Centroid Community CNC Support Forum

Ok, Finally posting here my CNC Router. Its aprox 2' x 4' steel frame moving gantry. I built it with a pretty limited tools, welder, grinder, jig saw and drill press. I used self leveling epoxy to create flat surfaces for the linear rails. I am using stepperonline closed loop steppers Nema 34 CL86T. When I first commissioned the machine, I purchased a small linear stage for the Z axis. I had some wooden parts to get things working. Eventually I used the router to build parts for an alloy Z axis. I recently upgraded the ball screws from 1605 to 2010. The original screws were bent, and would whip at speeds over about 100ipm. Also they had about 0.002" backlash. I can get 500 ipm with the new ballscrew, they are very straight, and less than 0.0005" backlash.
The machine is a hobby for me. I design custom surf hydrofoils. I machine female molds out of MDF. I use the molds to laminate carbon fiber in a top and bottom half to make the hydrofoils. I run a parallel tool path with very fine 0.010 stepover. Some cycles run 4-6 hours. Never had any issues. I use precision mill as the smoothing setup. The results are superb, almost no sanding required, corners are sharp and crisp as they should be. I learned a lot and had fun building the machine. I am very happy with the Centroid Acorn board and CNC12 software. As long as you follow the instructions, everything just works.

I am always learning new things and trying to improve accuracy. I am trying to figure out how to improve the roundness of a bore tool path. Currently I get an oblong hole. With a climb cut the long part of the hole is at +45 deg to the +X axis with Conventional the long axis is at +135 deg. Hard to measure accurately MDF with calipers, but The longest part of the bore is about +.005" to +.010" for a 0.750" bore. The slower feeds produce more round holes. Speeds are 75 ipm to 35 ipm. I am running a spring pass. I was expecting a large improvement by replacing the ballscrews with considerably less backlash, and a much more rigid Z Axis. What am I missing?

right on! thanks for sharing, nice work on the machine.
foiling is really cool!

out of round holes could be the overall turns ratio not being correct.
looks like you have metric screws 10 mm pitch. i'd go back to 2.54 rev/in and do you test cuts again.
are servos 1:1 with the screws?

backlash and something loose can also cause a shift when cutting, so review that as well.

Question: What is causing an inconsistent radius after milling a circle or arc?
Answer: Mechanical lash in the machine or improper backlash compensation setting...
Solutions:
• Adjust software lash if less than 0.001". See TB037 - measuring machine lash
• Repair mechanical if lash is over 0.001".

Drawing below show exaggerated example of Mechanical lash in the Y axis with a CW cutter path and no lash in the X axis.


- Elliptical circles are the overall turns ratio set in correctly review TB36 Measuring Motor revolutions per inch.

- Problem: Steps at 0,90,180,270 Cure: reduce the Backlash Compensation amount. Set to zero and remeasure backlash.

look for loose pulleys , missing keyways, missing set screws, and other lost motion in screw support bearings.

when i mill 3d shapes on a router i set all the backlash compensation values to 0.0000 . The tool path i use does a turn around 'off the part' and therefore the backlash doesn't come into play. having it on in this case sometime causes issues when 3D surfacing like you are doing.

Thanks cnckeith for the feedback.

The closed loop steppers are coupled to the 2010 ball screws 1:1 with oldham style couplers. I will experiment some more with the backlash and turns ratio settings. I came up with those values using Marty's process using 2,4,6 setup blocks. I will go back to 2.54. Even so, the difference in turns ratio in X and Y should only account for about .001" difference for a 0.750" bore. Likewise backlash is pretty minimal.

https://drive.google.com/drive/folders/ ... drive_link

I put some vids on a google drive. You can see the amount of oblong fitting the pin in the bore. Looks just like your drawing. Am I correct that structural flex from high accelerations would cause the same out of round? What sort of accelerations are people running?

Are those high quality couplers or an Amazon special? I use Ruland couplers for all my couplings.

Your Google drive link doesn't open your album. It wants to "request access" to your album.

The couplers are made in China, as are most of the drive train parts. They are good quality 40mm diameter. They clamp tight and there is no play. Measured backlash is only 0.0005".

I changed the permissions to the google drive folder. Hopefully that works now.

https://drive.google.com/drive/folders/ ... sp=sharing

"good quality" and " China" dont belong in the same sentence.

Are they locked on the shafts with keyways or simply a set screw? Did you use Loctite on all the couplings where they meet the ball screw and motor shaft?

If your measured backlash is only .0005 then I would agree with Keith that your turn ratio is not as precise as it could be. To get it dialed in better you need to travel as far as possible, going 2-4 inches isn't anywhere as good as going a foot or more. If you have multiple gauge blocks lay them end to end on the table and travel as far as your precision blocks will allow. Do this as many times as you need to to get precise, repeatable measurements.

I have not build a router yet, but should the slaved axis not have the same settings like the main axis?

Good way to get rid of backlash with cheap C7 screws is to use 2 nuts, one floating and spring loaded.

Are the bearings in the fixed side at the servo angular and with proper preload?

2000 steps for a 10mm pitch is IMHO not ok, I would go to 4000 with steppers and 10000 with servos.

Uwe