Advanced Search

Doesn't qualify as walk through but here are some hints. I used mb2hal.

forum.linuxcnc.org/10-advanced-configura...h100-inverter#335652

can someone walk me through the setup on linuxcnc for this? i am seeing so many conflicting and incomplete instructions and components that are no longer supported or redundant (like vfdmod).

Also what USB dongle?

I have a rtl8192cu based USB WIFI dongle on this test machine
and it seem to have no effect on latency when plugged/unplugged

Have you done the usual optimizations?

DKMS driver for RealTek Ethernet chips

Turn off IRQ coalescing on Intel Ethernet chips

in grub file isolcpus = last processor

pin Ethernet IRQ to last processor


 It means that the network message could not be compete in one cycle.

Actually it's worse than that, with default settings, it means 5 read requests in a row timed out
so usually means multi-ms latency...
 

So, its definitely doing something different when the drive is on vs off. So that's a start. It does show exception 1 a fair bit.


[27/10/2025 15:29:13]
Modbus Response (COM5)
Address: 1
Function: 4 (0x04) - Read Input Registers
Exception Code: 1
Checksum: 0x82c0 - OK

---

[27/10/2025 15:29:13]
Modbus Request (COM5)
Address: 1
Function: 1 (0x01) - Read Coils
Starting Address: 773
Quantity: 0
Checksum: 0x2c4f - OK

---

[27/10/2025 15:29:13]
Modbus Request (COM5)
Address: 1
Function: 1 (0x01) - Read Coils
Starting Address: 779
Quantity: 0
Checksum: 0x4d8c - OK

---

[27/10/2025 15:29:13]
Modbus Request (COM5)
Address: 1
Function: 1 (0x01) - Read Coils
Starting Address: 912
Quantity: 0
Checksum: 0x3c63 - OK

How is the network for Mesa card set up, using network manager or by editing the /etc/network/interfaces file?

isolcpus=5

would be correct for a 6 core CPU

Hmmm. that's helpful thanks. its doing... something in both directions. so maybe the issue is the spindle talker 2 software.

Any advise on another way to check functions?

That is in your hal file for the X axis, it should beif you want X to be the first joint/axis.
Same for other joints and axis, and there are multiple instances for each joint to be changed.

I have been advised to isolate a core. I believe my Intel has 6. I have done the following:

I think I have isolated a core correctly.  cat /proc/cmdline BOOT_IMAGE=/vmlinuz-6.1.0-30-rt-amd64 root=/dev/mapper/debian--vg-root ro quiet isolcpus=1

OK, thanks. Time to start coding.

lorenzing wrote:

Thank you all very much for the support. The drives use the CiA402 protocol, with the jc2ktr setup I was able to see the drives and I can perform jogging, but I don't understand why I see Z1 on joint 0, which should be X. And when I try to jog joint 1, joint 0 always moves. I definitely made a lot of confusion in the HAL file.

Maybe your Hardware (ethercat ports, or motorcables) are wired wrong, your halfile looks okay to me?
I´ve had at work that if you swap in and out of the Ethercat ports you get weird behavior on the EC-Bus.




 

maybe www.com-port-monitoring.com/ or similar?

so its not talking to my delta vfd either... so im guessing its not the vfd side thats the issue. hmm

Hi,
I have seen the following issue occur on three machines that Range from 2.9.3 to 2.9.6. 
The hardware is all cases are Lenovo i5 mini PC's running MX Linux 23.x Bookworm. The WiFi dongle is a USB type

What seems to happen is that when there is USB activity the network to the Mesa 7i96S will get a read error. It means that the network message could not be compete in one cycle. It typically happens when the USB stick is removed or inserted, especially if the machine is busy executing a program. It will also happen if the WiFi is active like when a file is uploaded to the nc_files directory from a laptop on the network.

Is there anyone else that has maybe seen this or know of a solution?
 

It's giving kinamaticsInverse failed error. As per documents joints can't have offset values. So to align J1 and J2 these must be in line, arm j1-j2 should be horizontal to give offset values zero. But practically I can't make arm in this position.