Advanced Search

Happy to hear that!
And to get feedback that the dkms driver really helps.

I haven't had a chance, honestly I don't think I could justify one ATM. The old tech budget has taken a beating already this year.
The Intel NICs sounds like a good idea.

If it's this one www.stahlwerk-schweissgeraete.de/plasmas...t70-pilotzuendung-en 
I would look at one that doesn't use HF start. You don't want HF start in cnc table.
Generally, blowback start is what you want.

I got pretty reasonable results on the RPi5 even with Wayland and LXDE.
Enough that software step generation might be an option for a mill. I was surprised by that.
I've had to go back to the basics with pi-gen, was having issues with it deciding to switch from the graphical target to multiuser.
I've got a bunch of images to test, so stay tuned.
But I think once this is nailed it will be easier to get reproducible results. You've convinced me regarding the kernels. So much easier. Thanks

Ok cool, git gets a bit over my head. I ain't real smart, but I used to lift heavy things .

Hi,I'm currently on my journey to build a CNC Plasma Cutter. For the plasma source I'm looking into the Stahlwerk Line-Up of Plasma Cutters (Cut70 or Cut100). All Stahlwerk Cutters have a build in THC Connector. What I can't figure out is if this connector is already divided. Or if an external divider circuit is needed. The Manual just says that it is used to connect the cutter to a CNC controller. Does anyone have detailed information? Or own such a device at has already checked the voltage levels? Greetings
Nils  

EDIT: I see there is already a workflow, but it doesn't build for arm64 or trixie, nor does it publish a github release with built packages. I'll see if I can fix that.

github actions/workflows support both amd64 and arm64, and are free to use for public repositories. I'll see if I can cook up something to build the packages using that, and pull-request it if it works. it could also potentially replace the buildbot altogether - I'm not sure if that's a good idea or not.

that said, I'm a bit short on time at the moment, so don't expect the PR tomorrow.

interesting, the check was removed quite a while ago . I wonder what else does it want.

as to LXDE vs XFCE, I'd probably go with the latter, it's true and tried, and almost no one uses the former - Pi OS uses Wayland and some Wayland-based desktop environment by default, I don't think many people switch to Xorg apart from us.

I've been diligently building up a 5-axis machine for machining small analytical instrument parts like fluid manifolds. My work envelope requirement is quite small. It's common to drill 0.5 mm holes 10 mm deep, and have them connect with an intersecting hole on another axis.

 

The pictured machine is wired with a Logosol CNC package (5 Logosol LS-231 drives + 1 kW spindle drive + a supervisory controller that manages door safety and more). This package is likely in many small specialty CNC machines produced around the world. I've been using Logosol's Windows based CNC program, which is free (Thanks, Logosol!) but is not really supported, updated, or documented. I've been in touch with a Logosol support person, who has been great to work with, but the software feels like a dead end. It's been great for proving the concept, but I'm forecasting a move to LinuxCNC.

The Logosol drives and supervisory controller are nicely wired - so it would be a shame to alter the reliable wiring to use ±10V or step + direction. Logosol's protocol is via RS-485, and includes 256 point path buffering.

My thought is to write HAL components to interface LinuxCNC pins to Logosol's LDCN protocol. The servo drives run a 51.2 microsecond update rate, so 20 servo ticks is 1.024 ms. LinuxCNC's update rate is 1 ms. I can either alter LinuxCNC's rate, or downsample the path points to match (feedback here is appreciated!).

I'm based in Raleigh NC, USA and welcome helpers or onlookers in my shop!

Edit: Isel is the company that made the 3 linear axes and the 2-axis A&B trunnion. They made the vertical brackets, the extruded aluminum base, and many other components. I presume they integrated the whole machine as a 'box builder', and the CNC vendor is more of an applications specialist. The build quality is fantastic.

One more note: The spindle is a Jaeger, and supports tool changes. You can see 12 tool holders dismounted from the bed + a tool height sensor.

I'm going through old Logosol posts. I'd appreciate hearing where you are on this project.

You can communicate from a PC to a chain of Logosol drives with a USB to RS-485 converter. No Mesa ports to work with.
These converters typically use a chip like an FTDI for USB-to-UART conversion, plus an RS-485 transceiver (like a MAX485) to handle the differential signaling and half-duplex operation on the bus.
RS-232 vs RS-485:
RS-485 is half-duplex (not full duplex). It uses two wires (A/B or +/-) that are shared for both transmitting and receiving - devices take turns communicating on the same pair. RS-232 is full-duplex, using separate transmit and receive lines.
The key advantage of RS-485 is differential signaling, which eliminates common mode noise and allows much longer cable runs (up to 4,000 feet) compared to RS-232's typical 50-foot limit.
RS-232: Uses separate RX, TX, and GND lines. Single-ended signaling measures voltage relative to ground.
RS-485: Uses differential pairs (A/B or +/-). The receiver compares the voltage difference between the two wires to determine logic levels. Since noise affects both wires equally (common mode), it cancels out in the differential measurement, making the signal much more robust.