Crash not showing in simulation

[frankTMJQD]

I have a rouhging path where I use 3d adaptive to clear off stock. I run half of the part and turn in 180 degrees and do the other half with similar paths. I use a copy of the first 3d adaptive operation for the second job and it crashes in the exact same spots on the stepdowns but not on the finishing pass at the "bottom" depth. I run "minimum retraction", 10% stay down level, 0.5mm lift height and 3000mm/min no-engagement feedrate. On lead in/out I have 1mm set on hor/vert lead in/out. I suspect it's related to lead in/out or linking but the error isn't showing in simulation like you can see on the attached picture. I've always trusted simulation before so this thing kinda got me scared since I don't know about crashes like this before they happend.  

Down to the right on the picture of the part it shows that the first adaptive operation has 10mm roughing stepdown so the error only apears one time. On the second operation I set roughing stepdown to 7mm and then it hit the part in 3 stepdowns instead of just one from what I can tell. 

The machine this code is run on is a 3 axis machine with Fanuc OM-C controller if that matters.

[HughesTooling]

Is this something you can share, if you can export as a f3d and attach. One thought is have you enabled model selection in the op and only select some faces and not all of the body?

Mark

[frankTMJQD]

Yes. It's very messy tho but if you look at the last setup named "LONG_third_op_side2" it crashes in the first adaptive toolpath there. Attaching  the f3d file here. 

[HughesTooling]

Using the standard Fanuc post it looks OK if I backplot using NC Corrector. I have seen problems with very short arcs, on the post dialog what is the minimum cord length set to.

Can you upload the gcode from just the adaptive cut, you'll need to ZIP it to attach.

Mark 

[frankTMJQD]

Minimum chord lenght is set to 0.01. Everything is as default except this: 

Preload tool : no

Separate words with space : no

Use sequence numbers: no 

Attached is the nc post that crashed. 

[HughesTooling]

---

@frankTMJQD wrote:

Minimum chord lenght is set to 0.01. Everything is as default except this: 

---

That might be the problem. The sizes on the post dialog are in mm so 0.01mm, try set to 0.05 also set minimum circular radius to 0.05.

The back plot of the g code you attached looks good. Some controls do a full circle if you program a move to the position it's at and they see a very short arc move as a request for a full circle, I think this might be the problem.

Mark

[frankTMJQD]

Thanks. I googled the problem you suggested and it seems like it's the problem I'm experiencing. I tried going through the gcode but I can't find out where it messes up but it must be something like this.

What do I sacrifice by setting it to 0.05mm? Will the post still make the code so that my CAM accuracy is kept? 

Might be slightly off topic but I also notice that it makes many small arcs where there is no reason to not just use one arc for the whole movement. Is there any way of making lead in-out one arc instead of many small? Or other movement that doesn't matter. Never noticed this before but if it keeps chopping up arcs where it doesn't need to this problem can appear again.  

[HughesTooling]

---

@frankTMJQD wrote:

What do I sacrifice by setting it to 0.05mm? Will the post still make the code so that my CAM accuracy is kept? 

---

All that happens is the arc is turned into lines, so a 0.05mm arc length is probably going to end up as one or two short lines.

---

Might be slightly off topic but I also notice that it makes many small arcs where there is no reason to not just use one arc for the whole movement. Is there any way of making lead in-out one arc instead of many small? Or other movement that doesn't matter. Never noticed this before but if it keeps chopping up arcs where it doesn't need to this problem can appear again.  

---

 

I think you had the tolerance and smoothing set to 0.1mm, a tighter tolerance will smooth to arcs better. The calculation time will be longer but try 0.01 tolerance and 0.02 smoothing or if you don't mind the calculation time try 0.005 and 0.015

Mark

[frankTMJQD]

Ok then it's a good thing. Is 0.05 enough? I'd rather bump that up to even 0.1 or 0.2 if that could save me from more crashes in the future. 

I always thought that higher tolerance allowed for more slack (running more in arcs) than lower numbers there. When I went from 0.1mm tolerance and 0.1mm smoothing to your suggested 0.005/0.015 it increased the moves from 4922 to 5373 and from 1108 arcs to 1569 arcs. 

Edit: I now increased the min cord lenght and circ radius to 0.1mm in post and generated it again. It gives me exactly the same amount of arcs and linear movements as before (post I crashed with) with 3536 linear moves and 1108 arc moves. I kinda fear that the problem isn't solved then 😞 

edit2: increasing min cord lenght and circ radius to 0.2mm gave the same result also. Bumping it to 0.5 increased linear moves and reduced arcs kinda like expected. 0.5 sounds high tho. 

[HughesTooling]

Although the code size might increase with the tighter tolerance it is smoother and I'm pretty sure you'll get less short arc moves.

This image is with your tolerances, note the sharp corners and the yellow loops.4

And here with a tighter tolerance. If you open the attached file I duplicated the adaptive cut and tightened the tolerance so you can select each one while viewing from the top so you can easily see the difference.

Not sure what to recommend for minimum cord length, you probably only need to see 4 or 5 less arcs to fix the problem.

Mark