Advanced Search

If your test machine has no external e-stop circuit you shouild simply connect the e-stop out signal to the estop in signal in your hal file. Something like this:

net estop-out <= iocontrol.0.user-enable-out
net estop-in => iocontrol.0.emc-enable-in

Then the e-stop buttons on the userinterface will work. and you will need to toggle them when starting Linuxcnc.

I don't think you would need to change anything in your servo drive. The only changes should be how it is seen and commanded inside Linuxcnc.

How is your servo currently commanded? I presume it is by step/dir commands in position mode? That would not have to change, you would simply have to change what signals are commanding the Mesa hardware stepgen so that they are commanded by the spindle commands instead of the joint that it is currently commanded by.

For weeks, I've been working on my own CNC machine and have made good progress, but I need help with some small details from experts.The machine is a MicroCarve MV3 with 3x 23-305-DS8A steppers in bipolar parallel mode, connected to a TB6600, which is connected to a Raspberry Pi 4. I don’t have a stop button yet.I’ve managed to wire the motors and Pi to test the axes with a Python script. I also created a minimalist INI and HAL file and started TkLinuxCNC. However, I can’t move the motors because the software shows an E-stop, and despite my efforts, I can’t get the machine to move.I access the Pi via VNC without a connected monitor. The image used is rpi-4-debian-bookworm-6.1.54-rt15-arm64-ext4-2023-11-17-1731.img.xz, updated to LinuxCNC 2.9.6.Friends, I need help. I’ve spent many hours on this and am close to getting the machine running.My test script works; HAL and INI files are attached.Thanks,
Dragan

I found a safety issue with the toggle2nist function in my gamepad.hal file.
if you e-stop while switching in pause, than switch on and start program again, pause will often not work.

---

this is a safety problem if you want to trust your pause function.

---

Therfore I use a special version of toggle2nist: forum.linuxcnc post 05 Oct 2025 18:50 #335873 special  toggle2nist.comp from user Aciera. You have to compile that in by hand.
www.forum.linuxcnc.org/24-hal-components...ton-at-the-same-time
With "pulse-length" parameter set to 50 the problem of the hanging pause is gone.
Therefore, the default toggle positioned before the toggle2nist component must be removed across all implementations to ensure correct toggle functionality.

also changed: e-stop with 13 to stop with 13
 
 

Try this small example config. That should make one axis work.
If it doesn't it is much easier to troubleshoot.
github.com/dbraun1981/hal-cia402/tree/main/example

 

Sounds like a user component.

There are some examples, for example the mitsub_vfd.py here github.com/LinuxCNC/linuxcnc/tree/master/src/hal/user_comps

have a look at the example config in the 'sim/axis/remap' folder:
github.com/LinuxCNC/linuxcnc/tree/master...emap/getting-started

Hi All,
I have been playing with this for days and have not cracked it. I was using AI for answers but every solution failed. I want to do the following. I have the python code sorted but getting LinuxCNC to call the python scripts seems to be the problem. The python script sets the Modbus TCP registers on the VFD.Spindle forward, pass the value 1 to my python script as an argSpindle reverse, pass the value 2 to my python script as an argStop spindle, pass the value 5 to my python script as an argSpindle speed, pass a value between 0 and 200 to a python script.can anyone offer a solution. Cheers
 

The last cool thing i remeber reading or putting any effort into was the gentleman from Proma on a Monokrom thread asking people to beta test his. Capactive THC Analog DC converter, I wonder how that went? And then ther was the makerspace that hacked together a usable motion control to build their fiber system from some clever manipulation of QTPlasma . other then finding a laser power source which will happen sooner then laater "fingers cross" 

Since I was last asking for assistance, I opted to switch over to a 7I96s board. I am having issues getting a cheap MPG pendant to jog axes in the Axis GUI. I have confirmed that the encoder works as well as the dead-man switch allows the signals I've designated as axis selectors and step values to trigger appropriately. I have seen forum posts about joint vs axis jog being toggled through the use of homing. (joint vs teleop) I utilize the Home all button to zero out the display before attempting to jog any axis. Using Halshow, I am able to see the joint-select-(a-d) and joint-incr-(a-d) trigger as well as the axis-selected-count value count up and down. I used the PNC wizard to setup the pinouts and didn't see "Axis select" as an option but did have "Joint select" available. I assume, based on what I've read about the 2.9 update that this wouldn't be an issue as I'm just using a standard 3 axis cartesian XYZ setup with possible expansion to XYZA once I have more confidence in using Linux CNC to actually run my machine.
It's probably something overly simple like linking joints to axes in a different way than PNC did by default or just a need to add a switch to convince the machine that it's been homed.
I've attached the Hal and Ini files for review.

I am running LinuxCNC 2.9.4 from the official repository, but it seems that modbus functions 1 and 5 aren't available. (loading my partial mb2hal config file at halrun throws "[fnct_05_write_single_coil] out of range") I read that Linuxcnc 2.8 didn't support those functions, but that 2.9 should. What am I missing?

I've gone back to 2.9.4 because of a similar problem with 2.9.5/6.
At first all buttons on the pendant work fine until I try to call an MDI_COMMAND_xx. The command itself is not executed but afterward there is a zombie process and most of the buttons on the pendant stop working. In halshow all buttons and connected signals still show activity - even MDI_COMMAND_xx are getting called - but nothing happens. After restarting linuxcnc everything ist back to normal. With 2.9.4 this does not happen! Seems like something is wrong in 2.9.5/6...

I have been fighting these iX7NH drives for several days and they still won't reliably communicate via ethercat. However, they talk ModbusTCP just fine. What's the best way to control axis drives via ModbusTCP? The machine being driven is a pallet changer for a milling machine, so path accuracy doesn't matter, only endpoint repeatability. I figure it won't matter that I can't have CSP mode.

In order to make the drive move the motor, I have to:

1. Set coil 0x12 == 1 (SV_ON: This enables the drive output)
2. Specify movement speed (register 0x6034), acceleration (register 0x6036), and deceleration (0x6038)
3. Specify the target position (register 0x6025)
4. Set the control word (register 0x6001) to 0x30, then immediately to 0x20 (I tried this without linuxCNC, in Python, with no delay between the statements, and the servo system executed the move correctly)

Step 1 only has to happen once, step 2 whenever the application requires it, and steps 3/4 whenever a movement must happen.
It seems that I could use MB2HAL to create HAL pins for coil 0x12, registers 0x6034, 0x6036, 0x6038, and 0x6025, and data from LinuxCNC's motion planner would sail right through to the drives, but what about the control word register?
I have to set and then clear bit 5 whenever I want the drives to make a movement.

You're welcome, I'm glad you got it worked out. On the topic of separate machines, the logic is there and holds water; it's almost like multiprocess welders, unless you drop some serious money, you're compromising. And I, for one reason or another in the past, needed both couldn't afford both, or didnt have the space, I know I'm basically handicapping the machine, whichever gui is loaded but not ot the point, parts are out spec, or customers bringing stuff back. The plan is to build another machine for a 3 to 5k fiber laser, ditch the plasma altogether, and mount an ATC spindle and vacuum holding to this current one. I've been really good this year, hopefully a smoking deal comes through on BidSpotter for a laser source. This is the 5th iteration of a CNC gantry-style machine I've built. I went HAM on it just because I'm tired of being tired of not being satisfied with it.  The .5" flat bar on top has a reference shoulder machined in and a skimmed on both sides im gonna use jack screws every 12" or so in pairs to get this thing as coplanar, square and level as pyhics will allow, me to and then use a metalized epoxy to bed the bar to the frame and finaly a few stitch welds to lock it in.  ce la vie i gues. Your set up is pretty gangster 

 

Hello, what is the content of toplevel.py?

import oword
import remap