Advanced Search

Also sometimes users want something that doesn’t work within Linuxcnc’s model.
Mostly this will boil down to USB motion cards, usually ones the Mach supports and using Linuxcnc to send gcode to an external board. Most are accepting this is not supported and not in the pipe line, once every often someone will spit the dummy and throw all their toys out of the pram.
For other support channels you may want to jump over to the Linuxcnc homepage and see what the other options are and links to said options.

Oh I forgot, development is very centralised via github. It’s a very mature project which can trace things back to a NIST proposal quite a few years ago.

Any issues related to the Linuxcnc code is well reported thru the github channels.

Lists and the like on the forum don’t really work as, I mentioned before, users think that their problem is unique to themselves.
We started a thread a while back to gather issues related to Parallel Port cards. Every so often within a month 3 to 4 users will start a new thread with the same issues.
Apart from not having a SPI driver for the RPi5, due to the RP1 chip (which caused issues for many utilities in general use for accessing gpio on the RPi 4 and below) there is no real issue.
There was a big push at the end of 2023 to update the existing RPi images and create one for the RPi5, Rod worked hard on that and bounced a few ideas around with myself. Now it’s at a stage where it is quite mature and very usable. The only real “issue” is that the RPi imager utility cant be used with it for reasons discussed in this thread with the solution.
So far the hm2_spix driver works quite well and the gent who wrote it made some code changes to the original hm2_rpspi driver.
Once new packages are built I’m sure the RPI5 will be fully supported “out of the box” for it does require building Linuxcnc from the current code base for either the master branch or the 2.9 branch. Not a difficult task if one follows the instructions as set out in the docs.
Every cnc machine is different so one can not hope to have a plug and play solution unless one buys a cnc machine that has been preconfigured by the manufacturer.
There are many boards available from Mesa that people use for a variety of applications. On top of that there are a number or alternative projects Remora and Linuxcnc-RIO are just 2.
Myself I’ve taken a different path by using a Spartan 6 dev board, designing some boards that take care of level shifting and built firmware to suit using the Mesa sources, for some of the firmware I’ve tweaked the existing drivers to suit the IO config. Using the same hardware I can also use firmware from the Linuxcnc-RIO project.
There’s a lot of choice available.

Hello tommylight,

That Google excerpt could be way off. I don't know. The only way to know what the current state of (LinuxCNC/Pi 5) is to poll it users and those attempting to use it. As you say they may have a working system and be doing something else.

Your comment about "Set it once and never again" is true for a lot of software until you change something (like a controller or the software itself).

I know newbies don't read things. For me "usability" includes reducing the need to read stuff as much as possible.

A issue list/process has several attributes besides being a pain in the neck. One is to track issues that surface and ensure that the issue is fixed and available for download. Another is for the developers to assess how to improve the user's encounter with the hardware and software itself.

I offered to help. If there is no issue list (for PI 5 for now) maybe I could help to create one. It would help centralize some aspects of the development work. Also, people age out. It may allow the current crop of wizards help the next generation of users.

Thanks,
Tom

Agree with Tommy.
Users skipping steps.
Users not typing in commands properly.
Users not relaying requested information.
Users blaming Linuxcnc when it’s a mechanical issue.
Users blaming Linuxcnc when it’s an electrical issue.
Users not bothering to look at the docs.
Cases of users editing config files with a word processor program.
Users not understanding the prepared Linuxcnc images are not meant to be a general purpose OS.
Users believing the fear mongers who will say you need to be a programmer.
Users thinking they are the first and only ones to have a particular issue.
Users who don’t know how to form a proper question.
The effing “Maker Community” that tells poor saps that a CNC machine is easy to use by anyone.
People that wear gloves near rotating machinery.
People that insist on soldering crimped connectors.
Ah fuck it just people in general.

Well that google sucks, hard.
There are no "instability issues" ever, you set it properly once and it keeps on working for years and years without issues, ever!
"steep learning curve"??? Yes of course, for someone with no prior experience, but so is tying shoes when we are young.
"compared to commercial CNC controls", i have no clue what they are comparing it to as from extensive experience i concluded the opposite, although some do come close to actually pointing issues in the right direction, some will absolutely make you take a hammer and destroy the thing.
"many users report frustration with the complexity of setting up a reliable system" Oh YES! This is absolutely true due to reasons:
1. They do not follow instructions! Read some random topics here, like setting up networking for Mesa cards, or servo tuning, most people will read and configure things and move on, but there are some that skip steps and come back complaining nothing works.
2. those who managed to set things up are not here complaining, they are making chips and using their machines, and most do not bother registering here, so google can not include those, and that is the majority of LinuxCNC users.
-
In short, LinuxCNC is extremely reliable, very easy to set up for normal machines thanks to several wizards that are included with it, but there is NOTHING easy for retrofitting a 15 Ton industrial machine without prior experience in electronics, and yet, there are plenty of members here who did just that with no experience at all, they just followed instructions from very experienced and capable members of this forum.

Hello cornholio,

Sorry for asking questions bass-ackwards.

I have the need to build a real boat but honestly I don't want to cut it out by hand so, I have been thinking of building a CNC machine.

Doing some research I found and looked at LinuxCNC for part of the software chain but the references I saw were pretty negative (see a citation below from Google). I was not sure of the veracity of the citation(s) but I really want to pursue the CNC route so I really needed LinuxCNC.

I looked at the forum and it seemed there have been persistent problems with Pi 5. Given my background I offered my help getting it to be more stable (forum.linuxcnc.org/9-installing-linuxcnc...dux?start=260#317739).

So I really like the time I do Linux and driver/kernel work and, I really want to build a boat and I would love to help here.

Tom

----

From Googling "linuxcnc reviews and complaints":

Reviews of LinuxCNC often highlight its powerful customization capabilities and open-source nature, but also point to a steep learning curve, potential instability issues, and the need for significant technical knowledge to properly configure and troubleshoot, especially when compared to commercial CNC controls; many users report frustration with the complexity of setting up a reliable system, especially for beginners, and concerns about real-time performance depending on the hardware used. 

Tom, if you had of set out your goals from the beginning, without having to play 20 questions, you would have had an answer in far far less time.

[rant]
This is what makes me somewhat miffed on the forum, getting information out of people to provide a concise answer is like pulling teeth. It's like everything is a state secret. Have people forgotten the art of asking a question ?
[/rant]

Hi rodw,

SheetCAM looks like it gets the nod. Many thanks.

It does have you start by scanning the plans on a (big) scanner and into an image file (.JPG, etc.). For that, maybe I can buy time on a scanner. I do own the plans and the scan would only be for my use.

On top of buying time on a cutting machine maybe I can get my boat realized (and keeping me busy) without actually having a CNC machine.

To reiterate my very first post: I have a fair amount of *nix/device driver and kernel experience and if a need comes up let me know how I might help.

Thanks,
Tom

TomAlborough wrote:

Maybe so, maybe so.

But part of me wants to use a Pi 5 for *something*.
 

You can only accomplish your object in life by complete disregard of the opinions of other people.
Aleister Crowley

Have a look at Sheetcam for all of  your CAM, nesting and scanning with its scananything addon.

Maybe so, maybe so.

But part of me wants to use a Pi 5 for *something*.

It would take the pleasure out of building and just turn it into a very tedious task.

I'm not sure I could make an image/picture that was not materially distorted but I will look into ways to do it.

I would like to run an option past you:

-- Tape the plans to the bed of the rig.

-- Replace the cutter/lens with a stylus.

-- I just saw a "CAM USB camera" online that gives you a view of the exact place where the (stylus) hovers over the plans. (I can not find it online right now...)

-- Then:

    -- Jog the steppers with a joystick to place the stylus over the "start" of the tracing

    -- Touch a button on the joystick to register the current location on the plans/bed

    -- Jog the steppers again to get to another point on the plans and press the button

Observations:

  -- Straight lines would be easy; curves would require help digitizing gracefully, but I have some thoughts

  -- The file of points would have to processed into a file of "cuts".

  -- Plans involving multiple parts would have to be digitized a part at a time. The parts would then need to be laid out on (a 4 x 8 sheet of stock say).

Thoughts?

Thanks,
Tom
  ​​​​

Much respect for the traditional way but I would like to come up with a 21'st century option. Digitizing from the image sounds good. I'll look into it.

Thanks,
Tom