Machining not matching simulations

[Anonymous]

I have a Tormach PCNC 1100, and have used fusion many times to make parts with no issue.  Now, The part comes out wildly skewed, twisted on the X axis.  Machine is not the problem, tested old parts, no issue.  Using pathpilot on the machine, simulation looks perfect there as well.  Called tormach, they said they have had customers calling with same issue on fusion.  Tried older post processor, helped.  Tried old post processor that worked before, and it was better again, but circles were very oddly egg shaped (extreme), and lead ins were in the part, which is bad. 

Drawing looks fine, simulation looks fine, simulation in cnc software looks correct.  Old parts tested were within .001" accurate, and ran a second time with same results.  New part with 3 different post processors gave 3 wildly different outcomes. 

One oddity is, I cannot select my "machine", since they only offer other models, like the 1100M and not the PCNC1100.  Does it matter? No idea.  3d printed the part, zero issues. 

[seth.madore]

Woah, that doesn't look good. 

Can you share your Fusion file? File > Export > Save to local folder, return to thread and attach the .f3d file in your reply

What machine you select doesn't matter (at this point in time, I wouldn't even select one). What does matter is what post processor you use. You are using the latest Tormach post, or is it one that you've saved to your Cloud? If it's an older one, can you share that here?

[Anonymous]

Here you go.

Tried newest post, last post of 2020, and last post of 2019. 

[seth.madore]

Was there any update to the Tormach side of things?

[Anonymous]

As far as the machine, no.  After trying the first time, I updated pathpilot (software that runs the machine), and it didnt fix anything.  Tormach was aware of the problem, they said it happened recently (not sure when), but fusion 360 customers have had the same issue. 

What we thought was, its something in the conversion to gcode, not the post processor itself.  I say that because, the old post processor did not work either, and they used to.  The file is fine, since it 3d prints perfectly.  Machine cuts old files from fusion no problem. 

Its a pretty dangerous issue, the simulations on fusion and pathpilot both show one thing, but machine does something wildly different.  Pretty scary.  Luckily its only messing with the X axis in my case.  University of Maryland has been fighting the same problem when using fusion. 

[seth.madore]

Can you share the NC code (not the Fusion file) from a job that you've run with success?

[racegrafix]

Yes, no problem.  All 3 of these are examples of working files. 

[seth.madore]

Okay. On the "Frame Insert PerimeterRED" and the "Turbo Mount Bracket Outer" can you post out that code with the current post?

I selected those two because they are the simplest bits of code to compare between old and new

[racegrafix]

Here you go.  Used Post processor marked "tormach" so, that should be newest, and had the known problem. 

[seth.madore]

And you know for certain that these new bits of code will not run proper?

There is no difference in the output code, with the exception of the added note of reduced rapid feedrates due to Personal License restrictions. In addition to the note, there are spots where G0 has been changed to G1. Otherwise, they are exact copies of each other

If the problem still persists with this new code, the issue is at the machine/controller, not Fusion

[racegrafix]

Old code worked flawlessly, new code is untested. I only did those for your comparison so far. 

[seth.madore]

Right. The point I was trying to show was that there were no significant differences between old and new code. It's only a couple data points, but it suggests (to me) that the problem isn't likely on the post processor or Fusion side.

[racegrafix]

Made a similar part, with same problem areas, and machined it perfectly.  Everything was correct.  The only difference was it was machined in delrin as a quick test.  If it was the machine,  it would make me believe it was a drive slipping (x axis) and maybe the delrin didnt give enough resistance.  But, on the bad part, even cutting air it had the mistakes, and there was a angle cut across the face of the part when the drive shouldnt have even been moving. 

[racegrafix]

I ran many tests, and with your help I can say without a doubt, it was not fusion 360 issue.  For my final test, I used a easy to machine block of plastic, on the problem file.  machined perfectly.  So it is indeed machine issue, and not fusion 360.  Just wanted to follow up and say thanks for all the help, it really sped up the diagnosis of this odd issue. 

[seth.madore]

That is good to hear! However, I'm still curious what the issue is, as it would be good to file that away in my memory should a similar issue arise with another Fusion user

[racegrafix]

So finally found the problem.  This machine has a billet machining bed on top of the regular table. The bolts holding the bed became loose enough it would move only in X axis under certain circumstances.  So the bed would slide one way, and sometimes back to correct location, and it would zero out fine.