Simple way to create REF points?

[crazy^millman]

Okay working on a large part with some undercuts. The machine at times has decided to move the head without warning in no predicable ways. To eliminate the possibility of a crash all index moves that are not full 5 Axis moves are done above the part in a safe place. I normally use REF points to control this if in 3+2. Most times a simple single Axis shift work nicely, but other times I want to come up exactly where a toolpath starts and end. Best thing I have found it go to the top view and put in 2D mode and set the Z where I want to be and then I just draw my points as close to those start and end. Then I go back in the operation and then pick those points. It does exactly what I want for motion, but time you do this about 30 or 100 times is gets real old quick.

Here is the question what do other do? We don't have a safe plane in 3 Axis toolpaths like we do in the Mulitaxis toolpaths. If we had that for 3+2 then I could pick the same plane using Z and done. I am pretty mimicking that same thing by what I am doing, but was hoping someone has stumbled on a better way to accomplish this without using MultiAxis linking. I don't want ghosted operations I wish we could link as an operation and not ghost the operations it works with. Curve 5 Axis would really help here.

Thanks in advance for any thoughts or suggestion anyone has or is willing to share.

[BenK]

I would love the ability to simply right-click and edit any linking move with full control. But for now I will have to keep using 5-Axis Curve and creating geometry to accomplish this functionality.

[gcode]

Have you tried 5X linking with point toolpaths to get in and out of the tight spots

[crazy^millman]

8 hours ago, gcode said:

Have you tried 5X linking with point toolpaths to get in and out of the tight spots

I don't care to do this in 5 Axis and using that way is more operations on a almost 250 op file. In the New 5 Axis toolpaths when you are doing 3+2 they now respect the active WCS. You can set the Plane to Z and a 12" value in them and the toolpath will come up to that point at the end of the toolpath. For this situation it is ideal. I could spend a week linking all the different moves in several operations if I wanted to control every single move. I don't have that kind of time and we still can't trust things since some near misses have gotten by everything expect the world class operators running these machines. I would like to use 5 Axis linking, but it is not a linking tool is a take control and lock the operation down to it's will tool. What I am doing it working great, but its just some extra steps I was hoping to speed up is all. Still much faster than back plotting the toolpath and saving the entry and exit moves to then use them for 5 Axis Curve toolpaths to link all the movements so I will just keep pressing on like I am.

Sounds like I need to put together another enhancement video and screen shots and send to QC.

[jlw™]

Let me know when you do, I'll send in the same request.

[gcode]

could you make an old school  point operation template  for a generic safety  routine

then transform the whole group about top for various C positions

Once you had the point tool paths worked out, a simple transform rotated about C would be your safe

entry exit motion

[crazy^millman]

That is just it parts may vary from 30" x 60" to 4' x 20' and have 1000 different index angles. Simple process is in the software and if I did everything through the Moduleworks toolpaths and never did anything in Mastercam then I have all I need to do it like I have suggested, but I like a lot of what Mastercam does so I want to use it better to make my job easier.

[David Colin]

Yes, that would definately be great to be able to define a safety plane (or cylinder/sphere) which Mastercam could use to automatically link operations in 3+2.

However, i'm pretty sure this safety zone was there in some previous Mastercam versions (in config job tabs). I don't know when it's been removed as i made a big jump from X3 to X9...

[Colin Gilchrist]

The Generic Fanuc 5X Mill Post has logic in it to define a "Safety Box", based on the Stock defined in the Job Setup. There is an "additional clearance" variable as well, so you can use the XYZ stock block, and make the post retract a certain distance "past" the box, before it indexes.

By activating some switches, you can force the Post to output a retract past the safety box, then change angles, and then approach the part.

It would be worth talking with Dave at Postability to see if he could implement something similar in his posts...

[David Colin]

I just reinstalled X3 to be sure and that feature was there and worked as expected... It's been removed (or relocated?) but why??

 

[David Colin]

[Zoober]

In the past, I've created a safety sphere on it's own level. Then I project points onto, and all around the sphere, in a grid pattern I think will suffice.

Then , I can select any of the appropriate points in any 3+2 plane to use as an escape point for any operation I may wish. It's easy and visual.

[Newbeeee™]

Hmmm, that safety zone feature which was removed looked cool...

I just did a 'safety zone' search in mcam and everything that came up was for multiax paths

 

[crazy^millman]

3 hours ago, David Colin said:

Yes this was a really nice feature I think about X7 or X8 this went away.

[BenK]

2 hours ago, Zoober said:

In the past, I've created a safety sphere on it's own level. Then I project points onto, and all around the sphere, in a grid pattern I think will suffice.

Then , I can select any of the appropriate points in any 3+2 plane to use as an escape point for any operation I may wish. It's easy and visual.

I like this idea. 

[Aaron Eberhard]

4 hours ago, newbeeee said:

Hmmm, that safety zone feature which was removed looked cool...

I just did a 'safety zone' search in mcam and everything that came up was for multiax paths

 

Yep, that's what it turned into..

My understanding (this was a bit before my time here, I was a reseller then) is that people didn't like it being applied across the board, and we didn't have all the information we needed to always safely do it, so it became an option like stock model.  That way you could control how THESE toolpaths linked vs THOSE, etc.    You could apply a different zone to each group, or manipulate the axis combo it'll use to get there.

I'm still not sure what Ron's aversion is to ghosting ops, but you guys know I'm always up for discussion if you want to see improvements.  Hit me up     I think ghosting the original op and including it in the "bulk" is the appropriate way to handle mass linking, as that way the original toolpath can still be updated and it'll be reflected instantly in the new link, order can be shifted without defining a new "link toolpath," etc. At the end of the day, Multiaxis Linking is designed to create the shortest safe link between two areas.  If you want something other than that, you will have to create it with a Curve or something equivalent.

That said, there has been discussions on how to handle things better when all toolpaths are operating on the same plane, so please keep ideas & discussions alive.

 

Thanks gents,

[crazy^millman]

2 hours ago, Aaron Eberhard - CNC Software said:

Yep, that's what it turned into..

My understanding (this was a bit before my time here, I was a reseller then) is that people didn't like it being applied across the board, and we didn't have all the information we needed to always safely do it, so it became an option like stock model.  That way you could control how THESE toolpaths linked vs THOSE, etc.    You could apply a different zone to each group, or manipulate the axis combo it'll use to get there.

I'm still not sure what Ron's aversion is to ghosting ops, but you guys know I'm always up for discussion if you want to see improvements.  Hit me up     I think ghosting the original op and including it in the "bulk" is the appropriate way to handle mass linking, as that way the original toolpath can still be updated and it'll be reflected instantly in the new link, order can be shifted without defining a new "link toolpath," etc. At the end of the day, Multiaxis Linking is designed to create the shortest safe link between two areas.  If you want something other than that, you will have to create it with a Curve or something equivalent.

That said, there has been discussions on how to handle things better when all toolpaths are operating on the same plane, so please keep ideas & discussions alive.

 

Thanks gents,

Aaron, thank you for chiming in your input and thoughts are always appreciated.

Ghosting ops when the tree is collapsed you don't have a good idea if they are dirty. I am still not comfortable linking 20, 30 or 50 or 500 operations with one linking process. Safety Zone was a good idea and had it been given more time I think it would have solved a lot of these issues I am running into. I got a way to get this done just like others who do this. I need the tool to retract from a 3+2 operation at the start and end straight up in the Z plane. Safety Zone did this I used it that way before it was removed. I am a control freak with my operations and my code. Multi-Axis link is an operation not a control and I and I want my operation to stay clean when linked to 2 operations that is just what I want to see. Problem is what happens when an operation is dirty? Is one, 2, 5 or 50 of them dirty if one gets dirty? Do they all need to regenerate if one gets dirty? To many things going on and not making things depend on others if they don't need to from a clutter standpoint is one of my thoughts. The other is posting or simulating that operation. How can I do that if it is ghosted? Stock models have now added process ghosted operation, but again if it were not ghosted then I wouldn't have to make sure that is checked. That adds another thing to think about and make sure I have included. Hard line to walk making software especially one made for type A personalities, but when you are doing the type of work we are doing everything that make your life easier is great. 

[Zoffen]

I'm not sure this will help with Ron's current situation here but i thought i would add this as it relates to the safety zone.

I have used the "stock" Tab in the operations manager to define a "safety" like zone using the rectangular stock option (i imagine you could use other shapes if you set it up correctly). I then customized my post, since all the data from those fields are available, to use this area for automatic 3+2 retracts. This took some heavy post customization but once i had it figure out i would define a box as a safe area and my post would automatically find the sphere the part will spin around in and automatically move it out of the way. This was on a Table/Table 5 axis(variaxis) so moving out of the way was a simple z move. It might be harder to implement in a head/head or other 5-axis style machine.  This saved me SOOOOOO MUCH TIME as i didn't have to worry about retracts at an index move my post would do it all for me. Some of my programs were for fairly complex hydraulic manifolds with some over 500 ops with LOTS of indexing so this was a real lifesaver. Plus if you reorder the ops you don't have to think about retract and approach moves as they are all done AUTOMAGICALLY!

If anyone is interested i'll try and dig up how i did it. Hope this helps someone!

[Aaron Eberhard]

Zoffen - That's a pretty cool way to ensure you're clear!  So your post was essentially fitting a spherical safety zone and filling in interpolated moves between angular positions?  That's nice   The issue people get caught on is that without some sort of controlled move from A to A45, THIS control moves XYZ THEN AB, THAT control moves all XYZAB at max rate, etc, etc.  It results in the motion being "unpredictable," and that's when you knock a spindle into a fixture.

Ron - On the backplot side, you can change whether backplot will show a ghosted path (Backplot settings > simulate disabled posting).  At the end of the day, we can agree that while Safety Zone was indeed a zone, but it wasn't very safe the moment you starting using multiple planes:

On the ghosting front, I believe we'll have to agree to disagree on the Multiaxis Link front    You can see when an op is dirty: 

Here's a tree with an dirty op with the branch expanded, then the same thing closed, all inside a safety zone:

You can see the state reflected pretty accurately and immediately.  No matter what you do,when you're slinging around 5-600 toolpaths like you regularly do, it's going to take some time to keep it organized, I don't see ML being an impediment to that.  If ML didn't do it the way that it does, you'd double the toolpath count because you're just using it as a curve toolpath generator that can't change the original toolpaths, nor would it allow easy editing/reorganization of the original toolpaths because if you changed the order or clearance parameters of the source ops, there'd be no "link" to the linking op.

I don't know that I would have relied on X6's safety zone to get me safely around a part, either.   It was nice for making sure your retracts were all to same plane, but doesn't really help you as soon as different planes are involved.  I think a different mechanism is needed in that case, and safety zone wasn't the best vessel for it, and the fact that it always just centered on the WCS selected wasn't very helpful.

Much like a stock model, transform, or anything else with dependencies, you're probably better off to do multiple Multiaxis Links if you need control, as opposed to fitting all 500 ops into one zone.

Good discussion thus far!

[Zoffen]

25 minutes ago, Aaron Eberhard - CNC Software said:

Zoffen - That's a pretty cool way to ensure you're clear!  So your post was essentially fitting a spherical safety zone and filling in interpolated moves between angular positions?  That's nice   The issue people get caught on is that without some sort of controlled move from A to A45, THIS control moves XYZ THEN AB, THAT control moves all XYZAB at max rate, etc, etc.  It results in the motion being "unpredictable," and that's when you knock a spindle into a fixture.

 

 

Yup. Mostly i just defined the stock box as my actual stock in my 5 axis vice and that turned out to be a good safety zone for the program with very efficient rapid moves.

I even had logic to determine different clearance zone z heights depending on where the rotaries were acutally at.

Example: a move from A0 C0. to A-90 C0 required less height than from A0 C0. to A-90 C90 if the part was a more rectangular then cubed. I programmed alot of 12"L x 6"W x 6"T ish workpeices so this did save some extra time not having to go all the way up as it would if i spun it around the long side. I hope that makes sense.

Typical Part:

I had a switch in the post to control the output of rapid positioning combinations. But, it was 99% of the time set to Z+(Clearance Move to safety Z) --> XYAC --> Z-(back down to part)

This seemed to work very well for the setup and parts that we ran on that machine.

[BenK]

What we really need is individual control between every toolpath. At the moment the only way we can do this is by using a Curve 5 axis toolpath. I really don't care for the way multiaxis link does it. 

[huskermcdoogle]

32 minutes ago, BenK said:

What we really need is individual control between every toolpath. At the moment the only way we can do this is by using a Curve 5 axis toolpath. I really don't care for the way multiaxis link does it. 

I couldn't say this any better way.  I think the main thing that would help ML would be to have it be kinematically aware based on MD, and allow for axis limits to be set.  Even if we had to select what rotaries we had available, and or could select a chain to follow, that would be awesome.

[Aaron Eberhard]

40 minutes ago, huskermcdoogle said:

...I think the main thing that would help ML would be to have it be kinematically aware based on MD...

I agree with this statement

 

BenK - Not to be flippant, but "I don't really care for the way multiaxis link does it," isn't exactly a helpful statement.  I can't go to the developers and just say "change things."  Well, I guess you I could, but you probably wouldn't like that, either. 

I know it's a lot of work on your part, and I do appreciate all of the feedback you give us, but if you make a statement like "what we need is individual control between every toolpath," my immediate questions are going to be "why," "when," and "can you send in a sample part that describes the motion you don't like and why?"  As you mentioned, there is a way to get individual multiaxis control between every toolpath, so we have to figure out why that amount of effort is preferable to the automated way.

[Zoffen]

Im not sure why this hasn't been mentioned but it looks like ICAM does exactly what we are looking for....

 

[jlw™]

That turns me on a little bit.  I'm not gonna lie, I just Googled that.