Advanced Search

This link into one of my videos explains it, Simple networking setup. Its not plug and play

rhscdn wrote:

Looks fantastic. Its great seeing more UI options beyond Qt. Just curious, is the display optimized for touch or keyboard input?

Being more of a C/C++ person, I’ve been building a similar pet project focused on a C++ interface layer for LinuxCNC. The idea was to wrap both the NML and HAL polling sides in a clean, multi-client-safe bridge so the same core can support a WebSocket interface for remote clients and a Node.js addon for local UIs. The main goal is to decouple machine access from the frontend layer, making it possible to run multiple displays and control interfaces against a single machine. It's proving to be about as much work as I anticipated

Don't stop this project.

One of my subscribers is having an issue. Unfortunately, I have not yet tried using the LAN-style MESA card, so I am unsure how to help him.

Jere is the message I got:

"Hi Jerry,I have a Mesa 7i92 card that is plugged into the Ethernet on my computer.  It is also connected to a 7i77 daughter card, and I have a 5v power supply going into the 7i92."

Do any of you have some tips that might be helpful?
Thanks,
Jerry
 

"Great work, Rod! This looks like a monumental step forward for LinuxCNC.

I believe this branch is the perfect strategy for implementation. It elegantly solves the real-time synchronization issues that have hindered advanced tangential control in the past.

For my part, I have an assignment to study this implementation . You have indeed laid a cornerstone for the system, and I am very excited to leverage your work for this project.

If I may add my opinion: I fully support your point of view regarding the modifications of G_1, G_2, G_3 equates in control.c. I understand the devs' concerns about real-time sanctity, but your approach to code readability (housekeeping) is crucial for the long-term maintainability of such complex code. Finding that balance is the key.

Thank you again for your incredible contribution."

Hallo Peter, T3D drivers exist with encoder output also. I have them in one machine, but I have used them in STEP/DIR mode with Mesa, without encoder feedback. Works fine, but I still hope that analog control from LinuxCNC and Mesa is better. Is it right or not? I also have other machine with Simodrive, there is +-10V control, attached to 7I33 - works perfect many years. But T3D has analog control 0-10V Do you know how does this work? Is it possible to control it with analog servo Mesa cards? I am planning to upgrade next CNC milling machine and looking for motors. I would like to use T3D, but milling machine I do not want to run without encoders feedback...

Looks fantastic. Its great seeing more UI options beyond Qt. Just curious, is the display optimized for touch or keyboard input?

Being more of a C/C++ person, I’ve been building a similar pet project focused on a C++ interface layer for LinuxCNC. The idea was to wrap both the NML and HAL polling sides in a clean, multi-client-safe bridge so the same core can support a WebSocket interface for remote clients and a Node.js addon for local UIs. The main goal is to decouple machine access from the frontend layer, making it possible to run multiple displays and control interfaces against a single machine. It's proving to be about as much work as I anticipated

This was originally supposed to be just a tough shape to pocket, it looks like it could be used for making a saw blade too.  The algorithm used to develop the perimeter would be prone to accumulate minor errors into major blunders since it accumulates multiple executions for sine and cosine functions to add up to a complete closed polygon.  The verbosity setting prints out the value that is computed by the program to be the "last side".  This one had a last side that was .065" bigger than the other 19 teeth.  repetitive editing of the apex angle of the 19 teeth eventually resulted in a "last tooth" that was .0015" too small. Verbosity set to 3 dimensioned a bounding box where width was .00065"  less than height so the shape is reasonably round, I did draw a circle around the tooth tips and couldn't see any big (or small) ones  so when I decide to mill this at least the database will be more accurate than my mill.  
I added the circular area inside the part since I had left it unfinished on my previous post, the screen shows the circular path but doesn't show the boundary circle I used to eyeball "roundness", nor does it show the center mounting hole which took a pile of g-code to finish.
the program is still g-code in its entirety, and its gotten pretty Hackey lately. 

It should be noted that Riocore can generate the gateware for the FPGA, as well as the complete LinuxCNC configuration (hal/ini/vcp).
In future, however, it will also be possible to build your own boards from this;
to this end, a KiCad template is generated, which can be turned into a complete board with relatively little effort.



already in dev branch, but still need some work:
* verify the kicad modules
* add more kicad modules
* some fixes while update the pcb

I am referring to executing an instruction like this G0 Z25;X0 Y0;Z0 or some OWord routines through external buttons, and I am referring to instructions not included in HALUI.

Hallo,

ich versuche einen Novusun NVEM mit STM32F207 CPU an LinuxCNC das auf einem kleinem PC (kein Raspberry PI) anzubinden.

Was ich bisher gemacht habe:

1. Installation LinuxCNC auf dem PC
2. Flash der Remora Firmware für NVEM
3. LinuxCNC Ethernet eingerichtet auf 10.10.10.100/24
>> Ping zu dem NVEM läuft
4. Gemäß Quelle DOC folgendes durchgeführt um eben Remora von GitHub auf den PC zu bekommen:
>> Mir ist hier aufgefallen das einiges fehlt unter anderem die Basic-Config "remora-nvem-basic" und unter den Components "Remora-nv"
Ich habe diese dann extra zum PC rüberkopiert in die entsprechenden Ordner.
5. Installieren der Componenten mit halcompile
>> Dabei ist aufgefallen das hier "Remora-nv" fehlt also nochmal nachinstalliert:
6. Die Beispiele der Configs in den Linuxcnc Ordner kopiert - dazu zusätzlich die "remora-nvem-basic"
7. Upload der Config zu dem NVEM - dazu musste ich tftp nachinstallieren 
>> Bei ausführen des uploads scripts dann bin ich im selben Ordner wie die einfache Config.txt von "remora-nvem-basic":
>> das scheint auch soweit zu klappen.
8. Start von LinuxCNC und auswählen der NVEM Basic die ja zur Config.txt passen soll. Alles Startet ohne Fehlermeldung.
>> ab hier kann ich zwar bei LinuxCNC den eStop aufheben und auch die Maschine einschalten. Aber sobald ich eine Achse X verfahren will kommt "Joint 0 following Error"
Ich denke der macht keine Pulse zumindest kommt bei den Treibern nichts an. 

Wie hast du die zum laufen bekommen?

The voltage range of the Mesa inputs is something like 5-32v and for the supply you could either take the 5V from encoder/stepdir outputs of the Mesa or have a separate small 9V or something like that psu just for the probe. Just connect the 0V together with Mesa field power 0V

Yes it seems external running of macros is broken. I will fix it asa I can.

Question though, Macro/MDI commands are not well defined terms , when you say macro, do you mean like Gmoccapy macros where you can have variables that you enter at request time?

Fair point, and I get the frustration — I've felt it too. But I'd push back a little on lumping this in with the VC-backed subscription-everything crowd.
I've been at this for twelve years, solo. No investors, no team, no safety net. I own a small machine shop, I run Mac & Linux (hate Windows), and I built JetCad3 because the tools I needed either didn't exist, didn't run on Linux, or cost more than the machines I was running them on. FreeCAD is genuinely good software and I have nothing but respect for it — but my scope is much larger, and I don't think the end goal is achievable as an open source project.
The subscription isn't the goal — developers are. I've written my own geometry libraries, my own NFP and Skyline nesting algorithms — I own all of it specifically so I'm not paying upstream licensing fees that force my prices up. I want this as cheap as humanly possible. My end goal is to put the middle finger up at Autodesk, not replicate their business model.
I'm blue collar as it gets. I already run two businesses and support my family — JetCad3 isn't a get-rich scheme, it's a twelve-year bet that someone should build serious CAD/CAM that runs on Linux and doesn't cost small shops an arm and a leg. If revenue ever lets me hire developers who've actually run a machine, I think I can genuinely compete with Autodesk, MasterCAM, Dassault, and ProNest — and that's a win for this industry, not a betrayal of it. I want to be so much cheaper than them with more/high quality features that it forces them to lower prices when they start hemorrhaging subscribers...
One thing I maybe didn't make clear enough in my original post: there will always be a generous free/hobby tier for every workspace in JetCad3. Guys building their automotive project in the garage will always be a priority for me — because that IS ME.

Thank you all

Currently I am only running first tests on my machine. But mostly in sim. Still a lot of bug fixing to do. I was recently working on some performance improvements on the development branch which I will merge to main soon.

To give you more technical background: this controller is the result of over 10 years of experience in industrial machine retrofits.

The project was accelerated by the massive Mesa card shortage in 2022. Instead of waiting for hardware that wasn't available, we decided to develop our own high-performance industrial-grade solution tailored specifically for complex retrofits. This isn't just a prototype—these units are already integrated and running 24/7 in machining centers, proving their reliability in real production environments.

Key Hardware Specifications:

Core: Custom FPGA running HostMot2 firmware for rock-solid real-time performance.

Fully Isolated Analog Servo Control: Precise ±10V analog outputs with a 7V/μs slew rate. The analog outputs are fully galvanically isolated from the logic to prevent noise and ground loops.

High-Speed Feedback: Isolated encoder inputs supporting up to 10MHz—ready for any high-resolution industrial encoder.

Industrial Isolation & Protection: Complete galvanic isolation between logic and I/O. The 24V field power is protected, and the logic power is fully isolated.

Onboard Processing: Integrated Raspberry Pi Compute Module 5 (CM5), providing a massive performance boost for LinuxCNC and NativeCAM.

Thermal Design: Fanless industrial enclosure. No moving parts, no dust intake—designed to last in woodworking shops.

The goal was to create a controller that is fast, precise, and, most importantly, available for professional retrofits when the industry needs it most.