by the possibilities. 

I ran into the same issues as discussed.  I tried your script on the Rpi 5 and it seemed to connect to their servers but it only seems to have grabbed 733 bytes. I may try some of the other methods mentioned previously.

 

___
System: Raspberry pi 5 + Leadshine EL8 EtherCAT drives
LinuxCNC level: first timer
Linux level: noob noob

I think there is actually a way to run linuxcnc using only forward or inverse kinematics using the 'kinstype=' parameter but I have not experimented with that at all:

See 'trivkins - generalized trivial kinematics' section:
linuxcnc.org/docs/stable/html/man/man9/kins.9.html

I'm not even sure if this would have any meaning in a non-trivial kinematic.

 thank you very much.