"(Built-in) Tolerance" settings make no change to post

[GalenYMW8K]

I have a very old control, Fanuc 6M. I have been doing everything I can to help my control speed up. The code that slows my machine down the most, by far, are linearized moves. Sometimes these are not avoidable... think Pencil, or an adaptive lead out (3D arc) , a micro-lift on an adaptive path, etc...

So, with some help, I found in the Generic Fanuc post a section of code that basically says "IF (allow 3D arcs= true) [some code] ELSE "linearize(tolerance)"

Thinking I had it nailed, so I can open this tolerance up and not ruin some of the linearized moves that are a cutting path, ruining the tolerances of my part, I change the "linearize(tolerance)" to "linearize(tolerance2)" and near the beginning, added "tolerance2 = spatial(0.01, MM); " This number is arbitrary, I will optimize it later with testing.

I was excited to try this but after posting, I noticed it made no change to the code. Thinking I simply coded it wrong, I went back to the generic fanuc post (that has not been modified) and in the Post Process window, tried posting with it all set to default. Then, I posted it again, but I change "(Built-in) Tolerance" field from .002 (default) to 0.10. This posted the identical code. I then tried 1.0 , .5, 100, 10.0 , etc, and they all post the exact same code.

I tried a few different post configs, other than the fanuc ones, and they all post the exact same code no matter how the tolerance field is set. Not only are the sequence numbers the same, but the XYZ moves are identical, even the .NC files are identical number of bites.

Help! I am really excited to see how much time I can pick up by tweaking this!

[Laurens-3DTechDraw]

That is weird.

Could you share a part maybe?

Did you know there is a Fanuc compact post special for older controls?

http://cam.autodesk.com/posts/?p=fanuc_compact

[GeorgeRoberts · Accepted solution]

Hello,

Thanks for posting! It's good that you are carefully looking into the post settings and outputted code. The tolerance setting in the post processor is used for linearizing arcs if needed. As a simple test, try setting the property 'allowedCircularPlanes' to 0, this will then linearize all arc moves based on the tolerance you have set. 

I tested this on the HAAS post by first setting the tolerance to 0.001 and then changed it to 1, the outputted code with 0.001 has 1000 more lines than the 1mm tolerance. Perhaps your code isn't changing as no arcs are being linearized? 

Hope this clears a few things up!

[GalenYMW8K]

Thank you Laurens, I will check that post out

George, 2 things- 1, you were right. I had made enough changes in the tool path I was testing this on that there were no longer any 3D arcs left. I added it back in, and tolerance does make a difference now. I feel stupid for this, but the reason I didn't consider was because I assumed that the post tolerance would also affect linking moves (think 2D adaptive on a micro-lift linking move), it's a large 3D arc. What I am realizing now, is that the smoothing and tolerance settings in the tool path configuration are what change this.

But that brings me to a second question. What, exactly, does the post tolerance (linearizing tolerance) affect? Literally only lead ins and lead outs?

Secondly, does the linearizing tolerance have the exact same affect as the tool path tolerance? If so, can we get even more compression and quality path by adding a smoothing to the linearizing tolerance?

I'm sorry for making such a dumb post but I was under the assumption that smoothing and tolerance for the tool path was only affecting the cutting portion of the path and that the linearizing tolerance would affect everything else (linking, ramping, leads, etc). So, naturally, I would have expected the linearizing tolerance to affect any tool path.

[GeorgeRoberts]

Hello

Thanks for the reply, I'm glad it cleared a few things up. The tolerance set inside of the post processor handles geometry which is created by the post (linearized arcs), these can be output from the lead-in moves or cutting moves (even linking moves in adaptive if they output an actual arc). To test, try creating a simple extruded circle and add a contour toolpath around the edge. You will notice that when you post, the path will be made up of several arcs. If you set the allowedCircularPlanes to 0 then post again, you will get a fully linearized toolpath (even for the cutting moves)... Changing the tolerance will then affect how many points are created to build up the linearized arc.

My example below has a linearized toolpath at 0.002mm and another at 2mm, you will see the difference 😄

I'm not sure about the smoothing option, I always believed this was used to create more arcs where possible... Something that couldn't be done in the post (as the arcs are linearized, so there is no point).

Cheers

[GalenYMW8K]

So, the tolerance set when we create the tool path, is again processed (twice from the original geometry) by the post processor tolerance? Seems odd..

Thank you very much for your time and your help, it is greatly appreciated.

[Laurens-3DTechDraw]

@GalenYMW8K

---

@GalenYMW8K wrote:
So, the tolerance set when we create the tool path, is again processed (twice from the original geometry) by the post processor tolerance? Seems odd..

---

Well only when your machine can't accomplish the move by the calculated path. So when the control can't do a 3D Arc the postprocessor uses this tolerance to make code the machine can interpret. 

But in theory, all tolerances you see could add up to the total error of the path.

[Laurens-3DTechDraw]

I've wanted to do this for a long time.

So I made a tolerance representation on how far the toolpath could be off form the wanted path due to the tolerance.

As you can see the blue line is spot on in the middle so it's usually far better than the possible theoretical error.

@GeorgeRoberts probably something you would like to store somewhere for future reference.