Circle trouble - beginning and end are offset

LightBurn Hardware Compatibility
1

Hello,

I read lots of posts regarding Circles not round and do not meet but couldn’t see the same issue I experience.

I am new to laser cutting and just assembled a Sculpfun S30 Pro (10W) with Y axis extension. When engraving / cutting the first test pattern, the cut circles are oval and there are strange lines off the circles (and rectangles too).

Sample-2
Sample-21200×1639 344 KB

I run some additional tests just on the cutting after adjusting the Y axis belt, but it made no difference.

Sample-3
Sample-31200×1732 399 KB

I then used LaserGRBL to make the same cuts. This eliminated the offsets, but the X axis is still too short.

Any ideas how I can fix the offsets in Lightburn and how I get a round circle?
Regards, Peter

2

Possibly looks like your wheels are overtensionsed. Check for binding by manually moving laser head in both X and Y directions. Do you feel excess resistance? Adjust eccentric nuts on the wheel axes to relieve tension in that case.

You may also be dealing with a loose pinion gear likely on the Y axis. Make sure the grub screws are well secured against the flat of the stepper shaft.

3

Thank you for your reply. When doing the first sample, the X and Y axis had been overtensionsed. I then came across a video showing how the tension is correct by having the laser head move on its own by tilting the laser cutter around 30 degrees in both axes. Tilt Test on middle of page

I also adjusted the eccentric nuts on the wheel axes in oder to make the print head move on its own when tilting the laser cutter.

In addition, I checked all the grub screws, and they are all well secured. Unfortunately, none of this made a difference. Regards, Peter

4

The issue is almost certainly mechanical. So look for anything along the mechanical chain where backlash could be introduced.

Start at the motors and check for backlash at every step along the chain. The pinion gears are still the most likely area for something like this. Make sure that the grub screws aren’t double-stacked, and that a screw is directly against the flat of the shaft. There should be no back-and forth motion of the shaft without direct movement in the pinion. Then proceed along the rest of the chain ensuring no backlash at each point.

5

I went through the whole laser setup again and corrected some slight alignment issues I came across. I didn’t encounter any loose screws and can’t feel or see any “visual” backlash.

The test print has improved, but there are still some slight steps most likely because of some remaining backlash in the system. The three samples are with different speed and reducing the belt tightness a little. But none of these made a diffrence.

Sample-4
Sample-41200×1464 311 KB

I think this remaining backlash are from the Y axis belts which in my case are 950mm long and I can’t see a way to eliminate it.

Is there a way to compensate for this in LightBurn? Also, the laser head moves very fast when moving between cuts. Is it possible to reduce this speed which may help a little as well? Regards, Peter

6

Another source of backlash can be the laser module mount on the gantry: it’s a large off-center weight secured by a couple of fiddly screws. Make sure the whole laser block has no play, particularly in the Y axis direction.

Reducing the Y axis acceleration can help: reduce GRBL parameter $121 by a factor of four.

After backing that off, you can reduce the Y axis speed by reducing GRBL parameter $111 by, say, a factor of two.

If dramatically lowering the acceleration and speed reduce the backlash, then it’s definitely a mechanical problem. If you can’t find the source, reducing the performance will work around it.

Software compensation can “hide” a mechanical problem by pushing it elsewhere in the pattern where it’s not so conspicuous, but if you depend on results that match exactly what the patterns call for, hiding the problem won’t get you to a Happy Place.

7

Also worth considering is the resolution on the stepper drivers. Resolution meaning how fine of a distance can the beam be directed to, measured by steps per revolution on the driver. Make sure the driver is set as small as possible or consider that the current drivers need replacement.

8

Definitely better. Surprised you’re getting improvements but unable to detect the backlash during inspection.

The amount of backlash you’re getting isn’t exactly small so should be apparent. Have you tried removing the belts entirely or getting them not engaged with the pinions? That should help you isolate individual components better. Also check the lens itself to make sure it’s not wobbling around in the module. I don’t expect that to be the case here but I could something similar happening if the threads on the lens or module were not properly cut.

@ednisley points out how to change the acceleration and speed. Traversal moves in LightBurn will go at rapid speed. So basically the max speed as defined in $110-$112.

You can enable “Hide backlash” in Optimization Settings in Laser window. If that does eliminate it then again reinforces that it’s a mechanical issue. If so, you really should try every possibility to track down the mechanical issue as you’ll have trickle down effects you don’t want to have to deal with.

9

Thank you all for the replies.

ednisley, The laser module had indeed some slight movement, but in the X axis which is now fixed.

I reduced the acceleration from 1000 to 200 and the max speed from 6000 to 3000. But no difference in the test pattern after these changes.

Dean, sorry, I don’t understand your comment regarding resolution.

berainlb, I did remove the belt when going over the setup and couldn’t detect any backlash. The lens is fine as well.

The main purpose at the moment is to cut some templets to help laying flextrack on my model railway. I then used the test pattern on all four corners and each one had the same backlash as previously shown.

I then used the laser for some larger template like pattern plus two 100mm circles and a 200x350mm square with a speed of 400.

Sample-5
Sample-51200×702 136 KB

Sample-6
Sample-61200×613 104 KB

There is basically no backlash visible. The pencil lines mark the beginning and end of the cut.

Sample-7
Sample-71200×758 231 KB

But there is an issue with the Y axes dimension I don’t understand.

Sample-8
Sample-81200×459 69.2 KB

It is too short by 0.5mm. But it can’t be a calibration issue as the error is not linear.

with a 10mm rectangle, the Y axes is 9.5mm. On a 100mm circle, the Y axis is 99.5mm and on a 350mm long rectangle, the Y axes is 349.5mm - always 0.5mm too short! The X axis is always perfect.

Does someone have an idea why this happens and how to correct it?
Wishing you all a Happy New Year, Peter

10

If I’m following the sequence of events correctly, you have now eliminated all the mechanical backlash: congratulations!

Fussy, fussy! :grin:

That’s small enough to come from the non-square shape of the diode laser beam: it’s wider along one axis than the other. Not by much, mind you, but enough to make a difference for fussy folks such as you and I.

The path around the circles and the rectangles looks darker on the left / right sides and lighter on the top / bottom sides, because the beam is wider along the X axis than the Y axis.

Soooo, you are now looking directly at the physical limits of your machine.

If you measure center-to-center, rather than edge-to-edge, across the patterns, the distances should be pretty close to equal. Next, you’ll get into very fussy steps-per-millimeter calibrations:

I feel your pain …

11

Thank you for your replies

ednisley
If I’m following the sequence of events correctly, you have now eliminated all the mechanical backlash: congratulations!

Unfortunately, this is only the case on larger objects. On smaller objects there is still some noticeable backlash!

Sample-10
Sample-101200×478 159 KB

ednisley
That’s small enough to come from the non-square shape of the diode laser beam: it’s wider along one axis than the other. Not by much, mind you, but enough to make a difference for fussy folks such as you and I. The path around the circles and the rectangles looks darker on the left / right sides and lighter on the top / bottom sides, because the beam is wider along the X axis than the Y axis.

That is correct, but the beam size is only 0.06 x 0.08mm in size, so I can’t see that an extra 0.02mm create a 0.5mm difference.

The Axis scale check is a linear issue = the longer the distance, the more the difference. But in my case every dimension in the Y axis is 0.5mm shorter - a 5mm rectangle is 4.5mm in the Y axis and a 500mm rectangle is 499.5mm! Indeed, no big deal on large items, but on small items it is a different story.

Sample-14
Sample-141200×581 140 KB

X axis on the left, Y axis on the right

When I increase the belt and wheel tension, the backlash in the ovals disappeared, but it is more pronounced in the circles instead. But the Y axis dimension is still 0.5mm shorter

Sample-11
Sample-111200×1647 343 KB

The same is the case if I make them looser. The rectangles and circles on the righthand side are 0.5mm longer in the Y dimension to compensate the error and it does work.

I could improve things by playing around with the belt and wheel tensions and using hide backlash option in LightBurn.

Sample-13
Sample-131200×1658 369 KB

Unfortunatly this doesn’t correct the 0.5mm error. But I did come across some forum posts of people having similar issues when using the Y axis extension like I do. So I guess you are right in saying that it is the physical limits of the machine! And the fix is to add 0.5mm to the Y axis dimension in the design. Regards, Peter

12

And the game remains afoot!

After a DIY extension of my 3018 CNC router, I discovered the longer leadscrew (which is not what you have) had an incorrect lead, off by one part in 250:

Which caused a linear error with distance, rather than the constant error you’re seeing, but it came from a mechanical error in something I would have laid big money on being impossible to manufacture so far out of tolerance. Of course, buying stuff on Amazon & eBay opens the door to many impossible outcomes …

Anyhow, the questions to ask your extended hardware now become:

  • Which absolutely fixed & immovable mechanical part (still) has a teeny bit of slack?
  • Might the gantry tilt slightly more on those squishy wheels during travel this way than that way?
  • Are all the belts ends as motionless under their screws as they should be?

Thinking about such things definitely keeps me up at night … :grin:

13

Try clamping / screwing the machine to the worktop. Looks like it is very light and may move slightly while in operation.

14

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.