Dust Collection Shoe with independent Z control

More
03 Apr 2023 04:42 - 03 Apr 2023 04:48 #268144 by finnstrom
I love Marius's channel and would like to replicate his dust collection shoe . It has a "Z axis" that is independent of the tool's Z axis movement, enabling finer control of the dust collection and air nozzle placement. However, in my design the dust shoe "Z axis" isn't referenced from the gantry, but from the tool plate itself.

I have attached two views of my dust shoe (with two tubes) and spindle with collet (no tool). One with the dust shoe extended, and one with it retracted for changing tools.

I have a homing reference switch on the dust shoe axis as well as the Z axis. When the dust shoe is fully retracted and Z is fully retracted (moved up) I know the distance from the bottom of the dust shoe to the top of the workpiece. My question is, what is the best way of controlling the dust shoe axis independently? My thought has been to add another real time thread that controls the dust shoe, but this thread would also need to have constant knowledge of the “raw Z value” (which I think is Z value in machine coordinates?) as well as the state of program (eg. cutting, or changing tools).

Another idea I had was to add some extra post processing of the gcode, but I feel much less comfortable with gcode (and understanding the various caveats/intricacies) than I do with multithreaded C/C++. My worry (with this latter idea) is that I miss what coordinate system is being used and muck up the correct dust shoe height. More importantly, it feels cleaner to not have to mess with the gcode for controlling dust collection.

Thanks in advance!
Finn




 
Attachments:
Last edit: 03 Apr 2023 04:48 by finnstrom. Reason: Images didn't attach the first time.

Please Log in or Create an account to join the conversation.

More
03 Apr 2023 11:51 #268168 by andypugh
I think that you can do this in HAL. If you choose the W axis to control the shoe, then connect the motor position command to joint.8.motor-pos-cmd - joint.2.motor-pos-cmd

(You could also do it in a kinematics module, but that might be a little overdoing it)

To subtract one number from another in HAL you use the "sum2" component with a negative gain for one input.

Please Log in or Create an account to join the conversation.

More
03 Apr 2023 11:59 #268169 by andypugh

I think that you can do this in HAL. If you choose the W axis to control the shoe, then connect the motor position command to joint.8.motor-pos-cmd - joint.2.motor-pos-cmd
 

Actually, with LinuxCNC 2.8 you would generally configure Axis W to be Joint 4. (It had to be joint.8 in 2.7 and earlier. 
The following user(s) said Thank You: finnstrom

Please Log in or Create an account to join the conversation.

More
04 Apr 2023 02:33 #268229 by finnstrom
Fantastic, that's an even cleaner solution. Thank you!

Please Log in or Create an account to join the conversation.

More
04 Apr 2023 19:31 #268284 by finnstrom
Looking at the HAL solution further, would that not fix the W axis to always be dependent on the Z axis?

I don't always want the dust shoe to be extended fully, such as when changing tools it needs to be retracted, or moving rapidly I'd want to pull it off the work piece slightly more.

Could I create something conditional in HALTCL? Also, what's a good way to "know" from LinuxCNC when the CNC is cutting? I was considering remapping M7.

Thanks
Finn

Please Log in or Create an account to join the conversation.

More
07 Apr 2023 23:20 #268562 by finnstrom
Seeing your post regarding IF statements in HAL  it seems this is not possible. I take it HALTCL is more of a parametric way of setting up HAL, not a method of adapting HAL during execution.

I did however come across MUX  which takes in two float values and assigns either to an output based on a pin input (mux2.N.sel). I'm thinking I can map an M1xx code to mux2.N.sel in HAL. That way I can use gcode to control if:
  1. axis W (joint 4) is controlled based on a negative sum of axis Z (joint 2) like you mentioned, or
  2. to fully retract (zero or home)
Next step is how to control that behavior. Initially I was thinking about remapping M7 and M8 but that doesn't seem possible . Maybe I'll remap O (spindle control).

A couple of questions:
  1. I've read about position prediction . Does the HAL sum2 mapping solve for that? Ie. if Z decelerates in anticipation of a new gcode command, W will match that motion.
  2. Are there any potential caveats with remapping the O code? I've read some about execution sequencing and modal groups but don't fully understand.

Thanks again
Finn

Please Log in or Create an account to join the conversation.

More
15 Apr 2023 08:56 #269075 by andypugh
(Catching up)

Yes, my solution would leave the shoe hovering a fixed distance above the table, whilst the Z moved about. Which would be fun to watch, at least.

As you have surmised, you could use a MUX to switch that axis between W control and a fixed position. Maybe it would make sense to use the vacuum-on signal to switch the behaviour.

You don't need to re-map the M7 / M8 signals, you can just use them as-is and repurpose the HAL pins iocontrol.0.coolant-flood / coolant-mist.

You probably need a bit of extra logic so that the shoe-axis homes properly. I think I would do that by adding extra select bits to the mux.

ie
sel-0 = coolant
sel-1 = W-homing

Then connect inputs 1,2,3 to the PID and 0 to a fixed retract position.

Note that you probably want to disable the PID when it is not in use and you will have to "lie" to LinuxCNC about the W-joint position feedback to avoid following-errors. I would just short-circuit the motor-pos-cmd to motor-pos-fb and leave LinuxCNC unaware of what the shoe s doing.

Please Log in or Create an account to join the conversation.

More
23 Nov 2023 19:18 #286360 by gardenweazel
I have also created a shoe using  the W-Axis. I'd be interested in knowing how you implemented yours.
I have the W-Axis riding on my Z-Axis with printed parts that I created and a NEMA17 driving the shoe.
I can home the W-Axis without any problem but, I'm not certain of next steps other than what functionality has been mentioned 
in this thread.

The one thing that is different on my setup is the W-Axis is actually mounted to the Z-Axis so it will move with the Z-Axis as well as move
up and down. My intent is to simply use an offset but, like mentioned would like to have over-ride control for tool changes, etc....

Any best practices or feedback for such implementation?

Please Log in or Create an account to join the conversation.

More
03 Jan 2024 20:46 #289695 by finnstrom
Thanks Andy.

GW:
You've got pretty much the exact same setup as me re. W on Z.

I took a long break on dust collection but am back at it and only a couple weeks from testing. I'll post my results.
The following user(s) said Thank You: gardenweazel

Please Log in or Create an account to join the conversation.

More
09 Jan 2024 22:05 #290312 by gardenweazel
I haven't tried running a job yet but, what I am going to test is as follows:

I wrote a few lines of code into my POST that simply negates the position of my Z-Axis and applied that to the W-Axis.
I included the min and max positions for my W-Axis and if the setting goes beyond those parameters
I just zeroed the W-Axis. Realistically my W-Axis should never go out of the min and max.

I'm sure that I'll need to tune it or even make changes.

Please Log in or Create an account to join the conversation.

Time to create page: 0.461 seconds
Powered by Kunena Forum