Is it normal for an idler to have this much vertical play? The one on the left has lots of vertical play. The other idlers don’t move up and down like this (such as the one on the right)?
Reason I ask: I have diagonal artifacts in only one side of 3d prints, and belt tension and all the other suggestions didn’t resolve it, so when I was inspecting the idlers to clean them and checking the path. I noticed this one moves a lot, and I’ve read that it can cause the type of artifacts I see? Thanks for any help.
It’s hard to see the right idler because your thumb is in the way, but I assume it’s tight.
In early SV08 some users put a washer there because they thought the same as you.
I don’t own an SV08 so I don’t know for sure.
There are YouTube videos out there on how to install a spacer.
Great questions, I didn’t get a chance to observe it put back together yet - that would hep diagnose. I’ll put the shims on top or bottom based on the position of the belt feeding into adjacent rollers, it looks like you could be right, but I’ll double check when it goes in that it doesn’t send the belt skew into the adjacent ones.
Great point, and thanks for responding, I was looking at all the idlers thinking…so much good quality went into other parts, why introduce so much low precision and sliding with idlers instead of bearings? (That screw also holds a top plastic plate on that would have covered the idler on that side, which is why it’s temporarily out.)
The main question is, could that affect or create the diagonal artifacts I’ve been seeing? At this point I can’t see anything else in the belt path that isn’t clean or standard, and I’ve checked tensions are right and equal.
In my opinion the 1mm axial move of the idler could cause visible “salmon skin” but not an artifact you can feel with a fingernail.
I still cannot come up with a root cause for a bad side with a parallel face that is good and doesn’t include Z axis moves. The mechanical and motion environment for 2 parallel vertical planes should be identical.
What happens of you move the print such that the “good side” is not in the plane that the current “bad side” is in. I’d also be interested in rotating the model by 15° and 30°. Does the effect wrap over the corner?
That’s what I mean, no matter how I rotate the model, the artifact is always on that particular orientation (front, left side, z1 rail). That hit’s strongly at one belt impeding, since during that movement on the xy plane that is drawing walls facing that direction, only one belt is active (since in Corexy a diagonal move is only one belt). However, as you said, it’s confusing why the other side doesn’t have it. My only thought is because on that side, it’s proceeding in a different direction on the diagonal, so the interference the belt is encountering is only one way. That might hint at it riding up a non-perpedicular or oblong idler, since moving in one direction shifts it to a flange. However, I made sure the front idlers aren’t riding flanges, and I can’t see any impediments. Here’s what the artifact looks like:
I don’t see any real issue. Photo is from front of bed? the “bad side” at 7:30 on the clock?
I’d hoped to see a diagonal pattern on that side.
I’m really at a loss but I’d print this next. Put the flat side on the 45° plane opposite the bad side. The slicer should make each layer a polygon with 4 straight sides. Goal is to see how far off axis the surface has to be to not show artifacts.
Lol wait…I don’t mean this in any type of way, I mean honestly: you think these are not immediately noticeable or no issue??? Photo from front of bed: These sides facing the camera and a bit to the left are drastically rippled and horrible surface texture compared to the other sides of the cylinder which are smooth. Look how horribly wrinkled the surface of a smooth cylinder is lol. This is a huge, very very noticeable artifact to the point of useless prints!
Also the problem with this test shape you provided - great in theory. But because you decimated it and the triangles are very large in comparison to the curve (low poly count), that alone on this scale is going to register as facets and cause some artifacts that aren’t smooth and make it harder to see clearly. I could easily whip up the same test shape in Zbrush with more tris…But I think we can skip ahead to this finding: we have isolated it now to one belt movement. Because only one belt turns and runs when the nozzle moves along this diagonal. It is strongest at the exact 45 degrees when only that belt is running and it fades at it reaches the front and the side, meaning as the other belt starts to engage, it starts to disappear.
So the main question now is, what is causing this? Seems like belt obstruction but I’m looking at the entire path and dont’t see hang ups? And also the less severe but noticeable z-banding I’m getting on all sides besides the one having this huge artifact, do look similar to OP’s. They may be related or may not. Some kind of mechanical frequency unresponsive to input shaping in both our cases?
I included the .step file (which Orca handles natively) to get the minimum shape errors.
I agree BUT neither you or I can come up with a plausible root cause. I spent 20 years as a test engineer and have been here MANY times before. The scientific way forward is to learn how to turn the issue ON and OFF. The fact that the parallel side is unaffected strongly refutes your theory that it is the one belt/motor system.
Given enough different tests eventually the parameters that turn the issue ON/OFF will converge on the root cause.
You don’t like the helical face test try these;
Print a square spiral vase, reverse the travel direction and print it again.
Print the original cube off a raft with no Z correction enabled.
Print cubes near the 4 corners of the bed
Or go the non scientific route and “try everything”
Reverse the position of the motors and belts.
Swap the right and left linear rails.
Raise one side of the printer 12", then the other.
I see your point: Here’s my theory as to how that could work. Let’s say for instance a belt is drifting toward one flange in one direction and the other in the other direction. That’s one belt, and the error only activates when it’s traveling in one direction. In general when the slicer does outer walls, it travels in one direction around the walls, in a continuous loop. So this very much fits and would suggest one belt still right?
I agree with your point that we need more information, I’ll try these prints. Right now the whole thing is taken apart to try to search for belt path problems.
Okay this is really interesting diagnostically and the first time the orientation of the artifact changed:
I manually made the wall loop direction switch directions (clockwise instead of counter clockwise) - now the artifact is primarily on the front and right side of the bed instead of the front and left. In fact, it is a mirrored opposite, meaning switch the direction of the wall loop made the artifact mirror itself exactly over the center line, from left to right.
I also did what you said and printed the cube in four corners, unfortunately it’s exactly the same in all corners.
I also tried to print the step file you sent, unfortunately it also had large, irregular triangles when importing it default into orca, so not sure how to get about it.
Good idea, to give it a full pause after a full z hop to rule movement totally out…The thing is if I paste this into the layer change g-code, it runs after the layer step up automatically runs and there’s no code I can see to stop that, meaning the z axis has already raised to 0.2 mm higher and then this runs (going to 0.4 mm higher than the last layer). I could run it as Z+2 and then Z-2, since it’s already moved up. But since it already lifts, in any case it wouldn’t isolate z moves from print moves. It would only stop any z vibration from contributing (the zhop and pause), but we’d just be ruling out z vibration right? I think your idea is smart, but I’m just not sure how to implement that correctly?
Or maybe your idea is that the raising movement is slightly trailing into the start of the X-Y movement of the new layer? I think basically this just lets the vibrations settle before the new layer starts right?
Okay update, I’ll keep this off the other topic so that it can focus on z resonance: I ran the code and observed it stopping and lifting and waiting before starting each layer (so no z movement anywhere near the x-y movement, totally still) and the issue is still produced. So we’ve ruled out Z completely, and it’s something with the A-B belts. Additional info would be, the z-resonance would have calmed later on in a given wall path, at least it would occur stronger more near z movements (right?), so that’s another indicator, even without this, nothing to do with Z. Lastly, I did re-tension and equal all my z-belts, still happening, and no difference in pattern (which again, if it was off even, with this heavy adjustment, we would have seen a slight change in pattern produced). So this is all useful data, everything pointing to A-B belts.
