Raster Disaster (Or - How to avoid Conventional direction)

[eitantal777]

When I do 3D Finish with an engraver, I want to avoid the Conventional direction at all costs. Or else, shavings/grains are still attached to the piece. This problems happens no matter what overstep % I choose, and no matter the tip size of the engraver. (This is what I call Raster Disaster). This problem is entirely a climb/conventional issue. When you climb, the grains get shredded and chopped. When you conventional, the wood grains are carved away, but leaving them connected at the end

There are solutions to this problem, but none of them are perfect.

The obvious one is to do 3D finish on offset mode, but there are 2 problems: with it. 1 - Witness marks, and 2 - I'm limited to a very simple vector choice. Vectric this quirk where half the time it goes in the wrong direction. Do a 3D finish in a selection of 2 vectors (Square-within-a-square). It will split the work into 2: One half is in the right direction, the other half is in the wrong. A way to solve this problem is generate 3D finish for the outer square, and delete the GCode for the inner square that I don't want to machine

Another obvious solution is to do a second run at 90 degrees, to take care of these pesky shavings. It is time consuming, but it works

The ideal solution to this problem is a special raster: Go all the way up on the right side, rapid to the left side, and go all the way down, rapid to the right again. The problem is no there's mode for this. The way to achieve this would be to post-process the GCode.

Unless someone knows of a perfect solution, I'll write a python script that achieves the ideal solution: Raster Disaster GCode comes in, Good Raster GCode comes out

[scotttarnor]

I use a tapered ball nose for all my 3D finish tool paths and have no issue , is there any particular reason you want to use an engraving bit?

[eitantal777]

Sorry, I meant exactly that: "Tapered ball nose", is what I use. I just called them Engravers for some reason.

This is what I use: https://www.amazon.com/2-Flute-R0-25-R1 ... 08867Y5CM/

If you don't run into such problems, Perhaps the problem is I use a low quality bits. May I ask what toolbit are you using?

[Rcnewcomb]

I use similar bits but uncoated.

What RPM are you running?
What is the feed rate?
What is the plunge rate?

[adze_cnc]

...Or else, shavings/grains are still attached to the piece...When you climb, the grains get shredded and chopped.

The last two 3D finishing cuts using raster toolpath that I made were is cherry and a poplar core plywood. No "shavings/grains" attached. I was cutting across the grain which works so much better then with / against the grain.

[eitantal777]

...Or else, shavings/grains are still attached to the piece...When you climb, the grains get shredded and chopped.

The last two 3D finishing cuts using raster toolpath that I made were is cherry and a poplar core plywood. No "shavings/grains" attached. I was cutting across the grain which works so much better then with / against the grain.

I go with the grain, not across. That could have been it.

There's a reason why I almost always go with the grain: If the work piece is rectangular, chances are the long side is the grain. To minimize changing directions when I 3d-finish, I choose the raster to line up with the long side. In this particular job, it was a rotary piece. In this case, and I figured the rotary axis is much less accurate with regards to things like backlash. I wanted to minimize travel on the rotary axis

[eitantal777]

I use similar bits but uncoated.

What RPM are you running?
What is the feed rate?
What is the plunge rate?

It varies, but usually this:
RPM: 18k
Feed rate: 90 ipm
Plunge rate: 1/2 Feed rate

[ZipperHead55]

It seems to me that you have a solution, and are in search of a problem.

I do a lot of 3D carvings, and I rarely encounter this "raster disaster" that you speak of. I use hardwoods (cherry, maple, oak, exotics) exclusively and only the "softer" hardwoods (western figured maple) cause me any grief. I use 1/4" ball nose for roughing, and then 1/16" TBN for larger flat(tish) areas, and 1/32" or 1/64" TBN for very fine detailed areas. Any areas that have "fuzzies" are cleaned up fairly easy with a brush (stiff plastic, brass or steel, from least aggressive to most) and sanding mops. Occasionally I break out an Xacto-knife and/or dental picks for tough areas.

I think someone who has more experience with the nuances of the software can likely help you with your "solution" that doesn't involve using an external script (I assume Python scripts can't be run in Vectric software, just "gadgets" using LUA). There are many strategies that I don't understand (self taught with Vectric products) that could be contained within the existing software, so it might be beneficial to wait until one of the maestro's weigh in with their experience (before coding a solution).

Allan

[adze_cnc]

In this particular job, it was a rotary piece

Ahh, the plot thickens! Why wasn't this mentioned at the start?

[eitantal777]

It seems to me that you have a solution, and are in search of a problem.

My problem is very real, or else I wouldn't post this discussion on the forum. The "fuzzies" are the problem, and they don't always come off easy.

Looks like fuzzies are the worst when:
* You go in the conventional direction
* You go with (or against) the grain
* You use softwood. I used pine
* Your model is detailed. This prevents easy removal of the fuzzies

Xacto-knife and/or dental picks for tough areas.

Right, this is how I ended up salvaging my work, that would otherwise go in the garbage. Even so, with the fuzzies removed, the end result is still worse, than if I used the offset mode

1/4" ball nose for roughing, and then 1/16" TBN for larger flat(tish) areas, and 1/32" or 1/64"

Your process is different than mine. I usually do this:
* 1 Roughing with 1/2" end mill
* 1 Finish pass with either 0.25mm, 0.5mm or 1mm TBN
It is possible that I get "fuzzies" because of a big jump from 1/2" to ~0.5mm, as the finish tool ends up removing more material

[eitantal777]

In this particular job, it was a rotary piece

Ahh, the plot thickens! Why wasn't this mentioned at the start?

I didn't mention it because I didn't think it was relevant, as I had the same effect happen to me in normal jobs.

This time it was relatively tame, actually, despite being rotary. The most severe raster-disaster happened to me when I made a picture frame. On that attempt, the step-over was large and the ball radius was small. I therefore concluded that the effect itself always happens, just with varying degrees of severity.

[dealguy11]

Wood species has a major effect on fuzzies. Some pines (especially eastern white pine) are extremely prone to fuzzies. I've never gotten a good cut from eastern white pine and avoid it like the plague. The pith between the growth rings is nearly impossible to cut cleanly.

You may want to try other alternatives. Some of the harder pines, redwood and cedar can do well. When it comes to hardwoods, I've had poor luck with some soft maple and some poplar. Usually have decent results from hardwoods like cherry, walnut, sassafras, birch, butternut and a variety of others. I don't carve oak because I don't like the graininess. For what it's worth, I like to the cut with the grain because it helps to mask the scalloping from the tool. I can see where cutting across the grain might help with fuzzies, but it leaves a scallop pattern that has to be sanded out. A sharp tool also makes a difference.

As far as climb versus conventional, how does it even matter on a raster strategy carve, or is that your point? Half the time it's cutting one direction and the other half the time it's cutting the other so every project has a combination of climb and conventional. Making every pass cut in the same direction would make every project take at least twice as long to cut.

My point is that while cut direction may affect fuzzies, my experience has been that wood species has a bigger effect.

[eitantal777]

As far as climb versus conventional, how does it even matter on a raster strategy carve, or is that your point?

That's exactly my point. If I choose the offset strategy, and strictly adhere to climb, I get no fuzzies. Reason being, the fuzzies are shredded before they even have a chance to become fuzzies. In normal raster, you necessarily go in the conventional direction 50% of the time, and you get penalized with said fuzzies. Raster doesn't have a setting of climb vs. conventional

You may want to try other alternatives

I don't want to opt for another wood material. I choose pine because it's cheap and plentiful. Selecting another strategy is free.

Half the time it's cutting one direction and the other half the time it's cutting the other

That's precisely the point. This is why I call it a raster disaster. I offer a raster strategy that is both "raster", and also climb 100% of the time: Start in the middle. Do a stroke on the left going up, then rapid to the right, and go down, then rapid to the left, start-over. This strategy does take more time than normal raster because of the rapids. To minimize this penalty, I could do small clusters of those. Vectric doesn't offer this strategy, so I may need to write a post processing script, that takes in a normal raster, (aka raster disaster) and modifies it to my suggested strategy

Note: The above strategy is not the same as an offset strategy on a square. The offset strategy on a square is a rectangular spiral, whereas mine looks like the "PAUSE" icon, and the two pause stripes drift away from each other

[Rcnewcomb]

For 3D rotary work look at the Milo Scott Continuous Turning toolpath.

It is substantially faster than traditional raster or offset, and is less likely to show up backlash in the rotary since it is continuously turning the same direction.

[ZipperHead55]

Forgive me, but I can't help but feel that you created this post since you came up with the witty term "raster disaster".

You are offered fairly simple solutions (sharp cutters; a variety of materials of something other than pine, which is notorious for this very problem ("fuzzies")) but insist that your solution (which isn't very clear, since you are talking about "Start in the middle. Do a stroke on the left going up, then rapid to the right, and go down. This strategy does take more time than normal raster because of the rapids". To me, this sounds like offset (a small dot (or square I guess), that gets bigger and bigger until....... it isn't a square, and then becomes a regular offset toolpath, with all the inherent issues that come with it (the witness marks in the shape of the objects).

You want faster toolpaths that cut cleaner, but you offer up something that the software already does, albeit starting in the middle as a square.

Me, I would rather let the machine do the heavy lifting, using existing built-in strategies (you are the very first person I have ever seen comment on this, but maybe because you came up with the pithy catchphrase of "raster disaster", so like an annoying radio advert, it sticks in your head far longer than you wish), using the proper tools, and the proper materials. I COULD make my house out of pool noodles, but I chose to use 2X4s, 2X6s, etc. My car COULD be made out of Lego, but I chose to go with what's available (steel, aluminum, suitable plastics/polymers).

TL;DR: the existing toolpath strategies seem to work pretty well, and the OP wants to make changes that aren't really necessary, if they followed existing best practices (proper toolpaths, tools and materials).