Advanced Search

Is anyone is actively developing Fusion 360 post processor?  I ran into an issue that someone should document if it isn't documented somewhere already.  

I noticed that there MUST be a lead-in defined in Fusion 360 or the torch won't fire (M3 command won't be generated) with open chain geometry.  This might also be true for closed chain geometry - didn't have time to look at that.  I was working on some art work that didn't have closed chain geometry and couldn't figure out for the life of me why the M3 command wasn't being generated.  Took a bit of experimenting to figure out how the settings worked, or, in this case didn't.

I've been resistant to give SheetCam a try a) because Fusion 360 has served me well enough up until now and 2) I didn't want to cough up the $145.  Maybe it's time to give it a whirl but it really would be great if there was a well supported Fusion 360 post processor.  Just say'n.

Actually in your case a simple single toggle component will do the trickt, just replace the hal signals input-togl and output-togld with your mesa input&output addresses:

loadrt toggle count= 1
addf toggle.0 servo-thread
setp toggle.0.debounce 5.0

net input-togl toggle.0.in
net output-togld toggle.0.out

For anyone following this thread later, www.oshcut.com offer a very good service in the US, I used them for my home machine components and other projects. No relation to oshpark.com PCB manufacturing apparently.

I run plasmac with a fiber laser now and while I have made a few modifications to the settings, the underlying plasmac component runs the same as it did with a plasma cutter. My THC is tracking to a capacitance value instead of voltage, which goes to 0 when the nozzle touches material (mainly tip-ups) during any movement. This usually leads me to needing to a do a reverse jog sequence and restart the cut - which works well.

The issue I am having is that the cut recovery panel is not consistently being enabled when paused_motion (plasmac state 19) is triggered. I can't narrow down why cut recovery is able to be used sometimes and not others. I have a component that monitors my capacitance (nozzle to work distance) and will pause the program if it goes below a threshold, or actually touches the nozzle to the work piece which prevents a crash.

To get cut recovery and reverse run to work, I usually (>90%) have to click the "REV" button, click Resume, click Pause, click "REV" again, click Resume, and then the "REV" button actually will back up the program. Other than doing this sequence, the "REV" button doesn't do anything, even while being in plasmac state 19.

I have tested cut recovery with both my custom component that will pause the program, as well as manual pausing and it doesn't make a difference for when cut recovery is not bugged.

A weird thing:
In the Value field there's 'change to gcode', this is the behaviour:
feature is Circle by Center, Diameter defaults to 38.1
right-click Diameter value and choose 'change to gcode'
value changes to 45
right-click menu now shows 'revert to original type', choose that
value changes to 1143
after this, 'change to gcode' and 'revert..' make the value swap from 45 to 1143 and back.

putting a different start value in e.g. 700 or 70 - change to gcode still puts the value to 45, revert to original puts it to 1143 and so on.

another test:
while the Value field is 'changed to gcode' the calculator style input field doesn't appear and this value doesn't change. 'revert to original type' allows the calculator to appear again.

A

Hello all,
It's good to see you working on NativeCAM, I use it all the time and have got to the point where I need to update to lcnc 2.9.x and the lack of ncam is the main problem.

I followed juergen-home's instructions with a couple of changes:

step 3. copied to /home/freemoore/linuxcnc/ncam/
step 4. open console in /home/freemoore/linuxcnc/ncam/NativeCAM-master
step 5. sudo bash ncam_debsetup.txt
python3 ncam.py --ini=/home/freemoore/linuxcnc/configs/sim.axis.vismach.5axis.table-rotary-tilting/xyzbc-trt.ini --catalog=mill

After this I have run each further step down to 14 and reproduced the errors and results in Juergen's post, but without getting tool library to appear in ncam.

Next I copied Giovanni's ncam.py from above in place of the one in Bogdan's download.

I still don't have a tool library, but this is outputting a program to the simulated 5-axis machine config of lcnc that I've set it up with here. And the program is running OK.

Not sure I know what would be useful to you but I'm happy to test new versions of your code here!

great to see the familiar ncam on a 2.9 installation, it gives me hope

cheers, Andy

you can do it like thisloadrt flipflop names=flip-is-on
addf flip-is-on servo-thread

net is-on       halui.machine.is-on  =>  flip-is-on.set
net is-on-reset motion.is-all-homed  =>  flip-is-on.reset
net go-home     flip-is-on.out       =>  halui.home-all

Thank you! This works like a charm. Machine is homed after enabling and ready to go.

I wanna see this actually moving a machine. Interesting about the CSV/CSP, but I think it makes sense in relation to what someone else said in another thread, that you are getting the reported position one frame late, so it always appears to lag, even though it "isn't".
 

I think I saw this post as well. But the following error was really high so I thought it can't just be that. But maybe I am wrong here. If anyone figures out how to properly run these in csp mode I would be very interested - no need for PID-Tuning in LCNC.

Sorry for the multiple posts, but editing in this forum is horrible.
 

Okay, Andy, let me give you a feedback.
I added the term to combine the signal to the output. Nothing changed.
Then I switched over to AXIS, because of the marvellous HAL Meter.
Compared signal with pin, pin didn't change.
Searching a little bit and found "PWMGEN enable" not defined.
I assume that was because I didn't took the pins cw and ccw but only the enable for starting the spindle.
Okay, changed, spindle ran. Full speed.
HAL Meter says, Pin and signal are identically and changing. Measuring always 10V output.
Check at Talla83, found
"setp hm2_7i96s.0.pwmgen.00.scale [SPINDLE_0]MAX_OUTPUT"
Tadaa! It runs...
Now I have to implement those changes into my gmoccapy_hal_file.
Thank you for your help and your patience.
BR Rex

Hello, thanks for sharing your configuration. I took a quick look and there is one thing I don't quite understand
net x-output <= cia402.2.velocity-cmd
net x-output <= pid.x.output
Is the direction above reversed? It should be
net x-output => cia402.2.velocity-cmd

Yes ths is a mistake on my end. But as mentioned it is for readability only.

This is given by:pitch of your ballsrews / Number of Encoder pulses per revolution.
In this case I would guess:
2^23 / 5   = 1677721.6
i.e. this tells you how many encoder pulses you get per physical movement of your machine.
This would foe example be for an sfu2005 ball screw with a 23bit encoder (23bit = 2^23 = 8388608)
(I'm pretty new to all of this so please correct me if I'm wrong)

Correct. And additionally for the velocity scale I have used 2^23 / 60 (as in 60 seconds) for the Lichuans and 2^17 / 60 for the spindle.
 

It is this DVS servo motor from Aliexpress. It works with the same XML configuration like the Lichuans. I've just changed the product and vendor ID and removed the IO section. It has a 23-bit absolute encoder but I have configured it as 17-bit incremental atm:

de.aliexpress.com/item/1005005929510135....gatewayAdapt=glo2deu

And the spindle:

de.aliexpress.com/i/1005006232728097.html?gatewayAdapt=glo2deu


 

Hello,

tks for the reply. I`m using MESA 5i25 + 7i77, with glass scales.

This enable setting its original from Induma, would you suggest to have individual axis enable (three outputs) or just one enabling all axis is enough? What would be the output name in PNCConf?
Also, I've one output to energize the axis controllers, dos PNCConf can handle it or should I use classic ladder?

I was reading some topics here about axis PID calibration and found interesting that the controller runs heavy on P (20..40) and almost no I (0) or very very little D values. Is this usual? Kinda different from other PIDs I've used in industry.

Sorry for the amount of questions, LCNC is new to me and Im very excited with it!

7i76e_7i76x1_Dpl.bit is the firmware needed for a 7I76E and multiple THCADs

The load resistor for Ohmic sensing determines the sense current
I would not use higher than 25K

Note the if you use a lower value you need to make sure that you do not exceed
the resistors power rating with the open circuit plasma voltage.

   

OK, I found an embarassing math mistake that has fixed my forward transformation. But I'm pretty mystified by my inverse tranformation. All of my constants appear be being set to zero.  

For example, in the following cablekins.comp file I have defined:

const double LEFT_DRUM_X = -100.0;
const double LEFT_DRUM_Y = 250.0;

These are being used in the forward transformation and working properly.  However in the inverse transformation i put in print statements:
 
debugPrint("Debug:     LEFT_DRUM_X = %f, LEFT_DRUM_Y = %f\n", LEFT_DRUM_X , LEFT_DRUM_Y);  

and the output is:

Debug:     LEFT_DRUM_X = 0.000000, LEFT_DRUM_Y = 0.000000

Sorry my C sucks.. any idea why this is going on?

thanks! 


 

I think your bit file should work (assuming your external RS-485 interface
automates RS-485 driver enable control)

The fact that you could not get the RS-485 interface on TB2 to operate
suggests a test issue.