Advanced Search

Hola. I want to use a CMOS laser sensor with an RS485 interface and find out if it is possible to use the RS485 terminal on the 7i96s board to read the sensor values. My goal is to use the laser to measure height and move the Z-axis to a specific initial height on a plasma cutting machine. I consulted Gemini AI, and it indicated that I needed to flash new firmware to enable this function; consequently, I visited the Mesa website and downloaded the hostmot2 file for the 7i96s. Could someone confirm if that hostmot2 file allows me to use the 7i96s's RS485 port in the way I intend? Thank you very much .

Input 8  (TB5 pin 1)  is

hm2_7i97.0.inmux.00.input-08
 

I recommend building a run-in-place (RIP) copy of the master branch.
linuxcnc.org/docs/html/code/building-linuxcnc.html
forum.linuxcnc.org/9-installing-linuxcnc...run-in-place-install

PCW wrote:

You need to configure the network on the host for the Mesa card.

Also 192.168.1.121 is not suggested for LinuxCNC use as it's  in
a commonly used NAT range.

Suggested setup is to use the 10.10.10.10 address on the Mesa card
(via the IP address setup jumpers) and set  a static host address of
say 10.10.10.100/24.

At this point you should be able to ping the Mesa card:

ping 10.10.10.10


 

I found the problem I had the whole time but Linux didn´t point it out enough for me. Since the 7i97t is connected via ethernet i ran into more and more problems when I approached wiring up the machine and testing in and outputs. Today i started wiring the Lathe to the Mesacard and i didn´t get any MPG Handwheel feedback at all, a few seconds into starting Linuxcnc I´d run into queueing errors.
Little research pointed me to major network problems, causing the 7i97t to indicate a fault with CR7 red whenever i try to connect and hold connection.

I turned the checkbox "show notifications" in the network tab top right on and viola i got an error "disconnected". So i went with your suggestion and set the Jumpe on the card to eeprom default ip=10.10.10.10
still the same problem. For sure i changed my own IP to 10.10.10.100. Still the same. I changed the cable. Still the same. I tried a lot of thernet configurations.

In the end, the fix was: The subnet mask was set to "24" and not to "255.255.255.0". After fixing that the card connected right away and the network symbol stopped spinning. Since I´m not familiar with Linux, that spinning symbol gave me bad feelings but there is no Tooltip showing up, telling me what that icon actually means.

Now got already both MPG wheels hooked up. Somehow I cna´t find the TB5 IN8 pin in Halshow...

Hi Chris,

thanks for looking into it.
I am currently just running the normal 2.9.10 LinuxCNC. I guess it came with a version of QtDragon? I have to figure out first what's the best way to run the Master branch.
Can I check it out and use it parallel to the 2.9.10 version and be able to switch back and forth?

Guido

I have build foam cutter a few years ago and not remember details, but I think it was as follows:
There are two possibily configurations:
1. XY with master-slave axes - no cones, but you can use G2/G3.
2. XYUV independent axes - can cut cones, but can't use G2/G3.
When you use XYUV you must convert circles to strings of short lines in your CAD/CAM.

I've never worked with a wire foam cutter, but I have made some Linuxcnc configurations that can use some custom M-code commands that will temporarily slave two axis together such that one will move with the commands of the other. The machines I was using this on were routers with two spindles each mounted on their own axis and configured XYZW. I had ran these for years just using XYZW g-code, but when the multi-line look ahead planer was first introduced in Linuxcnc, it only supported XYZ moves, any moves with a W command reverted to the old single line tool planner, and significantly slowed the wood carving we did with these. So I built my config with an M-code that slaved the W axis to the Z at their current positions when the command was issued. The W would then follow the Z until I issued the M-code command that released the W. (This had to be done when Z and W were commanded to be at the same positions relative to the position they were slaved at, or the W would give a following error.) Maybe this technique could be useful for you? Here is a copy of one of my configurations.
 

Ok try this one:

The first problem is Python.
Python is powerfull, but permanently developed, so python scripts are version sensitive.
LinuxCNC depends on Python...
Debian is conservative, Ubuntu is progressive, so there must be version mismatches.
Sometimes old programs not works with new OS, but worst is when works, but not properly, become unstable or some functionality gone.... 
Believe me, experiments like LinuxCNC+Ubuntu are not worth wasted time.

Mass - thanks so, so much!

This is sweet. I will dive deeper into it in the days to come and the project you linked to. But I am almost sure I will buy a (used) SK4/6 panel. I wouldn't be surprised if they all come with the PS/2 + matrix combo. (Some will possibly be USB+matrix but that would just be a win). I know, no guarentees - also in terms of pinout and such, but the main foundation is certainly there as you have shown. And most important, nothing locked behind a proprietary serial protocol either, it seems.

But any chance you could share the code, too?
Having looked at the github for the Arduino-LCNC project a bit, I now realize that the matrix never gets turned into USB-HID or gamepad protocol, at all. Interesting. Maybe a better more direct approach.

Not that I really know what I will be looking at but with the help of Gemini, I managed to make a nice custom version of the NanoELS FW (an open source electronic leadscrew project) for my little lathe. Had to get DRO scales implemented and add some other features - and I actually managed to do so on code level. So, hopefully the robots can help me on this project, too. 

Various Panels
As for SK4/6 I actually think they share the same layout 100%. I think I have even seen some SK6s with SK4 printed on the PCBs - but I could be wrong as I have looked at sooo many pics lately.
But there's also an SK5 which does have a slightly different layout. (Attaching pics of SK4/5/6).
Most often SK4 has membrane buttons whereas the SK6 often has tactile ones. I am leaning towards the latter.
But looking at SK PCBs get confusing very fast as there are so many different itirations of the SK4/5 plus a bunch of clones. 

Encoder Overrides?
BTW, it looks like you swapped out the clicky rotary switches for pots or encoders? So, no hardstops at each end but possibly better "sync" between screen and knob, is that it?
I think I will try to stay with the rotaries and see how they feel. 

We (this forum) do not close or retire topics as some forums do, i have no clue why they do that as the only thing that does is cause chaos and makes the forum unusable with many instances of the same Q&A that mostly never conclude or solve anything.

See what Tuxedo OS uses for touch, i am sure it worked on Mint also as i tested on a Lenovo X1 Yoga a while back, but do not recall what the package name was.

sorry, I posted on that one, but when nobody replied, I figured the post was too old and wasn't active anymore

Ha, that's spot on—I really did feel like a detective sometimes
The top part is a standard PS/2 keyboard; I connected it to the computer using a PS/2-USB adapter. I remapped the keys, mainly to access lowercase characters (Remap the Shift + button, just like in the original system).
The lower part is connected via an Arduino and an ArduinoConnector (forum.linuxcnc.org/show-your-stuff/48630-arduino-io-expansion).
On this keyboard, the pinout is rotated 90 degrees relative to the visual orientation, and not all pins are used. I added a pinout; I hope this helps.
P.S. The SK6 keyboard has a visually different circuit board; I haven't checked it, so all of this applies only to the SK4.

Hi Chris.

Here you have the file.