Advanced Search

I'm sorry for taking so long each time - distracted lately.
Maybe you could compress your configuration folder so I can look at everything?
Because you have custom ui and handler files makes things hard to know what is what.

Are you comparing master to 2.9?
The docs you quoted are for master.
The sims in master work for me.

Of course, hence my question.
To have it work properly, you need separate inputs for all drives, then LinuxCNC shows what drive faulted.

Just for fun I thought I'd try to use modbus - specifically gs2_vfd - to control my mill spindle.  VFD is a GS3 from Automation Direct, currently using 0-10v.

After some config problems (tty permissions), the USB-RS-485 adapter and LCNC are sort-of talking to the VFD.

I'm getting 'failed to set register - communication timed out' errors, but I think I have a more serious problem...

The way I've set up my electrical system, when estop is asserted, all the drives and VFD have no mains power.  24vdc is present for all sensors, Mesa, and so forth, but no drive bus power.

Until I de-assert estop, the VFD is dead and can't communicate with LCNC - so LCNC shuts down due to a 'no response' error.  I got around that for testing by taking it out of estop and energizing the contactors (having bypassed some HAL safety logic) and then starting LCNC.

Which got me thinking... is there a way to delay LCNC trying to communicate with the VFD until after I de-assert estop?

I'm not willing to change my estop scheme to permit power to the VFD when estop is asserted.  And the VFD has no separate logic power input like some servo drives.

Putting the 'loadusr Wn...' in the postgui file didn't do the trick.  I don't take it out of estop until the GUI is up and stable.

There's no point in trying to troubleshoot the read/write unless I can force LCNC to wait until after I de-assert estop.  Anyone have a thought on this?
 

qaywsxedc wrote:

 

Thx for your answer. I tried the linked version...still the same error, no ping possible.
but I am on the stm32nvem... i tried the rt version because i wasn't sure, if you are going do build it for both. It is my fault, but it's a bit confusing with all the different models.
I flashed the stm Version1.0.0 again...this works fine.

I have a gantry machine and need two Y axes and an additional b axis..but with the basic config of the nvem, i always get machine errors. So i need to the full config, or i am wrong.

I tried to connect via ftp to upload manually, but no luck(on 10.10.10.10 via port 69)...host is not listening. and the upload script times out everytime.
When i connect I can move my basic axes without problems. Maybe you can compile a Version with full config on github.


 

Hi, yes you will need the full config.

You cannot manually upload the config file. You need to use the python script "upload_config.py" as the script inserts metadata and pads the JSON file to 4 byte boundaries.

aDm1N wrote:

cakeslob post=308763 userid=25269

scotta wrote:

I've also had no luck with an out of the box STLinkV2. But ColdTurkey has got it to work

pyocd.io/docs/debug_probes.html

"Recent STLink firmware versions will only allow access to STM32 targets. If you are using a target from a silicon vendor other than ST Micro, please use a different debug probe."

If you have a couple of probes you can always flash one with the following firmware.

github.com/scottalford75/Remora-RT1052/tree/main/CMSIS-DAP


ok, so pyocd, CMSIS-DAP firmware, stlink
give me some more notes, im trying to do a docs thing
github.com/cakeslob/Remora-docs/blob/mas.../Ethernet/rt1052.rst
 
I'm a bit confused about the pins for the encoder. PE pins were for the STM, as far as I know the pins on the RT1052 are labeled differently.

Hi, the following example config has the pin information that you are looking for.

github.com/scottalford75/Remora-RT1052-c...0-rt1052-DMA-ENC.txt
 

Seems you are not actually using the custom homing component 'el8_homecomp.comp'.

[edit]

see this earlier post:
forum.linuxcnc.org/ethercat/51830-marco-...rcat?start=20#302562

[edit 2]
Note that the commands in the above linked post look wrong to me:
'sudo homecomp --install el8_homecomp.comp'
should actually be:
for halcompile you need to install the 'linuxcnc-dev' package.
 

1. 'less /proc/bus/input/devices' is fine. If you want you can install 'lsusb' with:
2. If you see pins for your gamepad instead of your keyboard under pins/input/ then that would indicate that the key word you entered in the 'loadusr ...' command is not unique to your keyboard device.
You would have to post the output of 'less /proc/bus/input/devices' or 'lsusb' for better analysis.

  Thx for your answer. I tried the linked version...still the same error, no ping possible.
but I am on the stm32nvem... i tried the rt version because i wasn't sure, if you are going do build it for both. It is my fault, but it's a bit confusing with all the different models.
I flashed the stm Version1.0.0 again...this works fine.

I have a gantry machine and need two Y axes and an additional b axis..but with the basic config of the nvem, i always get machine errors. So i need to the full config, or i am wrong.

I tried to connect via ftp to upload manually, but no luck(on 10.10.10.10 via port 69)...host is not listening. and the upload script times out everytime.
When i connect I can move my basic axes without problems. Maybe you can compile a Version with full config on github.

 

Hi all,
Is the description of QtDragon HD at linuxcnc.org/docs/devel/html/de/gui/qtdragon.html actually up to date? For example, when I compare the settings of the post-GUI-hal file with the same file from github, there are a few different designations and pin names. When starting QtDragon-hd, for example, I get an error message when loading the line “net eoffset-state qtdragon.eoffset-is-active <= motion.eoffset-active”. The message in this case is: “qtdragon_hd_postgui.hal:91: Pin ‘qtdragon.eoffset-is-active’ does not exist”. If I comment out this line, QtDragon starts. However, I don't know if everything is loaded correctly and completely.

Can anyone help me with this?
Peter

 

One mux2 would be the one that selects the stepgen velocity command to come from
the PID output or the scaled spindle speed command.

The other mux2 would select the joint feedback from either the stepgen feedback
or the joint position command.

The trick is not to switch the second mux back to stepgen feedback until
the joint command and stepgen feedback are zeroed (ideally by
homing to index)

Guys, tell me how to set up a search for a home position with a torque limitation on mechanical stops as in the video by Marco Reps, onceloved   helped me a lot to get the configuration up and running, but I still have one problem
onceloved , Thank you very much for your help!

Please see my configuration, it starts and can control the motors! But I can't set up the home position detection with torque limitation
 
 
 

PCW wrote:

The joint feedback could be looped
back from the C joint commanded position this case. Finally, a homing
to index move would be done to re-synchronize the spindle position to
absolute C coordinates.

The question is how? As I understand there is no command like UNlink signal with pin.
I cannot use mux2 to switch source of pos-fb (that was my first idea), because if I have position-fb = 300 in stepgen AND joint AND pid, and then somehow stepgen position-fb became 0.0 (or source of this value became 0.0 and joint/pid have 300) -> joint error occur.

 

After checking everything. I think I found the culprit. Since I am using a quadrature encoder. The direction of it was reversed. I fixes that and it works. Thank you a lot for your help.

Normally, I use the vertical head, which bolts on to the Y axis. But If I remove that vertical head, there is a second SK40 spindle directly in the Y axis, facing straight forward, so in that configuration, Y become -Z and Z becomes Y.
Cheers,
Mark