Addressing vector weakness

[deemon328]

I've been making a bunch of small parts for a project, and I'm a little disappointed in the output quality of vector cuts. I think a lot of us have experienced how hard it is to get a round hole or a round cutpath.

Here's a photo of one of my pieces. For some reason, my 1/8" bit depth was off. The outside pegs are 1/4" round, and the center is the 3/8" classical bit 3/8" deep in a 1.125" circle. I'm chalking it up to extreme corian particulate all over the machine.

The center piece is acceptable, but far from ideal. There's tooling marks all the way around the path (that you can't see in the photo clearly), and the path isn't round, alothough it's within the tolerance that I need. The center of the piece gets a 1/4" drill hole leaving 1/16" surrounding on the top, so if it's not pretty round, then the piece is no good.

http://photos.yoderwoodworks.com/images/A_2/4/0/5/15042/DSC04639_00171.Large.jpg

So, the discussion is how do we improve on vector paths for the CC heading into the future? Are there things that the software/firmware updates could do to increase accuracy? I'm all for lobbying LHR for new updates that could address these vector issues.

I read on an earlier post about the possibilty of slowing down the head speed to gain better results that's coming in a future update. I don't know if that's a real possibility or not since 1.126 just got released, but I'd gladly double or triple the time on my vector cuts to get better output.

How about ways to tweak the machine itself beyond the standard clean and calibrate? Are there things we can tighten up to reduce play?

[upcedar]

dustin, do you get this affect in wood also, thats a big-bite for any cnc. vector lines are used in cad/cam programs, it's the true line for the computer to pick-up to design a drawing. But, if you want software like the pro's pull out the plastic card - big buck's. maybe have someone redo u'r file, it might work better. Best Rick :)

[deemon328]

Thanks for the reply. Sorry for the error above; it's obviously a 3/8" deep path, since a 3.8" deep path would be impossible. I corrected the above post.

I'm attempting to address the vector carving function of the machine. If I draw a circle in Designer that's 1" diameter and do an outside cutpath, then the machine is supposed to follow that circle along the outside and cut a 1" circle within .005" tolerance(which I thought I read was the accuracy of the CW). That is what it is supposed to do, but the result is inconsistent, often producing a wavy oval. The part that I want to improve is the adherance to the path; I can live with runout errors because we can compensate for them in the software by making a piece larger or smaller.

Taking wood grain out of the equation seems like a good idea when testing for accuracy, since a solid surface material or acrylic would have a consistent resistance, but I could be totally wrong.

Anyhow, thanks for the continued discussion.

[upcedar]

now I know what your talking about after seeing your webpage, u'r a great woodworker! that's the kind of people carvewright needs to take the machine beyond the boundiers. I would look-up L/Q on this one. Rick

[Jeff_Birt]

You need to insure that your fixture holding the stock is not slipping on the traction belts or against the board tracking sensor.

[John]

Is it to much to ask for a "hobby" tool to produce professional results?? The QC (Quick Chuck, not Quality Control, although that point could be argued) on the CW is "hobby" grade. Look at what "automatic tool changers" cost on industrial cnc equipment and you will quickly realize the QC is a remarkable piece of ingenuity for its price. Unfortunately, with the low price, one has to have low expectations.

[deemon328]

Thank you for the further comments.

I'm as sure as I can make a poplar form with tape on the bottom for the aux roller for extra traction. Do you not have this problem, Jeff? I've certainly tried to do everything mentioned here to improve accuracy to this point. If there's more to the story, I'm all ears. Again, it's my goal to increase accuracy to it's potential.

No, I don't believe I'm asking too much of the tool. There seems to be a great paradox here on the forum. One school of thought is that it's an incredible piece of engineering and an amazing machine. The other school of thought is that it's a hobby piece of plastic that is supposed to be inaccurate. I happen to believe that the machine is built well enough to be able to expect good accurate work if properly tuned. Hobby to me only means that we must deal with the weaker parts to get the best results.

Regarding the QC (quickchuck), if the tolerance is such that it allows a little runout, does it also mean that the bit would wobble in the QC? If the bit is loose in the QC, then I can understand how chatter can occur, but would that even have visible results at 20,000 RPM and:


Movement Velocity:
Length Axis……………………………………………………..…2 inches per second
Width Axis………………………………………………….…….12 inches per second
Up/down Axis…………………………………………………....12 inches per second

What would be the difference in path quality if on vector cuts we were moving at 2 inches per second in both X and Y? When I watch the machine carve, I can see it go slowly around the X side of the curve then pick up speed as it traverses the Y axis. I think that this pickup in movement speed causes the bit to veer off path because of the load on the bit.

[DocWheeler]

Dustin,

If I'm following you correctly, wouldn't the bit-load be the same for the begining and ending (left/right) cuts and therefore cancel out that problem? OR, since the direction is reversed, will it double the error?

[deemon328]

That's a good point, and I'm just grasping at straws to find a reason why the path deviates. Here's what I'm noticing:

The outside cutpath is a .25" circle in .5" corian. I'm noticing on each depth pass that the bit moves slower along the X then picks up speed along the y. Now, the piece is always moving in two directions since it's a circle, but it definitely gets faster as it ends the majority of the X travel over the circle. It seems like when it speeds up it can't maintain the path properly, then I get wavy oval cutpaths. The inability to maintain the path is the root of the problem.

Another way to explain what I'm seeing: When you make a climb cut on the router table, you know how the board wants to shoot away from you because you're going with the bit rotation? That's exactly what these little speed up areas seem like they're doing. It's almost as if the bit is pulling the truck off path. The problem with this theory is that the bit is always contacting the surface the same way through the entire cutpath, so like you say, the force should be the same. So, is it possible that the Y/Z truck doesn't have enough stability vs. the belts and rollers of the X axis, and this difference causes the problem? I certainly don't know, but it furthers the discussion.

Shouldn't the speed around the path be the same for the entire carve? Why does it speed up and slow down at certain points? Can this issue be addressed via software/firmware by LHR to make things better for all of us?

Thanks again for the discussion :)

[DrBob]

Hi deemon328
I would say you are correct in what you are saying.
I am a machinist and the rule of feed rate is the softer the material the the faster the feed rate, the harder the material the slower the feed rate.
I want to carve some corian but am a little hesitant because of the feed rate.
I beleive that the machine was designed for wood with the grain flow along the length and that is why it speeds up on the x axes.
Bob

[deemon328]

Hi deemon328
I am a machinist and the rule of feed rate is the softer the material the the faster the feed rate, the harder the material the slower the feed rate.

Wouldn't that make a nice option for the CW...

Select Quality (draft, normal, best)
Select Material(soft, medium, hard)

Hmm, 6 options to have a finer control over cut speed and max depth per pass...


Regarding Corian, I've been cutting out a lot of small figurines in vector , routing paths in vector with the 3/8" classical bit and I've made some small 1/2" deep raster carves. Taking a full depth pass in raster does noticably slow the spindle down, but the results were totally perfect. The other two types of cutting I have done didn't pose any problems aside from needing to clean out the clingy dust. I've had to tape the surface because the speckled finish throws the board sensor into fits.

[twehr]

Shouldn't the speed around the path be the same for the entire carve? Why does it speed up and slow down at certain points?

When I first got my CompuCarve, I read someplace (can't find it now, of course) that there are different feed rates for the axes - two (Y,Z I think) are one speed and the other (x) is different. That being the case, the speed will appear to change depending upon which axis is being moved the most at any given time.

It is just a function of the geometry of the cut path.

To move from any point on a circle to the next point on the circle, there are two movements that need to be made, X and Y. The amount of X movement and Y movement varies with the position on the circle. Therefore, when there is proportionately more travel required in one direction, and the feed rate of that direction is slower than the other, the travel rate will be different.

Same thing happens on my mill when I have the traversal rates set different.

Just thought of this - A simple test or demonstration would be to cut a 6" straight vector vertically and a 6" straight vector horizontally. If the axes have different feed rates, the time required to cut one path will be longer than cutting the other. (Hopefully 6" is adequate to be able to time the difference.) Another possible test would be to have a long, narrow carving pattern. See if orienting it vertically and then horizontally will show a different estimated carve time. (Hopefully their estimate would take into consideration the different traversal times.) I think I will do both of these this weekend.

[twehr]

...when there is proportionately more travel required in one direction, and the feed rate of that direction is slower than the other, the travel rate will be different.

Just thought of this - A simple test or demonstration would be to cut a 6" straight vector vertically and a 6" straight vector horizontally. If the axes have different feed rates, the time required to cut one path will be longer than cutting the other. Another possible test would be to have a long, narrow carving pattern. See if orienting it vertically and then horizontally will show a different estimated carve time. (Hopefully their estimate would take into consideration the different traversal times.) I think I will do both of these this weekend.


I did some quick tests, and as expected, the predominately horizontal (X) cuts take longer than the predominately vertical (Y). Here are the results.

Carving Pattern: 11" x 14" - .600 depth - 0 Feather - No Draft - Best Quality

Vertical = 2:39:30
Horizontal = 2:55:51

Vector Path: 11.25" - .25 depth - 3/8" cutting bit

Vertical = 0:03:30
Horizontal = 0:04:08



Horizontal (X) movement of the workpiece is slower than vertical (y) movement of the cutter head. Therefore, board/head movement will appear to vary as it goes along a path of varying orientation (circles, curves, etc).

You can see pics of the test layout for reference below.

[DrBob]

Hi twehr
Thanks for that test, aside from the testing,
I love the Frame you tested with, any chance you could share the file, if you haven't carved it yet I am sure it will be awsome when done.
Bob