3D Roughing Tool Paths

This section is for useful tips and tricks for Aspire

3D Roughing Tool Paths

Postby garyalan59 » Sat Jul 20, 2019 4:58 am

It seems to me that the Aspire coding of a 3D roughing tool path is very inefficient. What I mean is this:
When I have a lot of surface detail, and some deep cuts as well, the program spends an inordinate amount of time retracing the already completed surface detail while stepping through the tool path to find the low spots to carve.
The current tool path I'm running has a projected run-time of 3+ hours, of which at least half (guesstimate) is spent retracing already carved surface details.
Wouldn't it be more efficient to "skip" the already carved surface details (say with a rapid z-retract) and "jump" right to the deep carving that needs to be done?
I'm using Aspire 64 bit, version 9.517 (Build 5169.5658.906), Built Apr 13 2019.
I'm using a 3D Raster roughing strategy along the X axis.
garyalan59
 
Posts: 3
Joined: Tue Jul 03, 2012 2:27 pm
Model of CNC Machine: Legacy Arty 58

Re: 3D Roughing Tool Paths

Postby TReischl » Sat Jul 20, 2019 12:49 pm

That is the nature of the Raster strategy. We sort of miss paying attention to the terms. Raster means back and forth over an entire area, like the raster scan on old TV's.

A recent thread also discussed this "issue". There is another strategy called Z Level. That is more like a pocketing strategy and does reduce roughing time considerably.

However, in my experience Z level can leave large stair steps which may not be desirable when the finish cut is applied. This can really become an issue when using an aggressive roughing approach.
http://www.tedreischl.com

Low Profile CNC Router Vise
User avatar
TReischl
Vectric Wizard
 
Posts: 2715
Joined: Thu Jan 18, 2007 6:04 pm
Location: Leland NC
Model of CNC Machine: 8020 Build 48X36X10 RP 2010 Screenset

Re: 3D Roughing Tool Paths

Postby garyalan59 » Sun Jul 28, 2019 1:53 pm

Do you have a link to the previous discussion?

I tried the Z-level raster - I don't like it for similar reasons to the X- or Y-raster. In the case of the Z-raster, the roughing path appears to hog out area based on the underlying structure. It looks like its going to carve the ultimate shape at each Z-depth, but ends up hogging out all the material in a "walk-in-the-park" strategy - "I'll carve this out here, then --- squirrel --- go over there, then --- shiny object --- come back here..." It's very frustrating to have so much machine time tied up in repetition. :?

I heard or read some time ago that "skipping" the already completed background graphics was the key to reducing the size of MPEG video files. Couldn't a similar strategy be employed in the raster technique? If the detail is already there, why repeat it? Just because we say raster, do we have to accept it's inefficiencies?

I don't have the necessary programming skills to implement such a strategy. I'm sure Vectric does. How do we get suggestions to them? :idea:
garyalan59
 
Posts: 3
Joined: Tue Jul 03, 2012 2:27 pm
Model of CNC Machine: Legacy Arty 58

Re: 3D Roughing Tool Paths

Postby TReischl » Sun Jul 28, 2019 7:10 pm

garyalan59 wrote:Do you have a link to the previous discussion?

I tried the Z-level raster - I don't like it for similar reasons to the X- or Y-raster. In the case of the Z-raster, the roughing path appears to hog out area based on the underlying structure. It looks like its going to carve the ultimate shape at each Z-depth, but ends up hogging out all the material in a "walk-in-the-park" strategy - "I'll carve this out here, then --- squirrel --- go over there, then --- shiny object --- come back here..." It's very frustrating to have so much machine time tied up in repetition. :?

Mine does not do an inordinate amount of squirreling around? Is it the most efficient? No, but it is not horrible either. I don't why you think it is going to carve out the ultimate shape at each Z depth, it is a roughing routine, so it hogs, that is what it is supposed to do.

I heard or read some time ago that "skipping" the already completed background graphics was the key to reducing the size of MPEG video files. Couldn't a similar strategy be employed in the raster technique? If the detail is already there, why repeat it? Just because we say raster, do we have to accept it's inefficiencies?

Video files are two dimensional, x and y if you will. 3D carving is obviously NOT 2d like a video file so it is not a trivial matter to do what you suggest, things like slopes and curved surfaces need to be accounted for. That is why the 3d raster version works the best but wastes a lot of time.

I don't have the necessary programming skills to implement such a strategy. I'm sure Vectric does. How do we get suggestions to them? :idea:


It is one of those software things. . . it LOOKS like it should be easy because we engage our super computers between our ears and the solution is soooo obvious. I have tried engaging my own super puter on this issue. Put it this way, if we can do it by using our own brains and doing it more or less manually then it should be able to be done in software. Give it a try sometime, I have. . . . after a couple of hours I understood what some of the issues are. It is not pretty.

Tell ya a fast story from years ago (feeling blabby again today). Had a customer who was drilling circuit board holes. His boards could have like 3000 holes in them. Of course his software ran all over the place wasting time and costing him money. He asked me to come up with a solution. I set up a simple test on my brute force algorithm that had 50 holes. Well, it ran for a few hours so I figured it was stuck in an endless loop I had created. Hmmm, nope, it was proceeding as it should have. I let it run all night and into the next day, still not done. I double, triple, quadruple checked my algorithm. Still no go.

Then I remembered an article I read in Scientific American, I think Aug '82 about the traveling salesman problem. Yanno, shortest route a traveling salesman can travel. Turns out that these sort of problem are geometric in nature, maybe logarithmic? Anyhow, the problem I was working on would not have finished in my lifetime given the computing power of the time. Anything over 7 locations tipped the scales.

So how come they can do it now (think google maps)? Parallel computing, they break the problem down into much smaller subsets and reiterate through that a few times and voila, you have the shortest route with stops in between.

Doing what you suggest is actually much worse than the simple traveling salesman problem.
http://www.tedreischl.com

Low Profile CNC Router Vise
User avatar
TReischl
Vectric Wizard
 
Posts: 2715
Joined: Thu Jan 18, 2007 6:04 pm
Location: Leland NC
Model of CNC Machine: 8020 Build 48X36X10 RP 2010 Screenset


Return to Aspire - Tips and Tricks

Who is online

Users browsing this forum: Google [Bot] and 7 guests