IJK problem

[bogusmill]

We are experiancing a small .0002 divot on some dog bone specimens for tensil tests. The divot is at the transition from arc to straight line X move on both sides. The straight line to arc transition is perfect. I'm gussing that it's caused by rounding numbers in MCX2. The .0002 is in the .nc code wether I use radius setting for an R output or Delta start to center for an IJ output it's there. If I use the absolute arc setting the arcs match but the Haas says it's invalid and wont cut it. Any ideas what's wrong?

[gcode]

quote:

---

Any ideas what's wrong?

---

Do you have the lookahead parameter set???

Its been so long since I've run a Haas that

I don't remeber what it is???

[Guest]

quote:

---

Any ideas what's wrong?

---

quote:

---

Haas

---

and

quote:

---

.0002

---

Just a guess though

[mkd]

FYI: it's setting 85 (corner rounding). G187 E.XXXX can also help.

.0002" is at the limit of the accuracy for a Haas in excellent shape. how would you even measure that small of a divot? A CMM might even have trouble.

[bogusmill]

Look ahead is set at 100 blocks. I can put a good .0001 indicator in the spindle at 0 for the horizontal cut. Move to the good end and the needle only rises. Move to the other end and the needle drops .0002 before starting to rise up the arc. The straight line to arc transition is perfect, why can't the arc to straight be perfect.

 

I created an arc and line, then mirrored them about Y, then mirrored that around X. So the code should have the same digits, only the the + or - should change, but that's not what I'm getting. It appears MCX2 is rounding one arc up and the other down.

[bogusmill]

Here is a code example.

 

N114 M08

N116 Z.11

N118 G94 G01 Z-.002 F45.

N120 X1.8918 Y-.8372 F8.

N122 G03 X.625 Y-.6097 I-1.2668 J-3.4128

N124 G01 X-.625

N126 G03 X-1.8918 Y-.8372 I0. J-3.6403

N128 G01 X-1.857 Y-.931

N130 G00 Z2.8

N132 X1.857 Y-.9308

N134 Z.11

N136 G01 Z-.002 F45.

N138 X1.8918 Y-.837 F8.

N140 G03 X.625 Y-.6095 I-1.2668 J-3.413

N142 G01 X-.625

N144 G03 X-1.8918 Y-.837 I0. J-3.6405

N146 G01 X-1.857 Y-.9308

N148 G00 Z2.8

N150 Y.931

N152 Z.11

N154 G01 Z-.002 F45.

N156 X-1.8918 Y.8372 F8.

N158 G03 X-.625 Y.6097 I1.2668 J3.4128

N160 G01 X.625

N162 G03 X1.8918 Y.8372 I0. J3.6403

N164 G01 X1.857 Y.931

N166 G00 Z2.8

N168 X-1.857 Y.9308

N170 Z.11

N172 G01 Z-.002 F45.

N174 X-1.8918 Y.837 F8.

N176 G03 X-.625 Y.6095 I1.2668 J3.413

N178 G01 X.625

N180 G03 X1.8918 Y.837 I0. J3.6405

N182 G01 X1.857 Y.9308

N184 G00 Z2.8

N186 M09

[Midwest]

I would analyze your line work..........

 

 

Maybe put on the FTP

[iowajim62]

quote:

---

I would analyze your line work..........

---

+1

also check arcs for centers and sweep

[bogusmill]

How do I send it to FTP site?

 

I have checked the lines and arcs, but as I said it's one line and one arc, mirrored around Y the result mirrored around X. They have to be the same radius and centers and lengths. Just different combinations of + and - depending on which XY quadrant they are in.

[Rick Henrickson 65]

can your HAAS read to .00001

you could change your mcam tolerace to mills

I doubt it is the code causing the problem though.

 

have you ever ran a ballbar test on a HAAS Hmmm it is quite an eye opener

 

that is a transition from single axis to multi axis move I doubt you are going to get much better then that

[Midwest]

quote:

---

How do I send it to FTP site?

---

email it to me............I'll be gone soon for the day.

 

 

[Midwest]

 

 

When you analyze you should see this window does the X and Y stay the same on each arc????? Also your rad and dia should match.

[iowajim62]

try posting with a stock no edits made post and see if it codes the same as ur post

[bogusmill]

Haas can't read to .00001 Rick, but it can read .0001 as the .nc file is written. Never ran a ball bar test but have run an ISO test where you cut angles and circles and bore holes to determine any weakness or inaccuracy in ball screws or alignments. I know the Haas has its limits but my issue is why don't similar arcs output similar code. I can live with what the Haas cuts if I can get good code to feed it.

 

Midwest, I've sent you the file, your probably gone by now. Check it tomorrow.

[vstreet]

I think you should skip the mirror part and draw the whole thing out. I think I have seen little screw ups like that before with translate or mirror.

[Surface]

I wouldn't think a dog would care a about a .0002 divot. You must be dealing with some very discriminating dogs.

In rare instance, I've had arc-rounding errors. They were easily fixed by forcing the post to break arcs into quadrants. Just a thought.

[bogusmill]

Quote: I've had arc-rounding errors. They were easily fixed by forcing the post to break arcs into quadrants. Just a thought.

 

The code I posted earlier is with arcs broken into quadrants. I've used radius and Delta center to start settings, even checked every arc check box there. Absolute setting got me symetrical code but the Haas wouldn't run it giving me an invalid arc error.

 

And yes Surface, these are very discriminating dogs. The dogs in the friction stir welding lab have achieved welds up to 96% of base metal strength so any little imperfection machining may cause the dog bone specimen to break someplace it shouldn't.

[Midwest]

Bogusmill, what are you using for a post??

 

It would be better if the file you sent me had the tool path.

 

 

I'm getting different numbers than yours with a .721 dia cutter as per your email.

 

 

try the mpmaster post

 

mpmaster

 

 

I'm not getting your .0002 problem in my post.

 

 

[mkd]

Switching to Metric can theoretically be more accurate; .001mm is finer than .0001inch.

 

this doesn't change the mechanical accuracy which is .0002" on a good day, on a Haas.

 

very discriminating Canine indeed!

[bogusmill]

Midwest, I just sent you the file with the toolpath. The post is version 11 MPMaster with a few mods by Trevor at San Diego CAD/CAM. I tried the toolpath right before sending it just now and got the same code I posted.

 

I've tried filter and tolerance setting changes but havn't tried metric yet. Might give that a spin.

[Midwest]

Bogusmill, It's doing what you are telling it to do...............see pic below.

 

 

 

 

You are using multi pass with the last pass @ .0002

 

 

[bogusmill]

Duh?! What was I thinking? Looks like your right. Got so much going on maybe I'm losing my mind. I need a vacation.