Advanced Search

Hi all, 

In case anyone else is interested in using inexpensive mini PCs and Mesa ethernet boards, I thought I'd share some details of my new setup. It's early days, but it seems to be working well--I'd be interested to hear about other people's experience with similar setups...

Mini PC, N95 $120 It's tiny, like 3-1/2" square!
Mesa 7192T $110 has ethernet and parallel port for the G540 (you need to build your own enclosure and supply 5v)
Gecko G540

I got the PC off Amazon:
www.amazon.com/dp/B0DGQ1P667?ref=ppx_yo2..._fed_asin_title&th=1
described as:
Mini PC Alder Lake-Ν95 (up to 3.4GHz) 4 Cores 4 Threads, Mini Computers 8GB DDR5 256GB M.2 SSD, Mini PC Dual Lan Support

Here are my somewhat cryptic notes about the configuration process. It's basically an assembly of techniques from other threads on this forum. I changed the IP address of the card from the default, but in hindsight I don't think that was really necessary.

Install linuxcnc from USB stick

Disable Turbo mode in BIOS (not sure this is really necessary)

(Note: there is no hyperthreading option in this particular machine)

Change mesa card jumper for "static IP from EEPROM"

edit /etc/network/interfaces to add static address for the card:
auto enp2s0
  iface enp2s0 inet static
    address 10.10.10.11
    netmask 255.255.255.0

sudo mesaflash --device 7I92T --addr 10.10.10.10 --write 7i92t_g540d.bin

change grub:
  isolcpus=2,3

===== At this point I was getting network timeouts. The fix for this has been posted on this forum, and is as follows: ============

Because of network latency issues it is necessary to install the dkms driver per this thread:
docs.google.com/document/d/1jeV_4VKzVmOI...ading=h.macj649sy0yq
Open a terminal window and Type the following lines: 
sudo apt update 
sudo dpkg -P raspi-firmware
sudo apt dist-upgrade
Note that the second line (sudo dpkg …) removes firmware for the raspberry pi incorrectly installed by Debian and is a bug out of our control

Type:
sudo apt update
you need to install some utilities. Type:
sudo apt install build-essential dkms 
If you have not installed a later kernel as described above install linux-headers. Type:
sudo apt install linux-headers-$(uname -r)
sudo apt install r8168-dkms 

Rob

109jb wrote:

I got my touchscreen monitor delivered yesterday and just wanted to confirm that the issue I was having with touchscreen button not working right was a consequence of using Flex GUI in a virtual machine on windows. The Flex GUI that I made works perfectly on a real machine with touchscreen. For the real machine I am using a RPi 4b and a generic 10" touchscreen monitor.

For anyone else designing their GUI this way, I am using VirtualBox on a Windows 10 machine. The virtual Machine has LinuxCNC 2.9.4 installed from the live iso download. Here are the couple quirks I have found running it in the virtual machine. I want to stress that these issues don't exist once the .ui file is put on my RPi machine.

1. touchscreen buttons don't respond as expected in virtual machine
2. The machine will unexpectedly go to "Power Off" as soon as a motion command (Jog, MDI, or in a program) completes. Doesn't do it every time, but pretty often. Again, only in the virtual machine.
   
   I also want to again give thanks for this wonderful tool and to give a peek at what I am working on. I have a 3 axis benchtop mill (RF-45 size) that I converted to CNC a few years ago. I did this as a hobby and use the machine for my hobbies. It has been run with many things including LinuxCNC on an old tower PC with ancient monitor, keyboard and mouse, with GRBL on an Arduino hooked to a Netbook with my own Visual Basic interface, and I tried one of those standalone controllers, but sent it back because it was junk. I finally decided to just do it better and got a RPi 4b, installed LinuxCNC and have been working on an interface that for most of the way I use the mill can be done without ever touching a keyboard. I'll still have a small wireless keyboard with trackpad available if needed though. This is the interface I'm working on and I think it will suit me. I have the UI sized to perfectly fit the display that I bought to use as much space as possible. Thanks again for the very useful tool.
   
   John B
   
    
   

If you want to add that gui to the Flex examples just upload the ui file and I'll add it.

JT

I have a turning application that's better suited towards a constant RPM on the rotary. I have my A axis functioning fine right now with position control.

I'd love to be able to just issue a M3 S1000 $1 instead of playing with the inverse time mode approaches.

I've searched the forum and tried several things without luck. Honestly, anything HAL related is an instant headache nightmare for me. Believe it or not, I'm an embedded programmer and did ladder logic in college, but the HAL stuff just does NOT click for me. I'd be willing to pay for someone's time to figure this out for me. I have a project for my church that is way behind.

I'm using Probe Basic for the GUI. I'd be willing to give up some GUI control as long as I can issue gcode commands.

Attached are my current INI and primary HAL.

Tried: 
- Changing control_type from 1 to 0, but had no idea what else to do
- Considered the rosekins but that doesn't really apply to my needs

Other forum posts I've read:
-  C Axis as Second Spindle - LinuxCNC
-  Using A axis as second spindle with stepper motor - LinuxCNC
-  Lathe spindle question turning and milling. - LinuxCNC

Thank you.

HI
I am wanting to add an optional rotary function to my table, to have a tubing cutter.  With another controller and software I just switched the Y axis control output cable to the rotary axis motor.  This is also the way it is done with my desktop laser to use its rotary optional drive assembly.
On my table I and using LinuxCNC software with a Mesa 7I76E card and QTPlasmaC GUI.  I have tried to leave out the Y home/limit entry in the pncconf setup, but it will not enable the Y axis.  I understand the need to do the homing operation so the software knows where it is for running a G-code.  After the homing I can position any axis to a location and then zero out the DRO and use that as a start position for the G-code. 
What I want is the Y axis not needing to home to be active.  This would allow me to home X and Z and position the Y to a zero position manually.
Thanks for your time
Tom V 

Confusing? For me for sure:) 
I'm now surprised by this hobbing.comp and its installation as i don't see you used "loadrt hobbing"+"addf hobbing.0" in the previously attached files.
 

With -1 you ask the master not to synchronize the dc clocks for the slaves. I would think 1000 could be a good number, every 1000 cycles the master will synchronize the dc clocks of the slaves. The clocks will run with good precision by themself in between syncs.
Sounds like you got a weird setup with EoE, never seen that before. It is explained in the IgH ethercat master manual. A way to establish ethernet connection to a slave over ethercat. Doesn't happen by itself must be something in your setup.
The message "Slave did not sync after 5000 ms" may not be fatal, most likely its clock got synchronized later.
You can enable debug with "ethercat debug 1" and check debug messages with "sudo dmesg". Maybe you can get something out of that. There will be more messages about the synchronization and how that succeeds.

I think that this could probably be solved by using TCP kinematics. Then the probe positon could be fed into the kinematics (via a HAL pin).
If you expect this job to run for months and thousands of parts then it is probably worth considering. Otherwise the mechanical stop as suggested above is probably more expedient.

If you have a spare lathe compund slide (or similar) lying about you could mount the collet block on a slide, so that you can align the end of the work to a probe....

spumco wrote:

 4th axis build based on similar reducer.

Do you mean this one? bodgesoc.blogspot.com/2017/05/harmonic.html ? 

This is often due to the stepper driver "enable" pins being wired, but not properly conttrolled.
The stepconf wizard doesn't dribe the "enable" pins, and most stepper drivers have an enable that is "enabled" when unconnected.

I would generally suggest saving pins and not wiring the enables. There is no real need for an enable control on a stepper driver (they can't run away in a fault condition, unlike a servo)

Actually the 5I25 has connectors P2 and P3 which are logical ports 1 and 0 respectively.

The info on that Wiki page is possibly a bit confusing as it combines a HAL config that I used initially, with the GUI that I use with a more developed version.

The .ui file is a GladeVCP and would be loaded by a line in the INI like:

I also seem to have fogotten to mention the custom "hobbing" HAL component. This is installed from the attached file with the command "sudo halcompile --install hobbing.comp"

This is then loaded in the main HAL file with:
(It looks like I need to re-write that Wiki page)

danke Dir!

Ich hatte im Netz schon gesucht, aber bin immer nur auf Sachen mit Zeitzonen gestoßen – das hat bei mir nicht gepasst, weil bei mir auch die Minuten nicht gestimmt haben.

Über das Setzen des Termins hab ich's jetzt hinbekommen:


Dein Link hat mir trotzdem geholfen – danke nochmal!

Gruß
Mücke

It sounds like you homing sequence might not be homing all axes.
You should see a circle/compass rose/ crash-test dummy target appear next to every axis position readout when homing is complete. Are you seeing that?

I am a but puzzled too, there was talk of adding it to LinuxCNC, and a suggestion that it might already have been done. But I don't see any "axis2" in the code.

I am fairly sure that it _isn't_ the UI in sim/gaxis, but perhaps that could work for you?

not easily so if it happens again i might have to rethink things. this was on the X axis so not too much mass being moved. the whole machine is pretty light less than 100lbs the X might hit 20lbs if that. most of the weight is the t-slot table and gantry beam.