Smoothing? <resolved, bad ballscrew support bearing>

[Ken Rychlik]

In wood cutting my program has the option of cutting climb or conventional. Is your outside cut clockwise or couter clockwise? If you can reverse it, that may be worth a try.

[JayCNC]

Post a video of it cutting, maybe someone will have some better suggestions after seeing the machine cutting. Just a suggestion since like Marty always says, we can't see what you see.

Maybe it will be something obvious to some of the experienced machinests that are on the forum? Also have you looked at the recommended feed and speed and chip load and sfm for your endmills? I know my endmills have a manufacturer calculator and it's way different than the g-wizzard calculator. Like not even close once I looked up the manufacturer specs. I don't use it anymore, I just use the formula based off my tooling recommend specs.

Have you tried using a larger endmill with a clean up pass just to see what happens?

I'll get some video tomorrow evening of it. Yes, I have looked up manufacturer specs, and if you slide the finish slide around in g-wizard it's close I've found, but way slower feed than they typically recommend.

I've used 3/16, 1/4, 5/16, and 3/8 and each has it's own pattern to it. The one pictured was with 1/4".

Tomorrow evening I'm going run a series of test cuts, starting at 10 degree angle, 20, 30 and 45, all with a .01 finish stepover and .5 doc and all with a climb and conventional finish pass and see what happens. Also going to get the indicator out first and check if I have any side to side movement at the bottom of the Z plate, I actually don't think I ever checked that way.

I forgot to mention earlier that I have checked spindle runout, and to my surprise it was less than .0005 (really need to get a tenth indicator).

[Dan M]

What software are you using for the cam, what grade aluminum, spindle speed for each tool, doc per pass? Also forgot to ask are you using carbide endmills?

Dan

[tblough]

The CNC code is not the problem. The CAM software is outputting a single G1 move for the straight lines and and a single G2/3 for each arc. It is not generating a bunch of short line segments. Smoothing should not be making any difference on this file. I suspect with smoothing turned off, the problems will still exist.

Run a 0.002" finish pass and see if that makes a difference. This is definitely a rigidity/backlash problem. My guess is if you measured the waviness in the problem area it's only one or two thousandths deviation. It doesn't take much of an imperfection for our eyes to see it.

[slodat]

Have you looked at the following error while cutting the diagonal? I found on my machine I had more following error in some cases with smoothing on. In the end I found that my servos weren’t controlling the axis well enough and I had the exact same quarter decay oscillation you are seeing. This was thought to be caused by an inertia mismatch of about 9. Everyone told me my machine wasn’t rigid. That wasn’t the issue.

[cnckeith]

Have you looked at the following error while cutting the diagonal? I found on my machine I had more following error in some cases with smoothing on. In the end I found that my servos weren’t controlling the axis well enough and I had the exact same quarter decay oscillation you are seeing. This was thought to be caused by an inertia mismatch of about 9. Everyone told me my machine wasn’t rigid. That wasn’t the issue.

ok..i'll add to the list .. "mismatched (not enough power/torque and or "geared" incorrectly so the axis motor is not operating in its "happy" zone and or incorrectly tuned servos" to my post. thanks!

[JayCNC]

Have you looked at the following error while cutting the diagonal? I found on my machine I had more following error in some cases with smoothing on. In the end I found that my servos weren’t controlling the axis well enough and I had the exact same quarter decay oscillation you are seeing. This was thought to be caused by an inertia mismatch of about 9. Everyone told me my machine wasn’t rigid. That wasn’t the issue.

Could you explain to me what you mean by “following error”?

[JayCNC]

Btw, Servos are direct drive with Ruland Spyder Couplings

[slodat]

Following error/position error = current position - commanded position.

The Clearpath software’s oscilloscope can display this.

This is the reference I was sent when I asked Teknic about looking at following error. He called it “tracking” error. Synonyms.

https://youtu.be/oJSp_1aNiSE

[JayCNC]

Ok, so I re-checked EVERYTHING on this machine and found that there was some play in the bearing block for the ballscrews. I took them apart and ended up shimming between the two bearings outer edges until they felt tight. I re-ran the same program with much better results but still not perfect. I’ve come to the conclusion that new better quality angular bearings are needed as I’m 99.9% positive this is the problem. Anyone know where to pick up a good quality angular bearings from? 12mm ID X 28mm OD X 8mm. I checked McMaster Carr but there’s no way I’m spending $204.32 EACH!!