I'm also NE Ohio (Wayne Co.) and I didn't think most of the farmers right around me had 3ph. But maybe that has more to do with the fact I'm on the edge of Amish country.

@retrofitcenter...

Hi , I gathered that you’re from Austria. I am too.
I think what you’re aiming to implement in LinuxCNC is great and exciting.
The multi-channel control option is definitely a real asset to the LinuxCNC project.
It makes it possible to successfully retrofit older machines—such as complete Homag woodworking centers with multiple table zones—using LinuxCNC. I work in the woodworking industry myself.
I also took a look around your website—you’re working on some great projects there.
What kind of training or education did you complete? Mechanical engineering?

Why? Cant you just lock it?

absolutely. I agree that exchanging files through the forum would get cumbersome very quickly once we start iterating.

I’m using GitLab for the development, so I think that would be the easiest solution. I can set up a dedicated repository/project for the testing where you can upload your LinuxCNC configuration, HAL/INI files, custom components, machine documentation, etc. We can then track changes through Git and use issues for individual problems or features.

If you don’t want your machine configuration to be public, that’s no problem, I can make the repository private and give you access directly.

A video call would also be very useful. Especially at the beginning, it would probably save quite a bit of back-and-forth if you could show me the machine architecture, channels, spindles, axes and what you want to achieve.

And yes, the 32 mm Swiss-type machine is exactly the kind of machine I’m looking for. For around $5–6k it could be a very interesting test platform for the multichannel system.

I’ll prepare the GitLab structure and send you access/details.

Todd Zuercher wrote:

Much of the USA is very rural, and in many places it can be more economical to buy and install a large generator than to get a large 3ph service from a utility.

All the farms around here can get 3ph drops pretty easily - or already have it.  It's really the suburbs and cities/towns that have an issue in this region (NE Ohio).

But I've no idea if those farms have "large" services or just enough amps to run some irrigation pumps and conveyors/fans/elevators.

@retrofitcenter...

This forum has no PM functionality.  Do you have a github repo or similar location where I could provide my config and related files?

I have no issues posting files here, but the inevitable back-and-forth with file transfers during iteration gets really clumsy when done through a forum.

I'd be happy to discuss whatever you'd like during a video call.

This is the sort of thing I'm talking about:  $6k for a 32mm swiss.  Probably get it for 5.

hgrinc.com/productDetail/CNC/Used-Star-M...C-Lathe/11240360356/

Much of the USA is very rural, and in many places it can be more economical to buy and install a large generator than to get a large 3ph service from a utility.

Dear All,

Thank you all for the warm response to our product and for all the feedback so far. The launch offer closed today and is now a regular price of USD 399 for a perpetual license.

Also 1.0.8 has been pushed.

Kind Regards,
George

Thanks for the pointers, and for the prebuilt packages. They make it much easier for others to try this.

For context, here is where I ended up and why.

Hardware and goal
- Beckhoff C6930 industrial PC (i5-7500, 4 cores, no HT), Debian 12, no Beckhoff software.
- LinuxCNC with the IgH EtherCAT master, using an AX5203 drive over SoE.
- Strict 250 µs cycle, plus a custom 5-axis TCP (G43.4/G43.5) and a modified trajectory planner.

What I tried
1. Xenomai 3.3.3 Cobalt on Dovetail (CIP 6.1.170 kernel), built from source with Skylake tuning:
   - Core isolation, no P-states or C-states, tsc=nowatchdog, an i915 workaround and pci=noaer for the PLX backplane switch.
   - The native latency test showed about 3.6 µs max idle and 2.8 µs under load, with 0 overruns.
2. LinuxCNC on that kernel. At the time I was on 2.9 without a Cobalt flavour. Its HAL threads fell back to plain SCHED_FIFO in the Linux root domain, and latency-histogram showed about 36 µs, worse than PREEMPT_RT.
3. Xenomai 4 / EVL, with AI help, using default settings. In my hands it came out worse than PREEMPT_RT.

What I chose
Stock PREEMPT_RT (6.1.0-41-rt-amd64) with core isolation. On the isolated core at 250 µs, cyclictest gave 6–7 µs max, idle and under load. EtherCAT ran for a 30-minute soak with desktop load and stayed stable.

Xenomai's best case saved me about 3 µs, which is irrelevant against a 250 µs budget. To get even that, I would have had to write and maintain a lot of custom code myself: the glue between LinuxCNC, the EtherCAT master and the co-kernel, RTDM-capable NIC drivers, and the tuning for each. I decided that wasn't worth the maintenance, so I stayed on PREEMPT_RT.

What I'd like from this setup
- A hard 250 µs loop with low jitter that I can maintain on a stock Debian kernel.
- No dependence on a patched co-kernel for every kernel update.
- Eventually a self-contained appliance, with safety handled by external TwinSAFE/STO hardware.

Why I'm posting
- It's a great project, and I'm glad it works well for you.
- For anyone with a similar setup, the overhead around the co-kernel is real. On this hardware it buys microseconds, not a different class of determinism.
- If you have a minimal EtherCAT-master-on-EVL setup, or numbers for igb or e1000e with EtherCAT cycles, I'd like to compare. I'll also look at the Xenomai 3 / 4 support in master (issue #3878) with a fresh install.

Regards

Thanks a lot for the detailed description and the offer – that's exactly the kind of machine I'd like to test with.

Your setup is actually a very good candidate, even if it's not a classic swiss. A main spindle with its own X/Z, plus a parting slide and a subspindle that need to synchronise for the hand-off, is precisely the situation where a second channel makes sense. Sequential operations are fine – the interesting part is the synchronisation at the hand-off point, and that's what I want to validate on real hardware. Step/dir with glass scale feedback is also a good test case.

Regarding axis names: the 9-letter limit stays as it is inside each channel. Every channel has its own task, interpreter and trajectory planner, so each channel has its own XYZABCUVW namespace. A G-code program running in channel 2 simply uses X, Z and C – the interpreter doesn't need to know about "C2". The names C1/C2, X1/X2 etc. are the global, channel-qualified identifiers used in the configuration, the HAL side and the GUI, so a swiss with C1/X1/Z1 (main) and C2/X2/Z2 (sub) is no problem. In your case the sub would become a real C axis in channel 2 and take part in that channel's trajectory planner, instead of being an extra joint.

Axes are assigned to a channel in the configuration, and the idea is that an axis can be handed over from one channel to another at runtime (e.g. for a transfer move). That part is still being worked on.

No coding needed from your side – I'll prepare a configuration for your machine and walk you through it step by step. To get started, could you send me:
- your current INI and HAL files
- which hardware/FPGA card you use
- a short sketch or photo of the machine layout (main, sub, parting slide)
- a typical part program, especially the parting/hand-off sequence

Then I'll come back with a proposed 2-channel layout and a test plan. I'll also share screenshots and a short video of the current state soon.

@spumco – thank you very much for the offer, I really appreciate it. I'm based in Austria, not in the US. However, I'll be in California from October 17 to November 17. I'm tied up with an important project during that time, but I'd be glad to meet up while I'm there and talk about how we could move forward – whether that's a machine from HGR or another way to get real swiss-type hardware for testing. I know California isn't exactly next door for you, so a video call would work for me as well. Just send me a PM and we'll find a date.

Thanks again – and thanks @tommylight for the support!

Yes, it will be open source. I’m just still finalizing the first proper release before publishing everything.

At the moment I’m integrating a lot of functionality inspired by industrial CNC systems such as Siemens, especially channel synchronization, spindle/axis synchronization and transformations for turning/milling centers, including face and cylindrical surface machining.

Another major part is the simulation system, including material removal, machine visualization and collision/crash detection.

The tool management is also being heavily extended with a proper tool database, sister tools, lifetime monitoring and 3D models of tools and fixtures.

I’d rather release a solid first version with a clean structure and documentation than publish something half-finished. Once that first release is ready, the source will be made available.

The error messages indicate that transaction 1 & 2 are bad but in reality all are bad.
the first register is a single 16 bit register. I would map functions to decimals. looks like it will become 0=off, 1= on
0 000 stop
1 001 run enable
2 010
3 011
4 100 coast stop
You don't have any hal plumbing.
You need to scale the desired speed to be a frequency in hal so S12000 should resolve to 200 Hz (2000 in register) - assuming normal 400 Hz VFD