EtherCat homing, limits, and probing

More
14 Nov 2026 19:09 #350698 by mwc
I'm aware there is a release imminent to make proper use of in drive homing routines, but it's also got me wondering about limits and probing.

Limits aren't that big an issue, as they should work just like any other limit switch by just using whatever input is suitable, but most EtherCat drives can act on the limit inputs internally by stopping motion. Is it worth the effort of utilising that functionality?

For reference, my current build has a dual motor gantry, and will home of the limit switches, so there would obviously be some parameter changing needed during the homing process.

Probing is a bit more interesting.
Probing can be handled like any other system with LCNC directly controlling things. but again EtherCat drives often have the ability to have a probe directly connected, with the probe trigger storing the exact position to an internal register.
Has anybody made use of this functionality yet?
I know the easy answer is just to probe slower to minimise how much any time delays affect the read position, so this is more about the theoretical possibility, as you'd then also have the electrical problem of connecting a probe to every drive that might need it.

Please Log in or Create an account to join the conversation.

  • Hakan
  • Away
  • Platinum Member
  • Platinum Member
More
14 Nov 2026 23:06 #350699 by Hakan
Replied by Hakan on topic EtherCat homing, limits, and probing
I guess it is a case where the drive moves outside of LinuxCNC direct control.
Just like homing. Mostly a question on how to interface to motion. Have you dug into that?
The drive will need settings for the probe action, that's manageable.
 

Please Log in or Create an account to join the conversation.

More
15 Nov 2026 02:40 #350704 by elizacommunities
Replied by elizacommunities on topic EtherCat homing, limits, and probing
Thanks for raising this, it is a useful question. On limits, I would keep LinuxCNC in control of the limit handling where possible, so the behaviour stays consistent across the axes, and treat the drive's internal stop as a safety net rather than the main mechanism. For probing, the drive's internal latch sounds attractive for accuracy, but as you say it adds wiring and configuration for every drive, so for most work probing a little more slowly through LinuxCNC seems the simpler and more maintainable route. I have not used in-drive homing yet, so I would be interested to hear how it behaves in practice once the new release is out.

Please Log in or Create an account to join the conversation.

More
15 Nov 2026 03:41 #350705 by tuxcnc
Replied by tuxcnc on topic EtherCat homing, limits, and probing
In most machines and their applications, using internal homing within the drive systems makes no sense and offers no benefits.
While it might technically provide greater precision, the question is whether that precision can actually be put to practical use. In amateur-grade machines, certainly not—and even in professional-grade equipment, not necessarily.
For instance, what is the point of homing a plasma cutter to an accuracy of 0.001 mm?
When turning or milling from raw stock, the machining allowance provided is far greater than the precision of the homing process.
Even when performing an additional operation on a nearly finished part—such as reaming a hole—you should still set your zero point using a probe or dial indicator; otherwise, a tiny speck of debris on the vise could ruin the entire job.

Please Log in or Create an account to join the conversation.

More
15 Nov 2026 05:55 #350707 by rodw
Replied by rodw on topic EtherCat homing, limits, and probing
I think homing is useful particularly where torque mode homing without any switches as on that said plasma, there is added simplicity engineered in by deleting limit and home switches. Limit switchers are part of some homing methods so if you do homing, you might as well use limits as well.

Probing to me  is  not useful in CNC but maybe for industrial applications. Applications like QTDragon and probe basic do not leverage the full capability of Linuxcnc homing because they don't probe onto a part and then probe away from it. Instead they repeat a probe at a slower angle.

From memory a 3d probe I had said not to exceed 200 mm/min dso something like this is how we should do ut
g38.2 Z -100 F200
g38.5 Z 5 F100 (or F50 even)
F100 = 0.00166 mm per servo period so surely that is as accurate as you need surely
The advantage of this is it removes any spring from the probe and gets a very accurate position much faster than the methods currently In use which seem to have been inherited from Mach3

Please Log in or Create an account to join the conversation.

More
15 Nov 2026 08:03 #350709 by mwc
Replied by mwc on topic EtherCat homing, limits, and probing
@tuxcnc for most machines, basic limit/home switch homing is perfectly adequate, especially on stepper driven machines, but on servo driven machines it is pretty common practise to use index homing, however EtherCAT makes that harder to achieve.
There's also the added benefit with closed loop EtherCAT drives (and especially now with the advent of multiturn absolute encoders) that the drives themselves maintain position through power offs, so in drive homing has become a much more beneficial thing to have and use.

Probing to me is the more interesting part. I have a homebuilt CMM, which I currently have no plans to upgrade as it does everything I need it to, so this more a discussion around what could be possible, and when I get around to retrofitting my little mill, I'll want accurate probing on it.
My CMM is stepper driven, with the loop closed using glass scales, and it all driven and controlled with a Dynomotion KFlop and Mach3 using the ProbeIt plugin. The exact trigger point is stored in the KFlop, then transferred to Mach3, eliminating any delays in Mach3 recognising the trigger.

@rodw I know you mention the methods being inherited from Mach3, but Mach3 can handle various probing methods. I know in ProbeIt, you can rely on simply the first touch, slow second touch, or reverse off for storing the touch point.

As a general comment, I'm well aware of chasing pointless accuracies.
The current plasma build, I'd be happy with a 1mm accuracy, as plasma isn't exactly a precision process, however on the two lathes I've previously retrofitted (both of which are now sold, as I have a commercial lathe with standalone controller), accurate X-axis homing made a big difference.
First lathe used a simple optical slot, which had 0.02mm repeatability, but there was noticeable drift with temperature, so if doing anything requiring high accuracy, I had to do a couple test cuts first, whereas the second lathe used servo index homing, and was always spot on.
The following user(s) said Thank You: rodw

Please Log in or Create an account to join the conversation.

  • endian
  • endian's Avatar
  • Away
  • Platinum Member
  • Platinum Member
More
17 Nov 2026 12:02 #350779 by endian
Replied by endian on topic EtherCat homing, limits, and probing

I'm aware there is a release imminent to make proper use of in drive homing routines, but it's also got me wondering about limits and probing.

Limits aren't that big an issue, as they should work just like any other limit switch by just using whatever input is suitable, but most EtherCat drives can act on the limit inputs internally by stopping motion. Is it worth the effort of utilising that functionality?

For reference, my current build has a dual motor gantry, and will home of the limit switches, so there would obviously be some parameter changing needed during the homing process.

Probing is a bit more interesting.
Probing can be handled like any other system with LCNC directly controlling things. but again EtherCat drives often have the ability to have a probe directly connected, with the probe trigger storing the exact position to an internal register.
Has anybody made use of this functionality yet?
I know the easy answer is just to probe slower to minimise how much any time delays affect the read position, so this is more about the theoretical possibility, as you'd then also have the electrical problem of connecting a probe to every drive that might need it.
 

Yes you right .. this problematics were open here many times and I think there are much more capabilities which can drivers provide to us.. Homing (see the link there it homes at internal probe register) and probing and internal latch mechanism is one of them which is must to have...
But it is tinkering, I can not implement it inside the linuxcnc without other users help a Luca was also asked for this and much more .

Much more means to have near zero lag at ethercet devices with linuxcnc with CSP(which are now lags (t+2) as ethercat system)... that means to send position(t+2) and velocity(t+2) .. velocity is practical to use from the PID and its fine.. but if we need the next next posiiton is much harder .. 

Probing is same song .. when we will be able reach faster servo scans we will be much more capable in the handling the edges but we will be not able to reach FPGA level ... therefore every manufacturer has that outside in the servo 

If we not solve these mentioned issues linuxcnc will be forever the tinkering nerdy app with great hidden potential .. I am sorry if I am little bit nerdy

Please Log in or Create an account to join the conversation.

Time to create page: 0.108 seconds
Powered by Kunena Forum