Advanced Search

Hello everyone,
I’m working on an unfinished DIY 5-axis CNC router project with a BC head. The machine was originally built by a company but never fully completed, and it has been sitting unused for about 2 years. It never achieved proper simultaneous 5-axis operation.
The current setup is based on Mach3, without TCP/RTCP, so true 5-axis machining is not possible.
Machine configuration:
X, Y, Z + B, C axes
Stepper motors with encoders on all axes
Dual motors on the Y axis (gantry configuration)
BC head driven by stepper motors with gear reduction
Total: 6 motors
I’ve decided to rebuild the control system properly from the ground up using LinuxCNC.
At this stage, I’m trying to define the correct approach:
What is the recommended kinematics setup for a BC head machine?
How is TCP typically implemented in LinuxCNC for this configuration?
Are there example configurations I should study?
Hardware side:
What Mesa configuration would be appropriate for this setup (6 motors, dual Y, encoders)?
Is something like 7i95T sufficient, or would another combination be more suitable?
My goal is to build a stable and correct 5-axis system, based on a proven approach.
Any advice or pointers would be greatly appreciated.
Thanks!

If all you did was autotune, you aren't done.
You need to set the feed forward. Also set the in position error allowance.

What could be happening is that 2 or 3 microns is still within the drives set error, so it just doesn't move.

So I have some L6 and L7 servos coming to evaluate for including with my machines. My understanding these are *mostly* the same as the EL6/EL7 models on their website, just OEM and 1/3 the price.
One notable difference is the 24 bit encoder on the L7.

Anyone set these up in linuxcnc? Issues? Success?



 

got myself an sgd7s mechatrolink-iii drive... tried to talk to it. although it's basically a 100base-tx ethernet, they are not using ethernet 802.3 headers. and worst at all - they violate the 60 byte minimum frame length. by 4 bytes i was able to convince Zynq MAC to transmit short frames by turning off FCS offload. i can see the drive answers, but they get dropped and there's no way to receive them.
NICs on linux PC are no better.
so the only way to do it with Zynq is to implement a custom ethernet MAC in PL.. unfortunately my board has PHYs connected to PS pins, so i can't do it. will try again with another board...

Are u using current version of translator?

Because I can clearly see subindexes of pdos are 0x00 instead of 0x01 which is clear issue to me l... 

Download latest translator version and try export again.. versions before were buggy 

Check it or edit it manually(fastest) and let me know

"Hi everyone,

Thank you all for testing the port and providing feedback! I've just pushed a few fixes to the repository (including the INI parsing errors and GTK deprecation warnings). github.com/cnc-proton/nativecam-py3-gtk3...vecam_2.0b-4_all.deb

Regarding Bug Reports:
As more people start using this port, it becomes harder to track individual errors and logs across forum pages. If you encounter a bug, a crash, or a specific error message, please report it via GitHub Issues here: github.com/cnc-proton/nativecam-py3-gtk3/issues

The forum is still the perfect place for general discussions, sharing ideas (like the Lathe Polyline!), and showing off your work. But keeping the technical bug reports on GitHub allows me to solve them much faster and keep the project organized.

Thanks for understanding and for your continued support!"

I've tried that too. Now I've noticed that I have pid.c.feedback-deriv connected. Maybe getting rid of it will help, Guess I'll try that next.

Hello andrax,

I tried to follow your guide but still getting .hal file error for missing pins.
I ran LinuxCNC from terminal and it outputs that the HomeComp started.
Then the terminal reports that pins "joint.0.request-cia-homing" and "start-cia-homing" are not found.

HEllo Endian,

thank you for the link i tried my best to fix it myself (and later with help of AI...) i hope i understood most of the problems.



I am trying to get a Beckhoff AX5125-0000-0216 running.

Current situation:
- LinuxCNC 2.9.8
- IgH EtherCAT master 1.6.8
- all other slaves on the real machine bus go OP
- only the AX5125 stays in PREOP / PREOP+ERROR

I also tested AX-only (only the AX connected to the LinuxCNC EtherCAT port, no other slaves) and it still does not go beyond PREOP. So it does not look like bus order / other slaves are the root cause.
It runs perfekt in TC3 also after  change from 2ms to 1ms i could jog it by hand.

What I already verified:
- TwinCAT project was switched to 1 ms
- exported again from TwinCAT
- converter run with c -i
- appTimePeriod = 1000000
- IDN 1 = e803
- IDN 2 = e803
- generated assignActivate 740 was changed to 730

Observed errors during the different attempts:
- AL status 0x0045: MBX_SOE
- AL status 0x001D: Invalid output configuration
- Writing IDN 0x80CC failed
- SoE configuration failed
- sometimes PDO entry 0x0024:00 is not mapped
- in AX-only test it still remains PREOP

Current result with LinuxCNC running:
- all non-AX slaves = OP
- AX5125 = PREOP E

I am attaching:
1. current ethercat-conf.xml
2. current initCmds/Antrieb-1-AX5-0-initCmds.xml
3. TwinCAT 1 ms export XML
4. ethercat slaves output
5. ethercat pdos -p 7 output
6. relevant dmesg output

Question:
Do you see an obvious issue in the AX SoE startup / telegram / DC setup, and what would be the correct minimal working config for AX5125-0000-0216 with lcec?





Thank you very very much for your time and help!

All the best
Tom

With open loop, (stepgen feedback) you need to have a _much_ higher P term
(typically 1000 with a 1 ms servo thread period)

 

Yeah, the title is a bit misleading at this point, but not sure it's worth making another thread for this.

I see what you mean but I've tried vastly different acc/vel values on the joint and stepgen sections and it still has the same kind of behaviour.

If it was a physical issue that should be ruled out when using stepgen.position-fb instead of encoder feedback, I'm thinking so? But that makes no difference still...

I tried to figure out how the joint comes up with the position command but with no luck either.

Thanks for the help,
I'll go through this step by step and get back to you 

This retrofit is possibly the easiest retrofit of all mills. I used a 7I97T with the existing servo dives and VFD.Notes:1. Aside from the servos and VFD there are only five 24V miniature relays to control.2. Since the servos utilized a tach (Aka Resolver) drive tuning took minutes to tune.3. The only additional board I made was transistor board to provide a signal to the J425’s drive enable relay. The drive enable optocoupler on the 7I97T cannot provide sufficient current for this miniature relay. This board consisted of a P-Channel Mosfet and a 10Kohm pull up.4. On the J425 all of the axis limit switches and the Estop are all connected in series to a single relay.5. I was able to use all of the existing wiring. I didn’t have to cut or splice any of the existing wiring.6. Almost all of the wiring to the 7I97T was through the existing DB connectors.7. Just like any other CNC, you’ll need wiring diagrams. They are available from zpsusa.com/kneemills.html I did not have any issues with the installation or performance of the Mesa 7I97T. I was concerned on its ability to thread mill since it requires 3 axis’s synchronized. It wasn’t an issue at all, in fact I cranked it up to 120ipm without any following errors.

Good evening, has anyone tried the new fixes in Grandixximo v1.41.0-pre2?
Are the slave pin values ​​read quickly?

Yeah, that’s definitely part of it as well.

I’ve touched on 5-axis concepts too, mainly around coordinate systems, tool orientation, and how things behave behind the scenes — even if most of my hands-on work has been 3-axis and some 4-axis.

It definitely adds another layer compared to standard setups.