Accumulated location errors in finished part

Sorry for the cross post, but I realized that some of my questions in the machine shop topic are better asked here in the Laser topic.

The part I’m working on is a template for manually drilling holes in some thin plastic protective sheets which go inside a metal enclosure with various PCBs and cables. I’ve cut several versions of the part with some success, however I’ve noticed a scaling or accumulation error which is affecting the placement of holes as well as the overall size of my template when cut by the laser cutter.

Here is a description of my workflow:
The original part drawing is in ACAD (dwg). I convert this to DXF using a discontinued Autodesk program called Autosketch. Then I import the DXF into Adobe Illustrator and save it as a .ai. Lastly, I import it into RDWorks and cut with the laser.

The center of the template seems to be perfectly aligned with the metal part, however, all holes in both X and Y directions from the center get farther and farther out of alignment toward the edge of the part.

The part is roughly 48" x 20". The accumulated error is aprox .125" in the X direction and .050" in the Y direction at each end.

Currently, I can verify that my tools seem to verify that the conversion from DWG to DXF doesn’t seem to introduce any error. I’ll have to get back to DMS and check further down the chain.

It may be that the software is all correct and that the laser cutter tool itself is introducing the error. I don’t know yet. I’ve only tried this on Big Laser since the part is so long.

Does anyone have any experience or suggestions?


Planning on using hardened drill bushings? You might not like the kind of hole a HSS twist drill makes in thin material. If the size remains constant between parts, I’d say you’re better off making a die set and just hit it with a hammer to blank the hole out. Just an observation, neat project!

Why are you converting the dxf to an ai? Every stage you ar adding error since dwg/dxf are unitless files and conversions have to take place based on the transferring program’s settings.

Has something changed that would prevent thunders from handling the dxf natively? It’s been a little while since I ran a non-artistic job, but typically I kept everything as DXF for my engineering project cuts

1 Like

Few things,

-You can save a dxf straight from AutoCAD
-You can import a dxf into RDWorks or Lightburn (lightburn is sometimes better)
-2.5 thou over an inch is not as great as I would hope for but also believeable for our lasers. We should get better by tightening up belts, but I don’t expect CNC accuracy out of these.

I’m more inclined to blame the machine than your files, but 4’ is a big part to hold tight tolerances on. How accurate do you need the holes?

What they said.
Try AutoCad -> DXF -> Lightburn

@Cwatkins Thanks for the suggestions! I am new to this kind of fixturing, so there is a lot I don’t know about, including the drill bushings you suggested. I looked these up and they seem like a great improvement and something that I am glad to know about.

I would also really like to understand how to make a die, but that seems like a significant undertaking. From time to time I have a need to quickly build a fixture for this kind of work because there is some unexpected lead time problem from the company which die cuts these for us. The ideal process would be one that I could complete and use in Production in a mater of hours once the need is detected.

Are there / have there been any classes at DMS on die making?

@hon1nbo I have tried to import DXF into RDWorks multiple times, but the scale is always wrong. If I remember correctly, the DXF is rendered as a ridiculously tiny version of the part.

There probably is a way to make it work correctly, but I do not know how. If anyone has some ideas, I would love to hear them.

@michaelb & @Kevin I will give Lightburn a look. I don’t have AutoCAD myself (I am not the mechanical engineer, just the electronics engineer), so I use AutoSketch (from Autodesk) to do the conversion to DXF. I assume that it knows what it is doing and have produced very accurate PCB assembly fixtures using this method (DWG -> DXF -> AI -> RDX) with no problems, but those fixtures were significantly smaller (~12 - 16 inch max).

I’m going to go back to DMS and try to check the measurements on the .ai and .rdx files to see if something shows up in the software before I start blaming the laser :wink:

Multiply by 25.4; you likely have your CAD and RDWorks not using the same standard (metric/imperial). DXF is a unit-less file so whatever imports the file will assume it uses the same units.

I had an issue once where Inkscape exported to font points as the units.

What precise flavour of dxf/options does this require? All varieties I’ve seen will either silently fail to import or arrive at a bizarre scale.

Wow! What a terrible idea for a file format!

I’ll try the unit conversion and and see if that is what is happening. Does RDWorks have a scale adjustment? I assume it does, somewhere.

We have AutoCAD in Digital Media

On the import you should be able to set a scale if I recall correctly.

Correct. Under Config(S) --> File Para Setting. Towards the top is a field that says DXF Unit with a pulldown to the right of it.

Sweet! I’ll check it out. This certainly would save me some steps and might just fix the accuracy problems.

After further review, all files measure OK in software. I have to conclude that it is the laser cutter which is not holding the accuracy over the full part somehow. More tests are needed to prove it (like burning inch markers over a large sheet or some such).

As a side note, I was able to import the DXF directly into RDWorks and measurements check out OK. Looks like the key is the DXF unit setting.

Thanks for the help, All. I’ll try to run some tests and post results here as they become available.

Aside from getting better accuracy out of the laser, what would anyone recommend for generating a large drill template such as I’m trying to do? Does DMS have other precise drilling tools? I’m even considering large paper prints and just drilling plywood with a drill guide.

I would suggest the following tests
First file: two holes, .250" diameter 12" apart X direction
Second File, 1" delta in Y: two holes, .250" 24" apart X direction
Third File another 1" delta from previous: two holes, .250" 36" apart X

Same as above but for Y, 36" may have to be adjusted to max allowed. Shift them 1" in X

All one file: Then run a 36" on on X-axis with .250 holes every 6", then back to first hole, go in Y direction every six inches.

The reason to have these run as separate files, it will eliminate any tolerance stacking in running holes. The final pattern may flush out any internal stacking or backlash. If possible, on the last file if you can control sequence, do origin hole, max X hole then directly to max Y - hole then start working backwards towards origin alternating X & Y holes. This should maximize stack up errors an backlash looseness in gantry.

No need to move material. Using.250 holes you can use gage pins to see what exact size is or variance plus you can use two pins to get accurate measure between centers (either add .250" or subtract .250" depending of measuring between inside or outside of pins)

n a second time or a different piece and when done, put a pin through origin hole, then stack second piece on that. Using a second pin through another hole see how well they line up. If mismatched keep reducing pin size till it fits. The difference in pin size is the delta between runs.

This will not require any sophisticated measuring devices but should get you within .002"< of true.

1 Like

It seems like if the bed was not level you would see “accumulated errors”. No idea if that is the case.

1 Like

So, I burned some test patterns over the weekend. One was based on the tests described by @Photomancer, the other was a simple ruler (1" markings along a straight line) in both x and y as well as staggered cross marks every 5 inches.

(ignore the random drill holes in the material. The sheet was used before I “repurposed” it)
After measuring the marks, I can say with 100% proof that the laser is not accurate over distance. Along the x axis (44"), the laser is off by about .1" (~3/32). Along the Y axis (25") the laser is off by about .06" (~1/16").

It is interesting to note that when I say “off” what I mean is that the laser is always adding more distance than it should. It doesn’t seem to be off randomly and it doesn’t seem to matter what the distances between points are (I let the software draw from left to right and also right to left), so I have to conclude that this isn’t a mechanical problem with failure to brake or locate properly. From all of the evidence I have, it looks like there is a calculation error in the tool itself, like a rounding error or a metric to standard conversion error or something else like that. My file was designed in inches and the output was in inches. I will try another test using mm and see if that affects anything.

Any other thoughts?


When you say “adding distance” is it linear, e.g. 24" is off 4X 6". If so, then just calculate what difference is, e.g. 2%. Do your designs as parametric. Design in actual, create a separate file, then decrease by that % and you should be closer to large target. Other questions:

  • Need to run test on other machines to see what their errors are. If not constant, then once you know which machine, adjust drawing accordingly.
  • What is difference between X and Y displacement? If they are the same, then life is easy. If different, is it significant? If yes, then decide how to split the difference and hopefully acceptable.
  • Using your drawing, print it in different saved formats. Are the results the same? If they are, then something is going on at conversion or how the laser software interprets.

This would make a great DOE - Design of Experiments Black Belt project!

Very frustrating when it’s “your problem” … interesting when it’s not. I miss doing these investigations, you can be chasing gremlins

Good Luck.

I design in moonwalker units but my production files are metric, which means going through the exercise of scaling by 25.4 so that base “units” are mm. For what it’s worth I also specify base units as mm in the DXF.