Slight stepping artifacts; any ideas?

I just did my very first test, and it looks really cool, and I’m super psyched about how well it turned out, esp since all I have is the .125" mill that came with the machine, which is pretty fat for the size of the thing I’m carving.

The only issue I see is that there’s some slight stepping. There look like there are two kinds of stepping artifacts here, both quite small. One is this kind of ragged step along the chest in the image below. You can see that the step doesn’t seem to conform to a single Z height plane, but kind of wanders up and down a little.

And then the other is a slight step that is lined up with Z height planes (so they are waterlevel steps), but they seem more apparent on the +X side of the bird - you can see them in this picture:

And less apparent on the -X side of the bird:

No real noticeable artifacts on the +/-Y sides of the bird.

Any ideas of anything that might be causing this, or anything I could tighten up to make this a little more precise? It’s… pretty amazing as is, but maybe there’s something I can do make it even tighter.

Thanks!

For the slight waterline steps, you might see a better finish if you were to reduce the tolerance value a little. This will cause MeshCAM to spend a little more time finding the exact point that the tool touches the part.

For the chest, I would want to look that the STL file. If it were the machine wandering in some way, it wouldn’t be able to get back on track as it goes around the part. I suspect the surface of the STL fie is slightly shifted- either by design or by the triangulation from the CAD program.

That said, great work. It’s nice to see the first thing people do when they get the machine.

-Rob

Thanks for the response. I will try the lower tolerance value and see what that looks like after I pick up more raw materials other than that renshape that came with the machine :slight_smile:

Regarding the funky chest stepping, it is definitely not in the STL file or model; the model is quite smooth (it’s a very low complexity subdiv model that I cooked up and subdivided a bunch, so it is very smooth). As I look closer, I noticed that it is also on the head a little; if I had to guess what’s in common with all the locations where that stepping occurs, I would guess that it’s where the curvature is exactly X degrees (so maybe in all locations where the curvature along the X axis is exactly 45 degrees or something). Does the tool path logic change at a particular curvature (like maybe it does x axis horizontal lines when the angle is <45 degrees but switches to waterline outlines when >45 degrees or something similar)?

It also seems to only appear on the +X side of the model…

So it looks like I have a (very, very slow) slip in the X axis in the positive direction given very long times (and I believe this explains both artifacts I was getting on the bird I did before).

I just did a fairly long project (maybe 70-80 minutes), and while the output is pretty awesome,


it’s definitely not “perfect”. The most obvious artifact is that on the +X facing interior wall, you can still see the roughing pass:

whereas all the other walls are smooth. In addition, during the milling process, it was clear that sometimes cuts to deepen the pocket on the -X facing interior walls were occasionally (and barely) digging into the wall formed by previous cuts, whereas cuts to deepen the pocket on the other walls were not.

Both of those (and the artifacts on the bird I did earlier lead me to believe that the spindle drifts to the +X direction very, very, very slowly, but noticeably. Is there anything I can do to fix this? Tighten a belt? Measure the drift and put a compensating factor in somewhere?

Thanks!

Hi Rob,
I have a similar stepping artifact as KJL’s bird on the #2 head tutorial which took about an hour - my first project. It is like the spindle shifted on the x axis by .003" This does not happen all on the same z plane. The meshcam Finnish angle is 50, waterline angle is 40. The tolerance was set at the minimum allowed (larger than the diameter of the endmill) of .135" Can you tell me how to solve the stepping artifact?

Thank you

Can you post or send me your gcode for that head ?

Rob

I had the exact same thing happen with my 3d head, but haven’t had the same issues on a number of other 3d cuts. So I wonder if it’s an issue with the gcode.

I’m betting it’s in the gcode.

I’m also betting on g-code, since I haven’t had a single drifter like that…

Do you have any opinion on my machine’s slow drifting in the +X direction? Would that also be a candidate for a gcode issue, or would it most likely be a hardware thing?

I think I see the scallop changing across the edge, which may signal a problem, but I’ve learned not to draw too much of a conclusion from the finish of materials like MDF, that cannot necessarily hold the detail. If there is a shift, I’d expect to see it on the interior of the part, maybe on the edge of one of the diamonds.

This is not to say that there’s no problem, only that I can’t confirm it from the photo. I’d like to see the cut in something like Renshape, wax, or plastic since those can hold better detail. Even maple would be better than MDF. Try cutting a simple cylinder with a small waterline stepdown and see if the walls end up with a taper on them.

Rob

I will do something similar tonight in a block of Renshape or wax (probably a big square pocket). What does “small waterline stepdown” mean?

The scalloped edge you see is the problem itself - the model itself has flat walls. The scalloped edge appears during the roughing pass when the mill is making rough horizontal line passes. The waterline pass should have smoothed it out like it did all 3 of the other walls but I think the shift has pulled the head too far from the +X facing wall.

In any case, let’s just assume that I have drifting in the X axis. Is that something that I can tweak? Is there a hardware thing I can tweak, or an offset fudge number I can stick in MeshCam or anything like that?

Using a stepdown of .005", or something small, in the waterline section is the best way to show a shift. That tends to jog the X and Y axis around a lot to cut the profiles and the small stepdown gives plenty of time for any shifts to occur.

Without seeing the CAD file, I’m not sure if the scallop should have been cleared. If that represents the edge of the stock in MeshCAM then it would not be cleaned up by the waterline. If that surface is modeled in the CAD file, then it probably should have.

There’s no reason the X should be drifting under normal use so there are no fudge factors to compensate for them. The only time a shift will occur in a “normal” machine is if you take a very aggressive cut. Sometimes the cutter will grab and shift just a little each time it happens.

Send me a photo of the test cut tonight

Rob

Thanks! I will do so. Hey, you might be on to something about why the scallop line is not being cleared. It’s not the edge of the stock in MeshCAM, but you’re right - it’s not modeled into the CAD file - it’s where I set the “area to be machined”. i.e. there are no high walls in the CAD file, just a low floor larger than the stock size. The area to be machined is defined to be slightly smaller than the stock, so the remaining area makes the walls. I should not have assumed that it would have cleaned up those outside edges…

The waterline toolpath will only follow the CAD file. If you have a machine boundary (whether you define it with the Set Machine Region command or use the stock boundary) that is not coincident with a part of the CAD file, the waterline toolpath will not go there.

Don’t worry about the assumption- you’re not the only one to have made the same guess before.

Rob

Hi, just an update - I’ve done a few other tests confirming that I do in fact have an X axis drift. I did a waterline test as you suggested on renshape and it does show taper - about 1/16" taper during the rough pass job with huge obvious chunks and scalloping and stepping only visible on the +X-facing walls, and then (after rehoming), about a 1/32" taper during the waterline-only job.

A previous test showed a shift of about 1/16" over an hour or so.

I also sent an email to the support@carbide3d.com (actually, a bunch of emails as I ran more tests).

Do you know if this kind of error is fixable by me, or will I have to send the machine back?

Thanks,

Can you post a picture of the taper?

-Rob

Sure thing! Did you guys get the emails I sent to support@carbide3d?

Anyway: I wanted to do a waterline test on a 1"x1"x.625" block, so the first thing I did was to do a roughing-pass only job just to get most of the other stuff out of the way, so I roughing-passed a 1.125"x1.125"x.625" block.

Here’s the result of that:

So it is drifting pretty aggressively to the +X. Here’s a view of the side where you can see the scallops and steps:

The bottom of the block measures a tiny bit larger than 1.125" (I assume because the roughing-only pass leaves a little bit of material on? But the top of the block measures around 1.0625", so it’s off by about 1/16" by the bottom.

After this I did my actual waterline-only test, with a stepdown of .005" as you suggested earlier. This also had some taper, but not quite as much as the roughing only job. I rehomed the machine but did not remove/replace the stock or rezero the machine, and I verified with jogging the machine around that 0,0,-.625" was where it should be. This is what it looked like several waterline passes through:

You can see that the waterline block is not centered on the roughing-pass block (because the left wall of the roughing pass was shifted over ~1/16" by the drift of the previous job), so the ledge on the left is smaller than the ledge on the right.

And here’s pictures of the taper and the tapered wall after the waterline test is complete: