Advanced Search

I created a diagram so that the offset value can be visually confirmed.

Yes, the machine is on, it jogs and the icon for homing is grayed out, when I get to my shop next week I will upload pictures. Absolutely appreciate you guys are really awesome. Super helpful and responsive! Much appreciated! If there’s anyone in this forum that is for hire, let’s talk.

Got it to home but had to change the end stops to just home not home/min, I'll post up my files once I drag the pc upstairs to give it internet, I'm sure there is lots to still fix and do

I'd like to share a structured and slightly modified version of `qt_auto_probe_tool.ngc` I'm using on my self-built 3-axis CNC with LinuxCNC 2.9.3 and the QtDragon GUI.  

---

### ? My Goals:
- Separate tool length offset from work coordinate systems (WCS)
- Make the tool table intuitive and readable
- Ensure safe and efficient probing using machine coordinates

---

#### 1. Set the Probe Top:
I defined the probe top using the tip of the spindle body (with no tool/collet) as a consistent mechanical reference.

- For example, if the empty spindle is lowered to Z = -120 (machine coordinates), I input `-120` as PROBE HT in the QtDragon Probe tab.(I know this is not the intended use.)
- I do not use BLOCK HT at all.

#### 2. Define Work Coordinates:
Once a tool is mounted (e.g., Tool #1 with 30mm protrusion), I:
- Add `T1 Z30` into the tool table
- Set up the G54 (or other WCS) normally

This way, the tool length stored in the table is actual physical protrusion and independent from any WCS — much more intuitive for manual tool changers.

Use G43 to apply the offset.

---

- Tool table values are stable and logical (they represent actual tool lengths).
- Easy to estimate and input values for new tools before probing.
- Tool probing uses machine coordinates, so:
  - Probe start position can be calculated from tool length
  - No dependency on current WCS
  - Safer: if tool table Z is negative by mistake, start offset is clamped to zero
- Multi-WCS workflows (G54, G55...) work seamlessly — no need to re-probe per WCS
- Tool probing is simpler and can be single-pass since start is close

---

Thanks,  

SanzuiWorks

G10 L20 A0

linuxcnc.org/docs/stable/html/gcode/g-code.html#gcode:g10-l20

I did some manual G-code to use a 4th axis rotary basically like a lathe.  The A axis doesn't have a homing switch.  It advances a specified number of turns as the X and Y does its thing.  

Part gets changed out, repeat.  

What's the correct G-code to reset the A axis?  Numerically, it's spooled up like 1000 turns in one direction.  Moves aren't relative and I don't need the thing to unspool 1000 turns to get back to the absolute A-axis value the next G-code run will start at.

 

So, does it not respond to the home button? Is the home button greyed out?
Does it jog? Are the motors locked or free?

I don't think that counts mode _can_ work with a Boxford style of toolchanger as the number of steps lost when the motor stalls against the endstop is unknown.

It could be modified to work, but it might be easier to start from scratch as a single-sensor position-mode stepgen changer wasn't really considered when the component was written.

You could even consider doing all the control in the G-code subroutine:

1) Scale the stepgen so that a position shift of 1 is one pocket.
2) Connect motion.analog-in-00 to the stepgen.2.position-fb
3) Connect motion.analog-out-01 to the stepgen.2.posiiton-cmd
4) Connect the home sensor to motion.digital-in-02
(Actually, you can use any numberered analog/digital IO, you would probably use in, out and digital 00 in practice, but I made them different for clarity)

toolchange.ngc:

I have also overpowered the 5v side of a 7i76 board for a second or so. I saw smoke at S1 and driver chips etc.
Freight is too expensive to post back and forth to USA for repairs. ( In Australia)
Can you make up a component set that would normally be damaged? I'll buy another board and the component set...... to try to recover this board for my next project.

Looks like your network adapter is not upped correctly. The master can't access ethernet.
Check "ip a", it should say UP for the adapter but I guess it won't.
Follow one of the standard debian guides to bring it up, will involve the /etc/network/interfaces file.
Can be a permission problem, check "ls  -l /dev/EtherCAT0" should say crwxrwxrwx
Try with "sudo service ethercat restart"
Also check "sudo dmesg" the last say 50 lines for anything that can help.

Good to hear, wouldn't surprise me if you change D1 and D2 to X and Z.
That config should be enough to move the axis. Try it.
Then you need to add limit switches and other switches if you have that, spindle control, pendant or gamepad and so on.
Now when linuxcnc is up and running you can easily inspect pins' values using halshow.
The drive and motors are closed-loop I guess, then you will see actual-position change when you rotate the motor shaft, even by just a little if the motors are active.

I think the installation is fine. The system reports POSIX realtime, and the pins for testing are accessible:

Owner Type Dir Value Name
4 u32 OUT 0x00000001 lcec.conf.master-count
4 u32 OUT 0x00000004 lcec.conf.slave-count

When I now call EtherCAT Master, the system reports:

Master0
Phase: Waiting for device(s)...
Active: no
Slaves: 0
Ethernet devices:
Main: 80:ee:73:fd:0f:8b (waiting...)
Link: DOWN

The network adapter is correctly entered in Ethercat.conf:

MASTER0_DEVICE="80:ee:73:fd:0f:8b"
DEVICE_MODULES="generic"

The permissions are also correct:
'KERNEL=="EtherCAT[0-9]", MODE="0777"'

However, I can no longer access the EK1100. The EK is connected to the network and shows no errors. As mentioned, everything worked fine with the same components before the reinstallation.
 

They are enabled if pin 9 and 10 are not connected. So in default state with no external relays connected the drives are enabled.

When I short 9 to 10 they become disabled, disconnecting power to the drives and locking the driver.

I´ll measure the pins that come from the drive.

Hello Folks,

just a quick update: At the Moment I'm in the process of aquerring the machine. The machine was listed in working condition. But sadly it has an error in the B-axis servo amp. The seller is realy cooperative and tries to fix it asap.

In the meantime i was thinking about to major "hurdles":

1.) Milling spindle orientation

The machine can hydraulically lock the spindle via hirth coupling in 12 positions. This is of course quite importent for tuning tools and broaching tools. Of course i want to implement this feature in LCNC as well. The first solution that came to my mind is using a custom M-Code that executes a custom HAL Component.  The M-Code posting needs than to be implemented into the post processor after every tool change.

But i would prefer the solution by Mazak. They used iirc a custom tool change command and added the desired spindle angle.

It looked something like this: T01XX while XX equals the desired spindle orientation.

2. Turning spindle + C-axis + clamping I saw this is done via the PP so never mind.

There are plenty of examples on how to implement a lathe spindle + C-Axis in LCNC. What worries me is the spindle locking.
There is the INDEXER available, but this works only exclusivly with G0 moves. As soon as the G-code contains G0 CX and i.e. G1 CX FXX it does not work anymore. How can i solve this problem? My idea was to somehow use motmod.
C-axis can be locked if the next line of G-code does not move the axis and unlocked if the next line of code does. The question is how.

The bridge the waiting time for new posts on this topic here are some pictures:
 

Yes. 7i77 please