Advanced Search

pgf wrote:

Another good perspective. You're clearly a LinuxCNC guru. How much time have you spent using Grbl or FluidNC?

You mention no need for coding skills: I'd say that any time you open up a HAL file, you're coding, designing arbitrarily complex digital "circuits" in a text file. An incredibly powerful feature, but where's the UI? It's a graphical concept, one that should be supported via a schematic capture tool. But there's no easy way to even display a HAL file, let alone construct it.

I'm not a guru, I've just made enough mistakes and been frustrated enough and most of the time it's a stupid typo or not interpreting the docs correctly, my issue, not anyone else's. Even with the tools we have for creating the amd64 & RPi4 & 5 images I still mess up when creating images (RPi4 & 5 are the ones I work, it takes a couple of iterations), and it's not a 5 minute job waiting for the tools to do their job.

Pen or pencil if you can't visualise in your head, or you could use a CAD package, although they're not really that complex. There was an attempt to do something with Eagle a few years back. I'd disagree, it's just a bit of configuration. Look at logically.....many of us when we first made our entry into electronics there was no readily available software for designing circuits, like the constipated mathematician we "worked it out with a pencil". I still use pencils, pens and notebooks, sometimes it's just quicker, every so often I have to wipe down my desk to make room for new notes.

"A little respect", how about taking a bit of constructive criticism, to tell the truth the comment wasn't particular to you, it was a general thing. Then again I can't be held responsible for how you take things. And if you're having trouble it's easy enough to ask for help.

Hello everyone,

i have completed the Ethercat installation according to rodw
forum.linuxcnc.org/ethercat/45336-etherc...-how-to-step-by-step

The installation went smoothly and the final test was positive:Component Pins:
Owner   Type  Dir         Value  Name
     4  u32   OUT    0x00000001  lcec.conf.master-count
     4  u32   OUT    0x00000004  lcec.conf.slave-count

The following version of Linuxcnc is installed:
6.1.0-30-rt-amd64 #1 SMP PREEMPT_RT Debian 6.1.124-1 (2025-01-12) x86_64 GNU/Linux

When I check if /sys/kernel/realtime exists, I get /sys/kernel/realtime
When I check the module cia 402 (halcmd loadrt cia402 count=1), I get Using POSIX realtime

Am I wrong as a layman if I assume that this is not real real time and that there may be problems with some modules?

I have connected everything and flashed the bitfile from PCW. But it looks i have not configured the second card in the hal properly.
Attached the output of mesaflash readhmi, the hal and the crash report from lcnc
any ideas?
thanks in advance.

I have a lot of mechanical work on the machine. Cutting and welding on the housing to get it smaller. maybe i will post pictures later if you are interested.

The plan for the drives at the moment is to use the SIEMENS Drivers if i can manage them to run... i think at least the Step Dir Config via SMC30 should be possible. But i would love to use ethterCAT like in this example.

support.industry.siemens.com/cs/document...p-pn-?dti=0&lc=en-WW

 

I have some experience with Mesa. After learning LCNC and help of great people like Talla83 and many more, my Deckel FP3NC runs really nice with it. 


But i am a complete NOOB with EtherCAT. And to be honest, i am a little bit afraid of it.
So i hesitate to buy a used few hundred euro PCB for something i have no glue of

"Not a Linuxcnc issue, but a user issue."

A little respect, please. There were fundamental changes, and a small attempt at automating the upgrade. I did read the docs, and I did make the manual changes I thought were required, and it did work... most of the time, and then it would stop working. I think maybe I'm trying to jog the jog the joints instead of the axes. Or else maybe that's fine, and the problem lies elsewhere. But either way, I have a debugging problem on my hands, and after a couple of hours, I'm still not done.

Another good perspective. You're clearly a LinuxCNC guru. How much time have you spent using Grbl or FluidNC?

You mention no need for coding skills: I'd say that any time you open up a HAL file, you're coding, designing arbitrarily complex digital "circuits" in a text file. An incredibly powerful feature, but where's the UI? It's a graphical concept, one that should be supported via a schematic capture tool. But there's no easy way to even display a HAL file, let alone construct it.

pgf wrote:

Thank you! Your's was exactly the voice of experience I was hoping to hear from.

I love/hate LinuxCNC. On the one hand, I spent hours last night trying to figure out how to get my USB gamepad "wand", which worked fine under 2.7.4, to work reliably under 2.9. On the other hand, it's likely that Grbl/FluidNC wouldn't support the gamepad at all.

I've wondered, too, about whether the Grbl comm protocol could be implemented on the front of LinuxCNC. I know the folks driving industrial mills would be horrified, but it might be a good solution in some cases.

Again, thanks for the feedback.

There were major changes to jogging from 2.7.x to 2.8, so any pendant would need work, that's why one should read the docs to see what has changed before upgrading. Once a again, not a Linuxcnc issue as such but a user issue.

If you're going to use a CM5 & carrier board you'd be crazy not to use an NVMe drive. It's fast, like really fast. I use one with mine Rpi5.

Hallo,

Kurz zu mir, habe mich heute hier bei euch regestriert weil ich nicht weiter komme und glaube noch öfter hilfe brauch.
Bin der Tom 35Jahre ich habe minimal Linux vorerfahrung, hab ein Prusa MK4 und eine Bernado 550 Dreh Fräs kombi mit nachgerüsteten DRO
Habe die Zahnräder zur Zug-leitspindel hin ausgebaut und will die jetzt mit nen nema24 - cl57t - mesa 7i96s - ace minicomputer - joystick oder controller/win11rdp steuern.

Hab alles fertig angeschlossen! Netzwerk zu Mesa eingerichtet jetzt kann ich die karte pingen mit ca. 170-200ms
Gehe jetzt in den pncconf wizard und mach nur das nötigste xz lathe der Motor test der X-achse funktioniert Motor dreht sich!
Speicher alles + desktop verknüpfung!

Wenn ich jetzt die Verknüpfung startet kann ich die Mesa karte nicht mehr pingen...
Karte leuchtet 4xGrün 1xGelb(der obere)
Jetzt kann ich den Motor 5min lang bewegen...
So dann nach ca. 5min leuchtet jetzt der unter dem gelben auf der karte Rot und ich bekomme im program unten rechts die fehlermeldung:
hm2/hm2_7i96s_0: error finishing read! iter=1256216
Jetzt kann ich auch den Motor net mehr bewegen.

Hab natürlich schonmal gegoogelt anderer Kernel macht mir schon zu viel probleme bei der Installation.
Gibt ja hier bei euch nen thread mit 34 seiten, wo ich echt nich mehr durch blicke...
Muss doch nen Fix geben weil es ja anscheinend erstmal geht^^

Möchte der effiezens wegen noch ne 2 Frage mit zu geben
Ich lass den originalen Schlosskasten mit zug/Leitspindel drin bedeutet das ich den schrittmotor einmal im dauerbetrieb
eine richtung fahren lasse(wenn ich nur den automatischen vorschub brauche) und einmal ganz normal mit eingelegter
schlossschraube zum gewinde schneiden...
Ist das möglich? Weil die position nach manueller arbeit ja sozusagen nicht mehr stimmt...
Kann ich LinuxCNC die Position vom DRO geben bzw. das DRO als endschalter benutzen?

Vielen Dank

Gruß Tom

 

I don't know where this idea comes from that you need any coding skills to setup Linuxcnc. This is one of the biggest lies attributed to Linuxcnc (Linux in general), it's bullshyte. You don't need to be a coder, even to type instructions at a terminal, no one ever said you need to be a coder to use MSDOS. That was the point of a OS to enable the system to be used by "non computer people" as well as a lot of abstraction & a million other things.

Setting up Linuxcnc for a basic 3 axis mill take no time at all using the Linuxcnc supplied config tools. Whether it's a parallel port or 7i92 based system. The biggest issue I see users have is not paying attention to the errors reported in the error report, things like pin xx doesn't exist, usually it's a typo. Or a user expecting a pin to exist that doesn't. Realtime errors are a PC issue, not really a Linuxcnc config issue.

Whether you axis, gmoccapy or your "grbl module" (whic I think will be a performnce hit in a bad way) you will still need to setup a linuxcnc config, the very thing you've been mentioning.

If you want to use grbl use it, but Linuxcnc is a completely different beast, but don't expect any grbl functionality from Linuxcnc.

Mesa, LlinuxcncRIO & Remora all work along a similar concept, step & direction signal generation is off loaded to the external motion control card.

I don't know why you would want to send a gcode file line by line to linuxcnc when it can open and read the file directly.

One thing that is very intriguing, you admit to poor coding skills, yet propose some major replacement of the Linuxcnc internal workings.

In reality all grbl is is a way to send data via a serial port.

Thank you! Your's was exactly the voice of experience I was hoping to hear from.

I love/hate LinuxCNC. On the one hand, I spent hours last night trying to figure out how to get my USB gamepad "wand", which worked fine under 2.7.4, to work reliably under 2.9. On the other hand, it's likely that Grbl/FluidNC wouldn't support the gamepad at all.

I've wondered, too, about whether the Grbl comm protocol could be implemented on the front of LinuxCNC. I know the folks driving industrial mills would be horrified, but it might be a good solution in some cases.

Again, thanks for the feedback.

I have a LinuxCNC image running on a CM4/eMMC, so it should also work on the Pi5.

Hello
I put my pi 5 into operation yesterday.
Unfortunately I get the locale error with Pconf

I tried to use the same settings as you.
My keyboard is german

Do you have an idea?

I have added the missing part to the INI file and change basic_probe.py, probe_routines.py and probe_subprog.py to the original files from v2.9 from github.

After performing additional measurements, it appears that the issue of bad synchronization between the measurement process and data updating in the widget mainly occurs when the feedrate value changes during probing. If the value remains stable throughout the measurement, the probe measurement cycle completes correctly.

The video shows four consecutive measurements:

• The first measurement was completed without issues.
• In the second measurement –> at 37 seconds in the video, the feedrate oscillate from 52% to 51% and back during the probe measurement, after which the value in the table for the Z axis changed. However, the measurement had not yet been completed.
• The third measurement –> similarly to the second, at 1:06 the feedrate oscillate from 40% to 41% (a very quick oscillation), after which the value on the axis Z changed.
• The fourth measurement was completed without issues (no feedrate oscillation).

Do you have any idea how a change in feedrate could affect the value update in the widget?

The Radxa Zero 3E is a nice small SBC very similar to the Pi Zero but available with more RAM and an ethernet connector. You don't want to use it for step generation, but with the latest Armbian running a 6.12 RT kernel, the latency is ok, especially for a headless motion controller running a servo-thread. Could be interesting for example in a OpenPNP setup. One problem is that USB is not working with current kernel, but this will probably be fixed soon.

Yesterday I ran a G1F1X100 command with 6 TMC5160 connected over SPI @ 6MHz and spidev driver; the servo-thread (reading all 6 position, writing all 6 speed, only one PID) stayed under 0.75 ms for the entire 100 minutes.