[Kiwi]

fkaCarel
On running your program, are the X_Comp and Y_Comp figures around the correct way? When I run some figures from the curve near the stall point I think the X and Y figures should be reversed. Apology if I'm wrong.
I've got a VB program running but I'm now working on reading the Gcode file.

[fkaCarel]

It is difficult. I wrote the program to design the algorithm. It is now correctly working in another computer. The basics are that it works in a normal coordinate system. 0 degrees is to the right, 180 degrees to the left, 90 degrees is up, 270 down. So my last drawing would be moves 315-0-45 degrees. I needed to fold this in the first quadrant (0-90), which has the consequence that all results are positive. So for integration, you can unfold it to 360, or do direction detection. The whole method, in development, can get quite confusing, because now you work in essence with 2 lines of G-code simultaniously, one real, one predictive. So do a lot of reality checks.

[Kiwi]

fkaCarel
The figures can sure get confusing.
I've attached a copy of what I've done to date. It finds the stall point in the spiral code but I still need to modify to cover all conditions and yet to look at G42.

[fkaCarel]

You're reasonable up to speed. G41 or G42 are just "walked in".
X20
Y20
G42 5
Y10 ;computer knows the material is on the right hand, stop at Y15
X5 leftturn and first predictive move, goto X5 and generate internal circular interpolation 270 180 millradius, because next command is Y0
Y0 next move is X0, intersection 45 degree, X_Comp = 5, Y_Comp = 5, stop at Y5
X0
I have already extended this with the connection on circular interpolation and offcourse program flow, but I am at the moment "overnumbered" and I am suspicious that there are more situations to be covered.

[HuFlungDung]

Isn't the universe easier to explain if we say the earth goes around the sun?

I think you guys are complicating this issue.

[fkaCarel]

I think you guys are complicating this issue.

I think you are simplifying this issue.

[HuFlungDung]

Heard of Occam's razor?

There is no magic to machine radius comp. If the path can be successfully trimmed at every intersection so that the tool radius will fit, then that path can be successfully offset to a zero radius, which is what the control must do with the part path.

[fkaCarel]

No, I have never heard of Occam's razor, and I am not in the least interested in it. Kiwi had a problem, I gave a solution and that seems to be your problem. And of course there is no magic, it only has to be done and analysed.

[Geof]

No, I have never heard of Occam's razor, and I am not in the least interested in it.....

Well you should be:

Occam's razor (also spelled Ockham's razor) is a principle attributed to the 14th-century English logician and Franciscan friar William of Ockham (Guilhelmi Ockam and Guillermi de ockam in Latin [1]). Originally a tenet of the reductionist philosophy of nominalism, it is more often taken today as a heuristic maxim that advises economy, parsimony, or simplicity in scientific theories.

Occam's razor states that the explanation of any phenomenon should make as few assumptions as possible, eliminating, or "shaving off," those that make no difference in the observable predictions of the explanatory hypothesis or theory. In short, when given two equally valid explanations for a phenomenon, one should embrace the less complicated formulation. The principle is often expressed in Latin as the lex parsimoniae (law of succinctness):

entia non sunt multiplicanda praeter necessitatem,

A colloquial phrasing is the acronym KISS which expands to KEEP IT SIMPLE STUPID; I do not like KISS because I do not like to imply people are stupid. I am with Hu in this case in expressing the opinion that you have unnecessarily complicated a simple answer.

[HuFlungDung]

FkaCarel
My apologies, I am not questioning your considerable abilities to program and create logic to detect these problem areas. It is indeed a clever idea.

But, my contention is that Kiwi is making an initial mistake in applying machine radius compensation to a geometric path which was not designed for comp and is essentially garbage geometry. That is a technical term for 'it don't work out right'

Even with the trouble spots in the code eliminated, one still ends up with a toolpath that does not follow the original geometry. Perhaps this is an allowable tolerance for some work, but in such a situation, then the toolpath interpolation tolerance could also be made larger, creating longer segments, and decreasing the number of trouble spots.

[fkaCarel]

So this is all an a point of view difference. I see somebody who wants to create a spiral in his own way and gets into trouble. I don't have an opinion how the problem is generated. I could help with the detection of the problem, which he is using now. We do not all have the possibilities, to do the things the way you are used to. Maybe your way is also not the only way.

So if I publicise a piece of software and I explain how things work, I overcomplicate an answer? Maybe I brought someone else back to math, maybe I explained someone else how you can calculate things.

And for persistency: Canes latrantes ne mordent.

[Geof]

......And for persistency: Canes latrantes ne mordent.

Rats; he won this one . Babelfish doesn't do Latin

[HuFlungDung]

Canes is "dog" something as in the constellation Canis Major
"mordent" looks like a derivative of "morte" or "mortal" which means death.

Best guess, dogs will not die.

[CountZero]

"Blaffende honden bijten niet"

from google.... seems this is a popular expression in Holland. Have no clue as to what blaffende means, so my guess: blaffende dogs does not bite.

[Geof]

Canes is "dog" something as in the constellation Canis Major
"mordent" looks like a derivative of "morte" or "mortal" which means death.

Best guess, dogs will not die.

I looked up mordent on answers.com and it is a term used in music for a rapidly alternating tone so I admitted defeat.

by the way..."Isn't the universe easier to explain if we say the earth goes around the sun?" The Inquisition is gonna get ya.

[Kiwi]

Hu
I'm fairly new (3 years) to the CNC world and self taught on top of that so I can accept that I am not following established practice.
I would have thought that the actual profile and G41/42 offset would be the correct procedure and not considered 'essentially garbage geometry'.
Getting the CAD program to create a offset to use as the tool path was only a work around because of the sort of problems I'm asking about. Is this path also 'essentially garbage geometry'?
What procedure do you suggest for roughing a 2D shape and then a finishing pass?

[HuFlungDung]

Kiwi,
You are correct, the original profile is what you should use, but I guess there are qualifications that the geometry must fulfill.

Trim must be correct and all radii must be tangent and trimmed properly. Internal corner intersections are not permitted (must be filleted), and in actual fact, cannot be machined to a sharp corner with a rotating tool. I'll qualify that statement to say that if the lines that join to make the internal corner are at least as long as the tool radius, the control may be able to make the internal corner properly.

From your description of short segment code, I understood that this was already an interpolated path, and not the actual part geometry that you were working with.

If instead the originating model was composed of a spline, which was subsequently interpolated into short line segments, then such a path is no longer suitable for machine radius compensation. This is because of the reasons I gave earlier.

Some CAM software (like OneCNC) can "optimize" a chain of entities, creating arcs out of segmented chains of short lines, where they will fit, and trimming to lines as will fit. This operation has a tolerance width to it. But, such a path could then perhaps be used for machine radius compensation. Still, a check would have to be made of all radii produced, to find the smallest one, because the tool must necessarily fit everywhere on a compensated path. While some controllers may calculate where and when the tool is too large and not alarm when that situation occurs, others may not allow it and simply stop wherever the tool cannot negotiate the intersection.

[Kiwi]

Hu
The paths of short segments are the CAM interpretation of the actual part geomentry.
I have no control over the drawing of the parts. I am given a 3D CAD model of the part to machine. I don't know is these are made up from arcs or splines or whatever.
If I use a offset path (2D) as you suggest and want to do a finishing pass what is the correct procedure?
Should I create a path with a 6mm offset for a 10mm diameter cutter, do the roughing (G40) then with G41/42, set the tool radius to 1mm and run a finishing pass or is there a better way.
I am looking for a solution to my particular problem and prepared to check the code before sending to the CNC if neccessary.
FkaCarel suggestion is a positive solution which I am prepared to work with. We are doing no different to what the controller will be doing.

[miljnor]

Or another solution (that may be unwanted) is to make your tool path always to the center of the tool and calculate accordingly.

Then when cutter comp is called it doesn't have to worry about the tool diam/radius and just worries about the Wear offset.

I've as Hu is trying to address that machines all have bad ways of comp'ing tool diam/rad. So to eliminate the mostly poor cutter comp you just end up not using the diam/radius offset.

I was force to do so when buying my own machines do to the HAAS machines do a recut calculation that pretty much errors out on all pocket OP's that CAD/Cam programs generate.

Canes is "dog" something as in the constellation Canis Major
"mordent" looks like a derivative of "morte" or "mortal" which means death.

Best guess, dogs will not die.

We have a similar saying in the south (if translate properly) We will "kick that dog till it dies" or in short beat the idea to death (this is what they use in Calif).

edit: I am having a dislexic day so sorry if this is poorly worded or typed. It ain't getting any better today!

[Geof]

"Blaffende honden bijten niet"

from google.... seems this is a popular expression in Holland. Have no clue as to what blaffende means, so my guess: blaffende dogs does not bite.

I took this one to Babelfish: The translation is: "Barking dogs do not bite

So who is the barking dog, fkaCarel, Hu or me.?