Advanced Search

Lcvette wrote:

You need to stop editing the py files, they are generated from ui file and your going to break things doing what you are doing. All edits must be performed in the ui in qt designer as things are designed. I'm going to go out on a limb and say your other issues come from allowing ai to edit files directly. A warning, ai sucks at xml and ui editing because it rarely has our understands the complete context. And if you don't know when it's hallucinating to guide it back to reality it's going to continue down the wrong rabbit hole. Just do your edits/work in designer where it builds the file correctly on start up and correctly uses all necessary code to initiate the used widgets. Your ai generated posts are mostly hallucinations where it may seem confident in it's conclusion but it's completely out in left field and no where near what's actually happening.

Ai seems wonderful until you realize it's extremely fallible.
 

Appreciate your take on it.

I've not touched any of the probe_basic py files, qss files, or anything outside of the config/user_nnn/ folders.

Unfortunately, I am not able to diagnose the various issues without assistance when I can't accomplish what I want in qtpyvcp/qt/PB.  And I would prefer not to pester you or other skilled folks unless absolutely necessary.

I have zero coding skills, period.  I can work out HAL logic no problem, but figuring out how to create a widget rule, stylesheet, or other text/terminal-based edits is a no-go without some sort of assistance.

I tried to add some hal leds.  Didn't work until AI helped me discover it was the layout (as you pointed out).

Once there, I tried to create a widget rule for the hal pins.  Didn't work.  Still doesn't, but without AI I wouldn't have discovered there's no widget rule channel for HAL items.  I would have either given up, or I would have fiddled with everything I could until I broke it and had a tantrum.

With AI help, I now have user-button panel & sidebar indicators that work exactly as I want, and a DRO that's as close to what I want as is reasonable.  Getting to this point would have been impossible without AI assistance, or at best would have taken weeks of back-and-forth over the forum.

I'm sorry if I'm coming off as a clown who's spamming you or muddying the waters with AI slop.  I genuinely hesitated to post anything, including the current GH issue I've got open, but I thought it might be helpful.  Even the "easy fix, you haven't formatted the sidebar..." was a non-obvious problem to me.

And thank you for all the work you've done.  Seriously.

I have tuned the servo driver to be totally still via the pots when no input is present. Its the balance (RV8) pot. I did this per the manuals instructions. I am unable to even try to do the mentioned tests because the servo has a mechanical advantage of a 90/1 worm drive and just runs... i would not dare to try to stop that thing.. LOL!!!

You did not clean the chips so i can not the the one close to the port.
It does seem like outputs for the relays, but no way of being certain without checking the wiring with a DVM, they might as well be inputs.
If the pins are wired to those ULN chips outputs, they should be outputs for relays.

So i tried changing the sign of the scale via Mesa CT with no change, also tried changing the analog scale so they were backwards with no change.. should i try changing something in my hal file or ini? I also tried swapping the +- leads going to the drive and that didn't work, i had to change some +- signs with the scale and encoder and basically gave me the same result but now in reverse...

You can use more than one parallel port, there are add on card with two ports on one card, so there are options.
If you can stretch the budget, the Mesa 7i96S should be perfect, 5 step/dir, 1 full spindle with feedback, i think 11 inputs and 6 outputs but i had 7i96 without the S.
And very easy to wire in as it has terminals for everything, and another and, it has a free IDC26 port you can use to wire in if you already have something wired to a DB25 port, just needs a flat cable adapter.

Tinker wrote:

When i disconnect the +-10V from the mesa card it holds position and is what im assuming is waiting on a +-10v signal.

Analog servos should turn slowly when there is no input signal.
Can you move it by hand? Does it kick back in place when released? Or does it actively refuse to move?

spumco wrote:

MODS - please mark [SOLVED]

Done, thank you.

You need to stop editing the py files, they are generated from ui file and your going to break things doing what you are doing. All edits must be performed in the ui in qt designer as things are designed. I'm going to go out on a limb and say your other issues come from allowing ai to edit files directly. A warning, ai sucks at xml and ui editing because it rarely has our understands the complete context. And if you don't know when it's hallucinating to guide it back to reality it's going to continue down the wrong rabbit hole. Just do your edits/work in designer where it builds the file correctly on start up and correctly uses all necessary code to initiate the used widgets. Your ai generated posts are mostly hallucinations where it may seem confident in it's conclusion but it's completely out in left field and no where near what's actually happening.

Ai seems wonderful until you realize it's extremely fallible.

@Lcvette - thanks, took me and the AI overlord a while to figure out why the pins were showing up in halshow but not being displayed.

MODS - please mark [SOLVED]

Is it possible you have the feedback reversed?
(check by changing the sign of the PWM scale)

Omg That would be great opportunity but I am in UK If You not require me on site and I can help anyhow I can try.

easy fix, you have not formatted the sidebar by adding a widget container or telling the current widget container to format vertically or horizontally so the widgets are there but out of view. need to add a layout format type by right clicking and selecting from the bottom options in the right click dialog.

I’ve released the first version of a small utility called CNC File Shuttle.
I originally wanted a simple way to send G-code files to a LinuxCNC computer with the same basic convenience I was accustomed to on another controller: select some files, choose the destination folder, send them, and see whether the controller and LinuxCNC are online.
CNC File Shuttle is written in Python/Tkinter and supports Windows, macOS, and Linux. I have primarily developed and tested the packaged Windows version, transferring files to a LinuxCNC computer over SFTP. macOS and Linux can currently run it from source.
Main features:
* Local folders, mapped drives, mounted shares, USB drives, or SFTP destinations
* Saved machine profiles
* Multi-file queue with destination preview
* Local and remote destination-folder browsing
* Requeue completed files without selecting them again
* Ask, always-overwrite, and skip-existing options
* SSH-key generation and guided public-key installation
* SSH host-key fingerprint verification
* Transfer progress and remote file-size verification
* Automatic read-only LinuxCNC status monitoring
* Display of the loaded program and current executing line
* Protection against overwriting the program currently active in LinuxCNC
* Failure-only diagnostic logs
The LinuxCNC status portion is intentionally read-only. It uses `linuxcnc.stat()` through a small helper installed under the remote user’s `~/.local/share/cnc-file-shuttle` directory.
It does not create a LinuxCNC command channel and cannot start programs, jog or home axes, reset E-stop, change machine power, or issue MDI commands.
Status monitoring also does not impose a blanket transfer lockout while the machine is running. Unrelated files can still be transferred, but the application blocks an attempt to overwrite the exact program path LinuxCNC reports as active.
SFTP is handled internally with Paramiko. The application supports key-based or password authentication, although key authentication is the intended normal setup. Passwords and private-key passphrases are used only for the current operation and are not stored in machine profiles.
The Windows release is packaged as a one-directory application, so Python does not need to be installed. The complete extracted folder must be kept together.
The project is MIT licensed. Source code, installation instructions, and the Windows release are available here:
CNC File Shuttle on GitHub
This is the first public release, so feedback is welcome—especially testing on macOS or Linux, different LinuxCNC versions, and installations using program-directory layouts different from mine. If something fails, the Activity Log and automatically generated failure log should provide useful information for tracking it down.
 

correct probe is stored tool with a z offset required and you also need to run the calibration to configure the trigger distance if you have not already done so.

Vielen Dank für eure Hilfe und die Zeit, die ihr für mich aufbringt. Das ist nicht selbstverständlich.
Meine xml sah schon so ähnlich aus, ich fürchte es hängt an der .hal Datei