Advanced Search

Hi Friends, 

I've spent the last few months modifying a chinese cnc with a mesa 7i96s and a custom cutting head I've made with the end goal making it a tangential cutter  I've long dreaded this part because i simply do not know where to start. I am working on tuning the error in the cutting motor but this is next on the list.  I am going to lay out what i would like the machine to do, and hopefully you guys can point me in the right direction. 

Attached is a cad screenshot with a visual of the cutting head. There are two pneumatic cylinders that can deploy the cutting head down or up with a solenoid valve controlling them both. I wanted to leverage some of the 7i96s auxiliary outputs or the torch up/down to control this valve. 

I am not sure what cad program that will be outputting these, probably fusion or solidworks in the long run so i am leaning towards a filter file or i guess this is technically a post processor, as mentioned in this post, but i didn't see really any progress so here i am. 

forum.linuxcnc.org/38-general-linuxcnc-q...n-of-travel-possible

My fundamental issue is i don't understand what node is doing what. I haven't ever uploaded a file to LCNC but i imagine its Gcode. 
So my simple brain wants to understand;
- Cad program - > outputs generic 2D gcode
- **NEW FILTER FILE** -> magic happens here?
-LCNC - > doesn't know the difference, its just following orders. 




Requirements (will probably grow):
-be able to set "cut depth' (the z position with 'torch down')
-set cutting motor to only find angle in one direction (not sure if i can or should do this on LCNC and not the filter file) because belts aren't very good pushers.. 
-I'd like all travel movements and find next angle to happen with the torch up, and it only go torch down only when the next angle is good and its in position
-BUT i would like arcs and angle to be smooth, i need to differentiate between lines with arcs an endpoints. 


also sorry i am rambling, can you tell i've thought about this. 
 

What did I do ?

Hi Rod,

The Cia-402 i have seen the source code. It does not much.
It sends and recieves position, velocity, torque  trough a can protocol?

Hi Tia,

when i change to tpmode=scurve
It is : tpmod_scurve

And check if you use values for the trajectory :  vel + acc.
[/code]

I just isnstalled debian 13 on the desktop pc and installed the same software, but the ethercat-master from the codeberg repo.
It also runs ok. So that's a good sign.


 

Today I ran some more tests. I can now reproduce the issue described in the other thread with 100% consistency. I noticed that when the Z-axis is moving, G96  doesn't seem to be working. I recorded a video showing that the spindle speed doesn't change at all during the movement.



PS. I think there might be something wrong with the HAL configuration. I just noticed that when starting LinuxCNC, I need to go into the calibration settings and re-save the PID values for the spindle (I enter test). If I don’t do that, the PID control is soft– under higher spindle load, it triggers a hold  .Also, if I leave LinuxCNC running idle (without running a program) for about an hour, and then start program execution again, the axis movement speed during a G1 command is once again higher than what was set.

That would be lovely - the CM5 has one but then you need a carrier board and I don't know if the LinuxCNC image would run on the CM5/eMMC? With the right carrier board would be perfect though...

Dunn why RPi doesn't add a connector for an external antenna.

Just for completeness and to save anyone having to mess around with compiling kernel modules, here is what looks like a fairly comprehensive (at least until 6.13) list of in-kernel USB WiFi devices: github.com/morrownr/USB-WiFi/blob/main/h...in-kernel_drivers.md

Discovered it a little late but I think only one has an RP-SMA connection.

tried here, it works with the original trajetory planer on the rip install, when i change to tpmode=scurve i need to desable homing on the gantry axis and i can do jog movements but when running gcode the machine seens to be stoped but it is runing with an extremaly low aceleration, near 1 minut to move 1mm

net estop-loopback => hm2_7i97.0.ssr.00.out-00

should probably be:
net estop-loopout => hm2_7i97.0.ssr.00.out-00
 

- Solved -

There is a startup sequence in control word,
set bit number 1 2, 0 to 1
then if you set bit 3, drive goes to enabled state,

Thanks a lot

Great! Don't worry about the cia402.comp as it needs work to work with the homecomp features. In fact it needs rewriting  from scratch I think...

Thank you mr Hakan,
I added them,

unknown wrote:

As the image is a custom one, only the kernel is built from RPi Sources.
This pretty clear, I would have thought.

To one well versed in building distributions, yes, but not necessarily to one who has only used distributions! Not until now, anyway

unknown wrote:

Where as you were attempting to install headers from the Debian repo

 
I was but at that point, didn't realise the significance of the kernel being sourced from a different place to the base distribution.

unknown wrote:

The kernel should load the drivers without any user intervention.

 
It does now load on boot.

Thank you.

Hi Tia,

It seems the .img won't work ok.

Today i tried the scurve + ethercat on a desktop pc and it runned ok.
It's using a EK1100 + 3x EL2124 on a stepgen component.
Jog speed was around 15000 mm/min wich i find not optimal, but ok for ethercat.

The config:
1x linuxcnc iso from : linuxcnc.org/downloads/
1x ethercat installation : ethercat install
1x linuxcnc + scurve, using makefile build system : codeberg.org/skynet/linuxcnc
1x linuxcnc-ethercat, codeberg.org/skynet/linuxcnc-ethercat.git

Running the config : config link

Now i can not try the cia-402 component myself as i dont have the hardware for it.
But maybe you are able to try this again.