Advanced Search

nice.. have you considered simply adding the probing tool number parameter to the toolchange call with an if statement? this way anytime you load or unload the probe tool it would activate a subroutine to either deploy the probe or retract it?

interesting concept on the ride along design. that is going to add quite a bit of unbalanced weight to the machine maybe? have you considered how that will affect high speed motion and inertia differences for the slaved drive motors?

I am about to release a fully integrated RACK type atc integration for probe basic, this could be a simpler and more economical method for atc maybe?

I think that you are likely to need some I to pull in the last bit of error. Have you tried adding I (very small numbers, .00001 maybe as a starting point)
Are the servos operating in velocity mode or torque mode? (ie, is there a velocty loop closed by motor tachometers)

First thing I would do is setup the LinuxCNC host computer with a static IP address of say 10.10.10.100
 This is most reliably done by editing the interfaces file as described in the hm2_eth man page:

man hm2_eth

Then after rebooting (or setting the interface down and then up) verify that the host Ethernet device
is set to the 10.10.10.100 static IP address with:

ip a

Then set the FPGA card to a 10.10.10.10 IP address and cycle its power.

finally check that you can ping the card:

ping 10.10.10.10


 

Sorry for the late reply.

I think that you can just load two carousel components.

loadrt carousel names=left, right

for example.

Then you can enable the relevant carousel inside whatever handler script / subroutine you are using. Presumably based on the tool number? 
I guess you will need some logic, if only to work out where to move the spindle to pick up the correct tool. 

Handling a random toolchanger, where the tool can swap between carousels would be rather more of a challenge. 
 

papagno-source wrote:

Note: Using LXRT realtime

 

Interesting! Are you using RTAI + uspace or Xenomai? I don't fully understand how to make LXRT work, or how to compile the wrapper applications. 
(Which is a bit of a problem if you are the release manager) 
 

What is the last MX version? did you try that version?
The latest version should work with just\
sudo apt install linuxcnc-uspace
But that would be 2.9.4 , not Master, so a manual upgrade would need doing
All this is based on it being a Debian derivative, so it might not work.
Just in case you are still trying to use the old how to for the 19.4 version, don't, it will not work on latest versions, as far as i know.
-
It is advisable to use Debian 12 or 13, or even Linux Mint Debian Edition 6, as all that is required to install LinuxCNC on them is the single command mentioned above.

Hello everyone, I’ve been trying to install master Linuxcnc, with mx 19.4 or mx23

Although in the past, I’ve done it seamlessly, currently I’m having huge trouble with dependencies because no matter if I activate the system or not upon installation, I always have problems with libudev or libgtksourceview.

if I try to remove the current versions, which are not compatible with Linuxcnc, i’ll end up removing other programs and creating even more dependency problems. Although I can install the version that linuxcnc-uspace wants for libudev, if I do it , it will remove other required software. Same for the other dependencies…

i’ve followed the guides, but I always get stuck with this problem. Does anyone have a different approach to resolve this?

thanks in advance

now i have another error message?


 

I am facing a very similar situation. (almost identical to the above descriptions)
I had some issue with the Linux installation to do apt-get, so today I tried to install the latest ISO image new.

Then the Mesa card was not giving any response at all.

I tried lots of different things. Now in my desperation I tried to ping the card from a windows computer and the 4 green LEDs on the 7i95 start with some life signs) aa ping. 192.168.1.121 gives me responses.

Verdict: Probably some mishap with the networkadapter on the NUC PC.

Any advise how to narrow down the Ethernet problem on the Linux installation?

I can also change the card to 10.10.10.10 and ping it from the Windows machine.
My assumptions

• 7i95 card seems not to be damaged
• 5 V power seems O.K.
• the patch cable is not causing the issue 

       Hi,
I am trying to assign to a macrobutton in QtDragon an o-word script that would allow me to toggle the button to switch on and off a laser.
The laser is connected to an output pin (hal_pi_gpio.pin-10-out) and this pin is connected to a signal (laser-on).  But, it doesn't seem to work. Enclosed are the custom M100 macro (to obtain the actual value of the output pin) and the o-word subroutine called by MDI-COMMAND = o<laser_on> call. Any idea why this won't work. Any suggestions how to do this?  Many thanks !!

Well, I went through the website.

But, I want to know about the exact hardware configuration. That is not given in the website.

 

I'm building the carousel at the moment, it's a ride along design with a geneva drive wheel.
Also with a retractable dust sucker, so should be fun to get this all to work.



 

Got it to work with auto extending the probe.
I've programmed the probe to automatically extend when the machine is in probing motion state.
The motion state 5 (probing) is setting a flipflip signal, which is bound to the output pin for a relay and pneumatic valve.
The flipflop bit resets when the machine program is in idle state (which for some reason seems to be inverted "program-running").

Altough I probably add a manual switch or button later, as this way you can't really accurately place the probe before starting the subroutine.
And it also extends when doing a tool offset measurement.

Here I was probing a set square:


 

Small question, after I restart PB I lose the values in the Probe screen and the settings for the Tool setter (like spindle head distance).
How are those retained normally? It didn't happen at first, so must have happened while editing .HAL and .INI.

Tool offsets and WCO do get saved, no issue with those.
 

I was asking as I was wondering what is the practical resolution. Everyone talks about better resolution, but do we practically need this "better" resolution? From mesa website: "10 bits at a 1 KHz sample rate and 12 bits at a 250 Hz sample rate"
is 250 Hz update slow? is 10-bits @ 1 kHz not enough? Heard somewhere that 0.01" THC control accuracy today is pretty good and pairing it with other info that 1 V arc voltage change relates to ca 0.004" (varies but can be used as rule) gives 2.5V arc voltage change. 300 V range @ 10-bit gives 300mV resolution. If 250 Hz is enough then 12-bit gives 73 mV.

I know 0.1 nV is better, but do we need it?
It's just because i have two plasma sources: Powermax 45XP and CUT 70P from Stahlwerk. I'm not going to buy CPC kit for 45XP so THCAD-300 will be more universal. I don't know if my 45XP has any divider installed inside. I don't know if CUT 70P has divider as well. I'm away so have no access to machines to open them and check internals. But I have now option to buy THCAD cheaper (shipping to US instead of to europe to avoid unnecessary and significant costs).