I am too, but only because I am not seeing (old tired eyes?) the “error” in the Messi image. I am thinking we have drilled down to the point where machine repeatability and positional resolution are coming into play. Remember most of these are not $250K CNC machines with rigid cast iron frames. If you zoom in far enuff, you will see wiggles and deflections you cannot remove or compensate for.
It depends on the type of layer (line or fill) line is almost every single time the one that engraves with accurate measures, the fill layer is the one that gets shrinkage on one axis. Here i did 6 50x50mm squares in the same job.
- Line
- Fill 0,2mm interval
- Fill 0,1mm interval (default)
- Fill 0,07mm interval
- Fill 0,05mm interval
- Fill 0,03mm interval
I attach the file for more information
Test.lbrn2 (5.8 KB)
Results.
At first i measured the X axis since all the latest missalignents were ocurring in that axis, but now all of a sudden ALL the squares measured 50mm in the X axis. Those that presented misalignment presented it in the Y axis.
Intersting to notice how the 0,07 interval (4th square) shrunk about 1mm in the Y axis while the 0,03 interval (6th square) actually stretched about 2mm in the Y axis.
some times it shrunks, some times it stretches. i want to throw this thing out the window.
Have you changed any settings at all during this time?
The fact that the artifact has shifted from X to Y makes me think this is related to the “PWM edge” issue.
Can you invert the Y-axis step pin ($2) and retest using the same test pattern. If that doesn’t work, can you test with every variation of X, Y step pin configuration. I’m assuming that Z is not inverted currently.
$2=0
$2=1
$2=2
$2=3
i had done this before with the messi test
The results didn’t change respect to the messi tests i was doing.
That said.
I have done the 6 square test with the $2 true on the X axis.
the top right area of the 0,2 interval square didn’t burn because of a fold in the paper, i tried my best to correct it but it seems it wasn’t enough.
Other observation to highlight, the roughness in the bottom side of the 0,07 square.
Last and most important observation, the shrunk and stretch for the different squares is exactly the same as it was before. Just like with the Messi test, i really don’t think switching any of the $2 for the axis will change the results. It didn’t change them before and so far we haven’t change anything in the machine for us to expect a different result.
I’ll be doing another square test, this time with only the $2 for Y axis set true. I’ll update with the results only if they do change something.
I’ve shown how to do this in my previous post. The Console commands go from X off/Y off, X on/Y off, X off/Y on, and X on/Y on respectively.
The Messi test makes it difficult to isolate specific issues since the image is very complicated. A controlled test should make it more visible.
Perhaps not, but you indicated a swap in behavior where the artifact changed from Y to X.
It’s hard for me to correlate the pictures to the test you’re running. Can you label them? Also, are the pictures in the orientation of X as width and Y as height? Are you only measuring vertical distance?
im only posting the results of the vertical distance since the horizontal distance is precise for all 6 squares. i measured it every test to make sure that’s still the case and they are all exactly 5cm.
The change from Y to X only happened in the Messi test when i changed the interval from 0,07 to 0,1. If i do another Messi test at 0,07 it’s just like before. So far the interval size seems to be the only thing that affects the shrunk or stretch of the fill layers, as well as the axis in which it happens. however in the squares test changing the interval size doesn’t have the same effect, since despite going from 0,2 to 0,03 there hasn’t been any axis change like in other cases.
I don’t think the Messi test is useful for anything but to see if the problem is solved. What I meant was that considering that the $2 test gave the same results with the Messi engraving (results being that no matter the $2 value, the shrunk/stretch was not affected) I am very sure that it will give the same results with the 6 square engraving as well (considering there has been no chages between $2=0 and $2=1). I am currently doing the engravings to confirm this, but i would not wait for the results to consider that the problem is related to something else.
What I think @berainlb was trying to say (politely) and I am saying for sure is that only you know what is wrong with the Messi test. If we see nothing wrong in the square and circle tests, it is difficult to see what you see in the Messi test. Putting a circle or arrow in the image identifying the problem would really help us help you.
the $2=2 test is done now. The only differences i see are that the roughness of the bottom sides of the 0,05 0,07 and 0,1 squares is worse, although my laser height adjustment for this test was kinda terrible, so it may be because of that. The other difference was a slight shrunk in the 0,2 interval square, but i think this is due to the laser height adjustment i mentioned before.
in the left is the $2=1 test and in the right is the $2=2 test. The bottom of the square being engrave different was because of the laser height adjustment being way off, so i pause the engraving, did a little adjustment and continued.
All of the other measures were identical to the other 2 tests ($2=0 and 1) same shrunk in the same squares, same stretch in same layers.
Now doing $2=3
results of the $2=3 square test. Identical to the others. precise on the horizontal axis and in the line layer. no visible offset in 0,2 interval, 0,1 interval and 0,05 interval. 1mm shrunk in 0,07 interval, 2mm stretch in 0,03 interval.
I am taking a sideline on this. Neither am I keeping up with what is wrong in the images nor am I following the intent of the complaint.
Is the left side of this wavy? Or is that an illusion? If so, X-off is better.
Is this slightly short? If so, Y-off would be better.
I’m having a hard time following and tracking which images correlate to what, but have you been testing with $2=0? Is that the default state? Sounds like that should be ideal.
it may be because of the paper, i can never get it perfectly flat for the pictures. When engraving i tape it down so its as flat as possible.
yes, the $2=2 was slightly short, abut half a millimeter, maybe less.
Yes, actually that was the configuration for the very first test of the 6 squares that i did here
In the follow up test with the same file, i followed the order of that first post to show the results. also every picture has a legend of what interval size it belongs to.
At this point I’m not sure what to think.
If this was only about variation in Y and was not able to flip to X I would say this might be an issue with the belt or gear or something in the driver since the interval size affects this. But that wouldn’t account for the artifact showing in X axis at certain intervals.
its not just at certain intervals, the shift in axis was present in the Messi test, but in simpler engraves like the 6 squares one, I’m not able to shift the axis of the stretch/shrunk. Maybe i should try something in between? more complex than the square but still with references to measure distance. but i don’t know how the conclusions of that would help us. If we observe more or less change in layer size or shift of axis from one interval to another, still there is no idea of something in the machine that could cause this problem following this behavior.
Just thinking out loud here, but perhaps a circle or hexagon would provide more data and still be easy to measure? Possible a “T” shape with known leg and cross sizes…
I can clearly see the problem with the Messi jobs you’ve shown and find it interesting that his fingertip on the left of the image is generally on-point and the error seems to get worse as you look further from that point.
Results of the first circle test. Behaves just like the square’s test. same interval presents same size changes in same axis. performing now the hexagon test
That has always been the case with this problem. I used to think that the whole thing would get off as it went through the engraving and by the moment it had to do another layer, the coordinates were off.
That wasn’t the case.
Doing the tests for this post i discovered that the problem is a change in the size (can be a stretch or a shrunk) in one axis of the fill layers. So far we can say that a same design with the same interval size will present the same change in size. I confirm this comparing the circle test with the square one, the changes measure the same in the same interval layers.
So the size itself can change with different interval sizes, BUT depending the design that is being engraved, this changes in interval can lead to a shift of the axis (this happened in the Messi test) in which the size change takes place. But we haven’t find out what pattern rules this behavior… yet.
After the hexagon test, which i expect to go as the square and the cicle one, i’ll be doing a Messi test with many tiny messi designs, each with diferent interval, to see if a shift in diferent axis is possible in one job. what is this information useful for? i dont know, i don’t know much about laser engravers, but you guys do.
At this point i’ll do any tests that come to mind and any tests that you guys tell me to (that i haven’t done before with the same config) in hopes to get you enough information about the behavior of this problem to come up with ideas about where the problem could be.
Shot in the dark here…
I was making some minor adjustments to my scanning offsets and noticed the “initial offset” field that I never needed to use. It seems to fit the bill here, other than the odd X/Y swap. Maybe worth a shot…
Continuing the discussion from Scanning offset adjustment fields:
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.