[Cocowboy]

I have routed edges of the subject wood both at recommended feeds and speeds, much slower and in between. I have a large WorkBee (Aluminum extrusion frame with belt drive Y axis). In all cases of a curved edge, I end up with an edge requiring significant smoothing with a spokeshave, card scraper, and/or sanding. The endmill seems to be deflected by the near vertical grain of the quarter sawn oak. I suspect that I am up against the rigidity of the frame. The smoothness of edge seems unaffected by feeds and speeds.

Anybody else encountered this problem? Solutions? Recommended experiments?

 

[difalkner]

Welcome to the forum! I have some quartersawn White Oak but haven't cut any on the CNC. What bit are you using (besides just 'endmill')? What feeds/speeds are you running? Are you cutting conventional or climb cut? What depth of cut? Got any photos you can share?

Are you able in your design software to make a conventional cut final pass and remove 0.001" to 0.002"?

 

[Cocowboy]

For the CNC I use metric units. Feed = 1250 mm/min at 12 Krpm speed. I have tried both conventional and climb. Conventional works better. I have used 1/8" and 1/4” square endmills with depth of cut ranging from bit diameter down to 0.5mm. When using a lower feed, I adjust speed to minimize sound. For example, 200 mm/min uses roughly 7 Krpm. In MDF, it is possible to use 0.001" depth of cut and feel the difference. Oh and the endmills are 2 flute spiral upcut. I doubt XY actual accuracy is much better than 0.007". Ball mills and rounded corner mills can "skate" when trying very fine vertical adjustments.

 

[Cocowboy]

Thanks for the welcome.

 

[Cocowboy]

As for the software, I have been using Fusion360, but for the edge work, I have manually programmed it, while using LinuxCNC.

 

[Cocowboy]

A recent example from an offcut.

 

[difalkner]

Yeah, that looks like either chatter or flex.

 

[old55]

Welcome to the forum.

 

[Cocowboy]

Flex it is. No chatter

Welcome to the forum.

Thank you!

 

[jw2170]

Welcome to the forum.

 

[maddog]

If you can use 1/2" shank bits I'd try that route. Otherwise I watch your X/Z gantrys for flex.

 

[Cocowboy]

Actually, I have done a little better than watch. I have avibration meter app on my my smartphone. There is a frequency peak at the frequency of which grain lines are encountered. What it doesn't show, of course, is the location of the displacement, which must be some combination of frame movement and bit bend.

A half inch endmill would be a good direction to try. It would be like the difference of a bike rolling over a big rock versus that of a rear tractor tire plus resistance to bend. Thanks!

 

[maddog]

If your Workbee has the rubber wheels I'd suspect there's a little give in those. Hard to detect, but as you've said very noticeable. If you get the same results with the 1/2" shank then the problem is definitely in your gantry. I've had customers experience the same flex results on the Next Wave Shark with the unsupported rails on the X Axis. I've never experienced this on the machines I design and build as I'm using 80/20 aluminum and 20mm HiWin rails, which do not flex.

 

[TimPa]

... versus that of a rear tractor tire plus resistance to bend.

now you got my attention!!!

you mentioned that it only occurs on curved edges (arcs?). do you get a clean cut on a straight line of white oak then?
if i have it right, i wonder if your program is leaving the arcs as small vectors that need smoothing out! just guessing - keep us updated on your progress.

 

[maddog]

TimPa also has a good idea. If your design program has a Node editing mode you might want to select the curve and see if it's made up of hundreds of nodes or if there are only 3 points (Bezier Curve). Some programs interpet a curve as a multitude of steps. This is sort of like how Photoshop makes a curve by changing the opacity of pixels, which is why bitmap images don't trace very well.

 

[Cocowboy]

When cutting circles, the "program" (meaning my g-code) is G2 or G3. When the arc goes essentially parallel with grain, then the routed surface is smooth there.

In the case of the dining chairs, the backs are hyperbolic (circle and parabolic curves just didnt look right to my eye). I used Fusion360 CAM for them. I drew them with Opens as, and imported the .stl into F360. It used G2, G3 and G1 for the edge contour. Towards the bottom and top ends the arcs have huge radius. Once a again smooth in-between grain lines, and not where distance between grain lines was smaller. The MDF models were smooth; No visible singularities that you might get with straight line segment curve fitting.

 

[Cocowboy]

I am going to conduct a test. A straight line cross grain cut conducted in X and then Y with 1/8" and 1/4” square endmills. 1mm depth of cut. 200mm/min feed rate to get close to max motor torque. For first bit of first cut tune rpm to near lowest noise with spindle pot.

Cheap to fail experiment. Make sense? Pic results to follow.

 

[Cocowboy]

Depending on results a trochoidal tool path to follow. Probably at higher feed rate to approximate straight line feed rate above.

 

[Cocowboy]

I have completed 1/8” bit cross cut parallel to Y axis and 1/4" bit parallel to Y and X axes. I also used my power miter saw to cross cut for comparison. The 1/8” parallel to Y axis and the 1/4" parallel to X axis are about the same to my eye. The 1/4" parallel to the Y axis (90⁰ to gantry) was the best, but did not equal the power miter cross cut.

I suspect the displacement evident in the parallel to X axis cut is due to the belt drive. The X axis is Acme screw driven.

Even though grain was not parallel to the bits, they were displaced at each grain line. I suspect that the 1/4" bit "bridged" between grain lines, which improved the "smoothness". There are roughly 10grains per inch in the rest board.

 

[Cocowboy]

A year ago or so, a local avionics installation company asked if I could make an aluminum pad for mounting an instrument nearly vertically in the outside of a plane. I went and looked at it and made some measurements. I came home with a piece of aluminum, and proceeded to make the pad over the next few days. The bottom was curved, so I tried to use a ball endmill for a finishing pass. The ball endmill just skated over the existing surface. So I used a square endmill and a really small step over.

There are two potential causes. One is inadequate downward force to keep the bit cutting. The other is that the bit is not sharp enough to slice into the material. In the case of planing endgrain on a shooting board, a plane blade that is not really sharp will skate over, too, especially with white oak. One strategy to alleviate this plane blade sharpness issue is to skew the blade, which is the idea behind spiral fluted router bits. Another result of inadequate blade sharpness is tear-out.

Mechanism of the Cause of the Observed Endgrain Cross Cut Surface

Based on the above, there are two possible mechanisms that cause the observed surface:

• the bit is forced out of its path by the harder grain lines, and is allowed to do so by flex of the CNC router framework.
• the bit {prabably the flute edges) is inadequately sharp, and tears-out the softer material between the grain lines.

If the first mechanism is in play, then why are the valleys between the grain lines uniformly "excavated' when the grain lines not parallel to the bit axis would not allow the bit to sink into the material? It seems that the second mechanism is the more likely one in play. One observation supports it.

When milling MDF at proper feeds and speeds, the machine only produces chips. No fine dust. When milling white oak, there is a significant amount of fine dust along with the chips. Thus, the bit (a new one) tears-out the material between the grain lines.

Confirmation of this could be confirmed with an adequately sharp bit, and, more specifically, adequately sharp flute edges. Another possibility would be to try a non-fluted bit.