CNC converted G0602 is all over the place.

Need some ideas, I converted a G0602 to cnc and I can't get the thing to cut accurately in the X. I have tuned the motor steps with a indicator and get consistent movements both in + and -. When I go to do a step cut test (piece with 2 different diameters) I get inconsistent sizes in the dimension,. ie. a programed diameter is off .018 on the larger side of the cut and .0115 on the smaller side of the cut. In my thinking these differences should be consistent. This makes it very hard to fine tune the steps. I have made adjustments to the gibs, checked for slippage on the pulley gears and the bearings are not too tight. I do not believe there is any slippage in the ball nut mount on the X slide either. Any other ideas of things to check?

Thanks in advance
Ed

Backlash?

What are the target diameters? It could be proportional thus meaning that the steps/inch are not correct. Look at the error as a percentage of the diameter(s), and apply that percentage to the current steps/inch.

By ''proportional'' I mean that if it is off 0.020 at 2 inches and 0.010 at 1 inch, then the error is proportional.

Or is could just be ''spring'' in the whole system.

Backlash accounted for

Quote:

---

Originally Posted by Jim Dawson

What are the target diameters? It could be proportional thus meaning that the steps/inch are not correct. Look at the error as a percentage of the diameter(s), and apply that percentage to the current steps/inch.

By ''proportional'' I mean that if it is off 0.020 at 2 inches and 0.010 at 1 inch, then the error is proportional.

Or is could just be ''spring'' in the whole system.

---

I will check on that, thanks

what material are your test cuts on? What is the tool? When you say back lash accounted for do you mean in software or are you approaching the cut the same direction for each (i.e. like you would on a manual lathe)?
IMHO if you are troubleshooting take all the variables out that you can - i.e. don't rely on software backlash comp. Try with an easy cutting material (6061 aluminum), with a shallow depth of cut / low feed rate and a sharp tool.
Also, run an "air cut" program that is fairly lengthy (couple of minutes) then see if the tool can return to a known position to see if you are loosing any steps.

Test movement.

Will x axis movements repeat consistently, and return to zero//same place consistently.
No load, just moving the x.
Make a simple program to move the x axis some distance, 1.000", wait 10 secs, move back.

Use an indicator clamped to the bed, and measure the x axis saddle, top see movement.

Now, clamp the toolpost with something, a large carpenters clamp or an extra vice.
Tug, push, pull the toolpost while repeating previous test.
As hard as you can (this is less than 1/10 of what (interrupted cut) turning loads create).

Does the x axis still go to the -1.000" position, consistently.
And return to zero.
--
Repeat for z axis.
--

Imho, ime.
1. You probably have slop in your ballnut, and in your x axis bearings.
2. Your z-axis carriage is probably loose.
This allows the saddle assy to twist.
Like rotating left/right, if you looked at the saddle from up above.

2.
Indicate saddle assy, with di clamped to bed.

Twist the toolpost, hard, some way.
Clamping a 12x12 mm steel rod 400 mm long in the toolpost, and then pulling it, about 20 kgf, 50 lbs, in z axis direction +/-.
Or twist the saddle.
Clamp a 2x4 on the saddle and yank one end, while looking at the di.

The saddle should only rotate/move minimally, and should return to the same position when you release the load.
It is likely the z axis gibs are loose or worn, and allow the saddle to climb.

This is exactly what shows as "bouncy" unpredictable results.
As the saddle rotates, the tooltip moves making for larger and smaller cuts in inconsistent ways.

The under-ways z-axis mechanism might be loose, and thus allows the saddle to rise a tiny amount (0.001" / 0.02 mm) and rotate (quite a lot).
It was fiddly and hard to adjust the screws/nuts under the z axis, for me, similar 12x24" lathe.

Thanks everyone for the great suggestions. Test cuts were light and taken in 6061 AL. I found part of the issue was Fusion was kicking out the wrong dimensions in the G code, due to a tool setting, The last test cut I took was much closer to target dimension, but ran out of time and wont be able to do any more testing for a little while.

Again thank you all for the insights,. it is much appreciated.

Ed

I have finally had some free time to look into this more. I am getting closer but still having some differences. I ran 2 test sizes on 3 different pieces of stock and getting some variances in the CNC DRO and actual cut diameters. All test cuts were in 6061 @ 720 5 IPM .005 depth of cut. Any other suggestions always appreciated.

Here are the results.https://www.cnczone.com/forums/attac...d=431044&stc=1

Quote:

---

Originally Posted by Petesmods

I have finally had some free time to look into this more. I am getting closer but still having some differences. I ran 2 test sizes on 3 different pieces of stock and getting some variances in the CNC DRO and actual cut diameters. All test cuts were in 6061 @ 720 5 IPM .005 depth of cut. Any other suggestions always appreciated.

Here are the results.https://www.cnczone.com/forums/attac...d=431044&stc=1

---

That doesn't look too bad to me. A lathe like that isn't going to reliably hold tolerances finer than .001", and if you're off by .0005" you have to double the inaccuracy on a turned part. Spend another ten grand or so if you want to go down another decimal place. Inaccuracy is the sum of all the slop, backlash, deflection, temperature changes, and measurement ambiguities in your system; including slight roughness of the turned surface. Declare victory and move on!

That is actually pretty good. If it's repeatable then just adjust the X cutter comp to bring it into the target dimension. When setting up, I almost never hit my target dimension exactly, but a quick tool comp adjustment fixes it right up.