When is a circle not a circle?

[Caesar S]

Wow, hit jackpot today I did a few techniques, the holes, triangles, and rectangles.

The holes were tricky, the rectangles were by far easier to do.

Also, flex is introduced by the end mill itself, the longer the shaft exposed outside the chuck, the more flex you have in the x/y direction (hence runout is worse).

So I exposed as little of the end mill as possible, and got good results.

[WillAdams]

We had a little bit on that on the wiki already, but I expanded it: http://www.shapeoko.com/wiki/index.php/ ... Guidelines

* Length
** Shorter endmills are more rigid, less-expensive and may provide a better finish (as a corollary, one should expose only the length of endmill beyond the cutting edge which is absolutely necessary[http://www.shapeoko.com/forum/viewtopic ... 551#p34551])
** Longer endmills, while not as rigid, afford the ability to cut more deeply, are more expensive and may not provide as nice a finish and may increase runout (see above note)

[samc99us]

I've been machining a lot of corian. Over 8 hours of machine time for just 2 parts. I go slow, 35 IPM on critical cuts, 0.05" depth of cut. This is with a modified SO2 (HFS5-4060 1000mm x-axis, fully supported 1500mm y-axis rails, stock x and y motor mount plates, z-axis NEMA23 upgrade, Dewalt DNP611, 1/4" 4-flute end mill, still only using Inventables belts). I notice eccentricity in holes that I plunge down into at high rates. Otherwise, not so much.

I tuned my machine by first squaring the router to the table using a set of 1-2-3 blocks (resting the router head on the blocks, and clamping the spindle mounts around the router), and also ensuring all the axis are square to each other. I then machine some 6" x 6" squares and measure them to compute the actual revolutions/inch values in LinuxCNC. Of critical note is the need to measure the actual bit diameter. The Inventables 2-flute 0.125" end mills are actually closer to 0.116" - 0.119" in diameter. That is a huge source of error if not accounted for!

Anyway, i'm within 0.005" after doing all this. The weakest areas I have right now are as follows:

1) No anti-backlash nut on the z-axis. This is responsible for a measured 0.005" error!

2) The 6mm belts. They stretch, but more importantly limit my acceleration values to 8.0 in/s^2. That's probably a large source of error for most. I'm upgrading to 15mm belts as soon as they ship.

3) The x-axis carriage. It twists about the x axis itself when I grab the spindle. The solution is more v-wheels, stiffer plates and even stiffer carriage fasteners. I'm also upgrading to a 4080 extrusion, which will help some more with the twisting and general machine stiffness. It means new y-axis sideplates though.

Best,
Sam

[Caesar S]

[Sorry for whatever got my previous post deleted, not sure why]

Thanks Sam, your post is very encouraging in that it is possible to get high accuracy at scale.

Now that I've figured out the squaring, tuning, leveling, tensioning, circles, x & y steps, feed rates, flexing, tightening, torque needs, etc. I've also mounted my Makita router back as my spindle.

I just cut a circle I'm happy with regular acrylic at 6mm depth. That's almost 1/4" thick very much sufficient for what I need. HDPE is easy at 3/4". Wood is just slightly slower than HDPE. I haven't cut aluminum or CF yet, that'll be another set of challenges I'm sure but then again I won't be cutting aluminum. Engraving is very simple.

There is an acrylic called Makrolon by Bayer which is really tough at 1/4", I have not been able to reliably cut that using my Proxxon Pro spindle. Maybe now with my Makita router I can.

I'm grinning like a kid in a toy store I'm ready to build my v2 SO2 at 1000 x 1000 mm span now.

I actually have the 1800 mm Makerslide so I can account for the extra wheels, plate widths, etc. Then use NEMA 23 motors, and the 9mm belt upgrade.

Would appreciate info on 15mm belt too (and pulleys, and idlers).

My end goal is to cut CF for production parts of my quadcopter aircraft designs. An aside is SO2 makes a great platform for a large scale 3D printer with space for multiple extruders (parallel printing of parts). Finally, at 1800 mm span, I can finally build a 100W laser cutter.

To save space, I'm thinking of placing all 3 subsystems on the same Y rails with separate controllers each and an A/B/C switch.

Whew, I've got at least 3 months of work to do. One thing at a time, I can work with my v1 SO2 now.

[samc99us]

Caesar S,

Glad you are up and cutting! If you decide to go for the 1000mm upgrade, let me know, I'll probably be selling my entire x-axis setup soon. Upgrading to bigger x-axis plates and likely a ballscrew z-axis.

LouisV has the 15mm belts. It's a pretty pricey upgrade for everything, ~$150-$200 depending on lengths and if you order extra belts.

[Caesar S]

I already have the 1000mm upgrade kit, with some additional some extra beams and rails.

I am worried about the Y motor mount not being stiff enough. I've been watching cutting in slow motion and there's definitely flex there while it's cutting.

Running the spindle at 30K rpm definitely helps push away material that would otherwise cause the flex combined with very slow feed rates (as slow as long as the acrylic doesn't start to melt and stick to the bit anyway, flyaway melted plastic chips are OK).

I do realize the limitation though, SO2 is an engraver that is CNC ready. Cutting is a nice bonus for me pushing the bleeding edge.

To address the Y axis flex I am considering stiffening up both Y rails using 2 rails each just as the X axis is constructed (so 2 rails, 2 motor mount plates, etc.) vs. lengthening the plate and mounting additional V wheels. I have enough parts on hand to do that.

For those looking this way, realize that the Y axis chassis plate mounts (at both ends of the Y rails) don't have the same width as the X axis, so a different length of spacers are required, I haven't measured them yet ... but attempted mounting the rails and found that I needed to spread apart the rails and needed longer spaces.

Will surface now and again ...

Cheers, Caesar.

[Caesar S]

OK, so I've been running in circles cutting different circles ha ha to isolate the flats.

No matter the software the flats are there. So that points to a HW problem (or controller).

I took apart the machine one by one and found that ta daaa the x belt was slack <dooooh>

I haven't tested it out yet, found it just now, time to sleep and test it tomorrow.

[cvoinescu]

Slack in the X belt would create flats, yes -- parallel to the Y axis.

[Caesar S]

Got to test it using raw g-code this morning. It is _beautiful_ .... love love love it

To be honest, it still doesn't cut round at small diameters (1" or less), but at 2" and larger I didn't notice anything out of the ordinary.

I've now tensioned both Y axis belts as well as X axis.

Edward, this really points to an improvement potentially to replace the belts with drive rods or instead of hand tightened tensioning, use a mechanism to tighten it say by turning a screw incrementally to the desired tension.

or this handy spring ... http://openbuildspartstore.com/belt-ten ... on-spring/