Bug in 4.3 2011-11-23? No pencil trace at the inner border

[wdaehn]

From a hexagonal shape (84mm diameter) I cut another (80mm dia) hex shape, removing 2mm of the material. The clipping pane is set to -0.01mm and 6mm because the top sheet of the material does not need milling, it is fine already.

What happens is that the inner hex shape is cut not entirely, the closest line is 4.3mm to the edge although the cutter is a 6mm (3mm radius). So I would have expected that either the z-level finishing step does that or at least the pencil trace.

roughing: 3mm depth, 0.4mm stock to leave. Obviously this step does not cut out the inner shape as this would be just 2mm deep. Fine.

z-level: 2.5mm depth, 0 stock. Not sure if the z-level should cut out the border of the inner shape or not. Is it following the same logic as roughing - only cut what is at full depth?

planar: all default except contour mode, it obviously deals with the inner hex, but does not go into the very borders. The closest it gets is 4.3mm.

pencil tracing: all default, it does not touch these borders at all!?!


Workaround is simple, set the depth for the z-level to 1.90mm, now it goes almost all the way down. works only because I have one depth. If the model would have different depths somewhere, it wouldn't work this way.

Am I missing something? Cutter is 6mm flat out of the default library, with no stock and 0.03 tolerance - I double checked.

(Failed to attach the zip file with 4MB size with the Rhino Model??)

[svenakela]

What happens if you:
* Remove the clipping planes?
* Make a Z-level toolpath with angle limit set to 90 degrees?
* Make a planar with angle limit set to 0 degrees?

I could have a look on the file if you PM me a link where to download it (or if you send it to me by e-mail).

EDIT: Your zip file is too big to attach to the forum.

[wdaehn]

I checked your suggestions, still the same thing. With a different project and in 4.2 I haven't noticed that behavior. There the Pencil Tracing went through all those areas where neither the planar nor the z-level finishing cut to.

A simplified version of my model, just a Boolean diff of one hex volume minus another, there all does work as expected.

[svenakela]

Werner sent me the file and we've found that the strategy had to be changed a bit, the restrictions of roughing and cutter tolerances made the toolpath generation very "strangled".
The distance measured in the screenshot Werner attached is a result of roughing, which will never touch a vertical wall on purpose. An extra Z-level with very small cutter tolerances will do the rest, including the very tight holes.
Or, go for a contouring in one cavity and copy that path to all the other cavities. Will result in a much faster path as well.

Happy cutting guys!

[wdaehn]

Thanks Sven, really appreciated. However I am still not sure the problem actually is on my side. Maybe, but I am not sure.

First of all, I did do roughing + z-finishing + planar + pencil tracing. Not sure why that got lost in the file though. Secondly, the question is less of how can I make it work and what would be an even better strategy than blindly go for the 3d modes as I did. It is the question on why a strategy would return the wrong end result.
I am worried that with other, more complex, scenarios like having multiple cavities staggered at different depth, the workarounds will not work well anymore. Or you might not even notice you are going to mill something with the wrong result, the cavity being of wrong size/diameter.

Also note, a similar - not identical - piece did work with 4.2 version I am stating. No proof though.

I have built three similar pieces as simplified test case and used different milling strategies. All contain all 3D steps, roughing, z-level, planar, pencil trace. All is left default, 6mm flat cutter being used. Only change is the roughing is set to 0.4mm stock.

It seems with the left two objects where I used a stock model, the pencil trace did even tell that no toolpath can be created. And clearly there is one needed when you look at the cavity. In my actual example it did create a toolpath but none at the inner cavity.
The right hand top object was without a stock, just a standard box+cutter_radius. There the pencil trace is doing what it is supposed to.

So although I get a different problem in this simplified case, it still is around pencil trace and stock model.

[svenakela]

We have a reduntant mail and forum conversation. Lets drop the mail.

See my input below, after your points (taken from the last e-mail). I saw you posted a model at the forum after I did this one so I'll post it here too for other peoples info. It's a simplified model with paths attached.
I don't know what tolerances you need but the attached model has a tolerance of 0,01 mm after pencil tracing. I didn't put any big effort into the model (have to work with my own machining too...) but I have no problem with pencil nor remachining. You can use the toolpaths by copying them into your large model.



· Roughing I fully understand, it goes through the outside only. There is no need for the roughing inside (except the holes) as the cavity is not deep, that can be done at the finishing step at once.
Correct.


· Z-level finishing I understand as well, if I would limit the depth to the -2mm, the z-level finishing would probably work. But just imagine there are staggered cavities, -2mm, -3mm, -1.75mm,… Should I really do multiple z-levels or set the cut depth to a tiny value? Can’t be the solution. So with my current setting I am not getting the cavity borders cut with the z-level finishing, that’s fine then – the pencil trace should do that then.
Well, it all depends how you want to do it, and most of all how the model is shaped. In this case you're doing 3D shaping on an almost vertical/horizontal model (except one corner). Yes, there are some extra work in small cavities and that is needed in any CAM system.


· Planar is obvious as well, it does not care about the very border, hence the 4.3mm.
Yes, correct


· And the Pencil Trace is really where I would call it a bug. The pencil trace is supposed to cut all the edges, shouldn’t it?
Well, it does. Check the attached model. I think you have to get the knowledge of configuring, that's where it all is - for every type of CAM system.



So really, my question is less on how do I get it to work, it is more if there is a bug in the pencil trace. As no matter what settings I use, the result should be the correct one. Not ideal but correct.
As I said, I see no bug.



However there are some statements I don’t grasp likely due to lack of knowledge:

· When I use a clipping pane at 0.00mm, it will cut the entire surface, even the top sheet, which it shouldn’t. Hence I used -0.01mm.
The Top Clipping plane states the MAXIMUM level, "from here an below" that is. Remember, there's no thickness at all in a Rhino surface (or any modeler surface actually). If you zoom into infinity you still can't measure a thickness of a surface. That's why the top clipping plane makes effect if you put it 0,01 mm below the surface. Then you exclude the top hundred of a mm, That's the feature of a clipping plane.

· It seems I failed to understand the impact of the tolerance field. Why would there be an impact in my model?
Taken from the help file:
Tolerance Saved with each cutter and is used for toolpath calculation.
As written before (in an e-mail) you are making small cavities with small cutters close to the size of the cavity. If you have a large tolerance, and when there's only a mm left in space a tolerance of 0,03 mm is quite uge. 6 % of the space left between cutter and edges actually. The tolerance is used in path calculations and now your are aiming for a pretty small volume so yes, it matters.


MfG,
Sven

[wdaehn]

Can you check my test file? It is quite simple. Why do I get pencil tracing with a block model correctly, with stock model I don't?
That is why I believe of a bug.


What you showed is how a toolpath much superior to mine would look like. Currently I am asking why there is no (useful) Pencil trace generated if a stock model is used, with a block model there is everything fine with the same settings.

[svenakela]

Originally Posted by wdaehn Can you check my test file? It is quite simple. Why do I get pencil tracing with a block model correctly, with stock model I don't? That is why I believe of a bug. What you showed is how a toolpath much superior to mine would look like. Currently I am asking why there is no (useful) Pencil trace generated if a stock model is used, with a block model there is everything fine with the same settings.

That wasn't what you asked for in the beginning, but that really is a bug. I've been talking to Joakim and there's a service release coming with this defect corrected, should be out today or tomorrow as it was already planned with other fixes.

Now, I still don't get why you want to run an almost vertical/horizontal model with 3D milling, this is way faster cut with a 2,5D contouring strategy.

Anyway, i salute you for finding this bug and not giving up in your trials to come to a conclusion.

[wdaehn]

Originally Posted by svenakela Now, I still don't get why you want to run an almost vertical/horizontal model with 3D milling

That's easy to answer: Lack of knowledge/experience. I will start looking into that.

Thank you so much for your help, really appreciated. Learned actually quite a lot via this thread conversation!


PS: Took me a while to figure out it all boils down to the pencil trace, hence the shift in topics throughout the conversation.