Adaptive Clearing wasteful extra motions and how can I optimize this carve?

[sprior913]

<Pardon is this post shows up twice, the first time I posted it it seemed to disappear>

I'm learning never to say "It could run more efficiently, but I'm only planning to carve a couple of them so it doesn't matter".  Last Summer I designed a workshop sign and I just can't seem to stop myself from giving them away.

I originally did a 2 stage carve with this design (1/4" straight bit followed by 1/8" ball nose), but switched to just using the 1/8" ball nose because it didn't make much difference and was less fuss with my X-Carve to not change bits.  So I'm doing a 3D adaptive clearing followed by a parallel finish pass with just the 1/8" ball nose.

What I'm seeing during the adaptive clearing operation is lots of vertical lifts and cutting moves which don't really seem to be cutting anything and are wasting lots of time.

  This 2 minute video shows this at 30 seconds and 75 seconds in.  I've seen the same behavior with straight bits, so I don't think it's any effect of the ball nose in the calculations.  I can also see these moves in the CAM simulation and they don't appear to be cutting anything there either.  I've tried changing a few operation parameters and haven't found anything that eliminates these.

I'm also looking for ways to make this carve go a lot faster, I think it's wasting a lot of time being fussy with the adaptive clearing which I've got set to leave 1mm extra material in both directions so the parallel pass has something to cut.  In the attached picture you can see the parallel pass on the left and the adaptive clearing on the right and clearly the adaptive pass is doing more work than it should, but I haven't found any parameter changes that claim to speed it up much in the machine time estimate.

So far I've tried:

Just for the adaptive clearing operation:
starting machining time: 1:21:52
stock to leave from 0.5mm to 1mm: 1:09:16
tolerance from 0.1mm to 1mm: 1:08:09
minimum axial engagement from 0mm to 1mm: 1:08:09
minimum retraction policy: 1:06:25

Attached is an export of the project, can anyone give me ideas how to speed up the adaptive clearing while still using the one bit?  This carve currently takes almost 6 hours of wall clock time despite the machining time estimate in Fusion of just a couple of hours.  I think that lifts and travel moves on the X-Carve are a lot slower than Fusion estimates.

[sprior913]

Here's the finished sign, took 7 hours to carve, about half that time was the adaptive clearing pass.

[daniel_lyall]

Just a quick note before I look at your file sometimes 3D and 2D adaptive are the wrong toolpaths to use on a router, quite often a 2D pocket will cut it a lot faster and just as good on a router.

[sprior913]

In this case it's a true 3D carve - the background is wavy.

[daniel_lyall]

That it is, you could do a search the same looking file is in this part of the forum it was brought up last year. there are a couple of tricks needed to get the cutting times down.

You almost had it, with your very first op the one you crossed out.

[daniel_lyall]

@sprior913 here's your file back you will still need to check it over to make sure it's cut depth is ok for your machine.

What I did was restrict the depth of cut useing the bottom height plane in the first adaptive cut, then the second one I did the same thing plus move the top hight down, this makes it cut at one height level and not go all over the show.

The parallel cut I used, I set rest machining and set it to minimum retraction in the linking tab. these 3 toolpaths a roughing toolpaths the finish toolpath is up to you.

[sprior913]

I think maybe I loaded too many topics into one post and this got a little confusing.  There were a few things going on.

The first is those extra cutting motions performed by the adaptive clearing operation which are shown in the video I posted.  I believe that these may be a bug in the CAM toolpath generation, so I was hoping that would be confirmed for submission to get it fixed.  This is part of what I was referring to when I said the carve took too much time because the toolpath does cause the machine to spend a decent amount of wall time doing these air cuts.

The second point is that the adaptive clearing operation is taking WAY too long and that appears to be because it was spending too much time doing the fine terracing that can be seen in the picture.  It wasn't clear how to tell Fusion to leave a mm or two of extra material, but not to be so precise about leaving EXACTLY that amount on the non-flat surface that so much time is spent making that perfect for the roughing operation.  daniel_lyall gave me a possible approach (and thank you) which involved two flatter operations, but I think I've found an answer that is easier - in the adaptive clearing operation the rough stepdown was set to 2 or 3 mm, but the fine stepdown was set to maybe 0.1mm - that's where I think the extra detail came from.  I think if I set that to 2 or 3 mm (the same as the roughing value)  I'll get the roughing operation I need without sweating the unnecessary extra detail which will be handled by the finishing operation.  The trick with figuring this out is that the machining time estimates are WAY off for the X-Carve, so in fact making changes to the fine stepdown value didn't make significant changes to the estimated machining time, but I suspect the differences would be a lot bigger in real life, and that's another issue.

The third issue is that there is no way to configure the retract and travel speed for the machine into Fusion's CAM, so for machines which have slower travel moves the machining time estimates get worse and worse for different operations which makes it harder to figure out optimizations like mentioned above without actually trying them.  Adaptive clearing seems to have the worst estimates because there are a lot of retract and travel moves - the estimate might be 75 minutes, but the reality is closer to 180 minutes!  Finishing passes like the parallel operation I've got at the end are pretty close to the estimated time because there is is very little extra travel.

So I'd really like to see if anyone can confirm that my point 1 is a toolpath generation bug or some other setting I've got wrong, and I'd really like to know if we can hope for an improvement on the third issue in the near future - it'd make figuring out the second issue a lot faster.

[kb9ydn]

---

@sprior913 wrote:

The first is those extra cutting motions performed by the adaptive clearing operation which are shown in the video I posted.  I believe that these may be a bug in the CAM toolpath generation, so I was hoping that would be confirmed for submission to get it fixed.  This is part of what I was referring to when I said the carve took too much time because the toolpath does cause the machine to spend a decent amount of wall time doing these air cuts. 

---

The biggest problem I see is that you need to set the "Stay-Down Level" to Most.  This tells the tool path generator to minimize retract moves as much as possible.  You may also want to increase the value of the "Maximum Stay-down Distance".  This lets the tool travel farther between cutting passes before having to do a retract move.  These two things alone will cut down the number of retract moves significantly.

Considering that your machine has very slow Z axis moves I would second Daniel's suggestion to try 3D pocket clearing instead of adaptive.  The adaptive strategies were really designed with higher speed machines in mind; the idea being that you take lighter cuts at faster speeds instead of the more traditional method of heavier cuts at slower speeds.  Because of this, 3D adaptive especially tends to be very heavy on retract moves.

Something else that might help too is to use a larger step over and smaller step down.  Especially if the higher level step downs can be done all at the same height, that limits the necessary retracts even more.

The machining time estimations could definitely use some work, although I thought there was actually a way to set the retract feed rate used for the calculation.

C|

[sprior913]

OK thanks.  Daniel had mentioned 2D pocket, not 3D, but I understand from what you just wrote that 3D pocket might work better with my machine than 3D Adaptive in most situations.  I'll try the other parameters you mentioned as well.

But aside from that, the air cuts I mentioned above still look useless to me no matter what machine they're running on, does anyone think this should be checked into?

[daniel_lyall]

The changes I did made the cut time for all 3 toolpaths to be less than your first cut, if you have a look through what I set that will show what settings to use.

Also that machine of yours is not very rigid at all, ( I will not put big cuts in a example for a small machine) if you do to big of a cut it will just chater the heck out of the machine, lose position or break the thing.

What @kb9ydn says is correct the Adaptive cuts job is to rip out as much material it can with constant cutter loads, If it only has a few mm to travel it will move to a new spot, there is no bug It can be improved in the back end.

A pocket toolpaths job is to cut within the pocket but sinces it is a 3D cut you will need to use a 3D toolpath hences 3D pocket.

If the sign had a flat bottom it would get cut very fast as there is only one Z height change to be done, with a 3D shape it has to do a lot of 3 axis moves there's no way around it other than getting a big machine that could do the cut in one depth useing a parallel toolpath.

[kb9ydn]

---

sprior913 wrote:

But aside from that, the air cuts I mentioned above still look useless to me no matter what machine they're running on, does anyone think this should be checked into?

---

In some situations the adaptive strategies can produce funny cuts that seem like they really aren't cutting anything.  It happens when there are little cusps of material left over in the calculations, which you will always get to some extent due to how the adaptive strategies work.  That particular spot you showed in the video didn't actually look that bad, and I suspect a lot of it was due to an excessively small fine step down.  It's hard to tell without actually looking at the tool paths in the CAM simulation.

That said, the adaptive strategies are not perfect and it's worth at least bringing it up if you see one that looks especially bad.  The developers are always trying to find ways to improve tool path generation.

C|

[Anonymous]

Although the post is pretty old I am curious to know if the problem of aircutting was solved or identified how it is caused. I have exactly the same issue with adaptive clearance cuts. After the rouging is done as how I intended the pass to use it keeps on passing the same lines over and over again to, yes maybe, take mini steps down to smooth. Since I will use a parallel or other pass to really smooth the 3 dimensional object I really don't need any of these as it looks like aircuts. Really, I could reduce the adaptive clearance cut by 75% if it would stop acting that way. I have it in with all possible settings as far as I know and with all objects so clearly others must know what the original poster meant and showed.

Hope someone has a clear answer on the case.

Thanks all 😉

[daniel_lyall]

You would need to show all your settings supplying a model always helps To do this Go to File -> Export and save as a .F3D Archive File and attach it to your next post, if you can not post the file you can PM it to me or say in your next post.

This post is to far out of date to have any real comparison to today a lot has changed in cam and fusion in general.