Vertical ridge on vertical sides

[lewis-levin 0]

This is a problem that cropped up as I tightened my belts. With looser belts, I did not see the problem as severely, but several things may have been going on...

 

I am printing letter blocks, 5cm cubes with a letter of the alphabet "extruded" all the way through the cube. To avoid a serious overhang problem, the cube is printed with the "hole" running vertically from bottom to top.

 

Three of the vertical sides of the cube print very smoothly (I've used 210c and 220c and 30mm and 50mm speeds). One of the sides always has a distinct vertical ridge. After tightening and aligning belts all the way around, the ridge is only a very small bump or identation (depending on print order and orientation of the cube on the bed). But, it is a very noticeable visual defect as shown below in 2 examples.

 

 

 

What you see (if you can see--it's not clear to me how to embed images that I previously uploaded in a gallery) is that the ridge on the vertical side always aligns with a vertex of the extruded 'L' in the center of the block. What is happening is that after the printer makes the loops/perimeter of the inner wall (the hole for the letter) the head shoots out to the outer wall to lay down the loops/perimeter for the outside of the cube. It is clearly overshooting or undershooting by a minute amount.

 

How can I fix this? I have tried wall of .8mm (pretty good), 1mm (pretty bad--probably because not a multiple of .4mm), and 1.2mm (better than 1mm but not as good as .8mm). The problem would just go away completing if Cura would put down the loops and perimeter for the OUTSIDE before going to the INSIDE (followed by the infill). Sure, there might be a bit of a bump on the inside, but no one could see it.

 

As for alignment, I have done a "hole cube" with rounded corners that pretty much prints perfectly--but it has only one surface. The problem with the letter cube is the movement of the print head from the inner vertical surface of the hole for the letter to the outer vertical surface.

 

Any help would be much appreciated!

 

Thanks,

Lewis

 

 

[gr5 2,177]

This looks a lot like the "z seam" which is where the UM pauses and moves the Z axis to go up to the next layer. I'm not sure you can get rid of it. Other slices (non-cura) will let you place the z seam at a random spot on each successive layer. The seam usually tends to be near the front left corner I believe. This means you can rotate the part so that the seam is "in the back" (but in the case of letter blocks it might be "on the bottom").

Now you mentioned about the order of printing. You can change the order. See the "expert settings" menu and then under "sequence" section you can play with the order and print the outside first/last/whatever. The term "loop" includes all inner "walls" including if your outer wall is .8mm and it is two passes, someone said the inner pass is a loop but I don't think that's true.

You could also mess with the "joris" setting but that usually causes more problems than it's worth (there are a few bugs with joris so if you are not printing a cup it will probably be worse).

 

[illuminarti 18]

Just to clarify, by specifying the print order you can control the print order of loops, perimeter and infill, but not the order in which individual 'islands' get printed, which I think is what would probably make the most difference in Lewis's case, and what he was asking about.

To test this, I created a simple STL of a square block, with a square 'ring' around it, and then sliced it in Cura 13.04, with a.) loops, perimeter, infill and b.) perimeter, loops, infill. I used 0.8mm wall setting to give two passes around the walls, and I turned off the 'force first layer' option that overrides print order.

Here's what the first layer looks like viewed in Repetier host:

The numbering shows the print order, when requesting that loops print before perimeters. It starts with the inner pass of the ring's inner surface, then the inner pass of the rings outer surface. Then comes the ring's finished outer surface, followed by the ring's finished inner surface.

Then in does the infill for the ring, so that that part gets completed before it moves on to the inner block.

Then it does the inner pass around the skin of the block, followed by the outer surface of the block, and finally the block infill.

If you change the print order to perimeter, loops, infill, then it prints the finished surfaces first, followed by the interior passes along the skins. So, reusing the same numbers, the print order becomes: perimeters first - 3,4,7 - then the block loop, 6. Then the block infill, 8. Then it goes to finish the ring - doing the loops, 1, then 2. And finally the infill 5.

So the perimeters can be thought of as any finished surface that touches empty space (even if on the 'inside' of the object). Loops are any additional passes around the circumference of the print that touch other loops, or infill (or would touch infill if the infill % was greater than zero).

 

[gr5 2,177]

Wow! Great explanation. I always thought #4 above was a loop. This really sucks that the most visible layer can't be printed last. You can print it first, but not last.

Soooo - what does the new beta cura/steam engine do?

 

[illuminarti 18]

Looks like the latest Steam Engine beta doesn't let you configure print order. But for this model at least, what you get is basically the same as 'loops, perimeter, infill' in the above example.

 

[lewis-levin 0]

Thank you for the very thorough replies. As a I watch the print I believe the problem is that the interior surface (it's still the outside of a solid model, but what we would see as inside the cube because I am referring to the extruded hole for the letter) skin is drawn first for both loops/perimeters; then the head zips to the outside to do the loops/perimeter for the outside skin of the cube. It is at this point the head appears to either overshoot (for most settings) or undershoot (with some settings). I have tried all of loops/perimeter/infill, perimeter/loops/infill, and infill, loops, perimeter (that leaves out several) but none cured the problem. Tight belts all round reduced the height of the ridge. But, it remains.

Illuminarti, your explanation is very clear. But, as you surmise, I have no control over which surface is laid down first, only which "ring" of the surface gets done first.

But, the discussion has clarified something for me. Since the problem is a bit of overshoot when the head does a long move while not extruding, perhaps I need to slow down movement from 150mm/s to something slower. I'll try and report back.

Would use of a different slicer give me more control over print order?

 

[gr5 2,177]

If you have overshoot then slowing it down might not help - instead you should reduce acceleration and possibly jerk. I would drop acceleration by a factor of 4 at least. And possibly jerk also.

 

[lewis-levin 0]

How do I reduce acceleration? I haven't seen a setting for that in Cura?

 

[illuminarti 18]

If you have an UltiController, you can change it in Control -> Motion. It is the Acc setting. Any change will only last until you power-cycle the machine, unless you save the settings (Control -> Store Memory).

If not, then you can change it using a gcode that you send via Pronterface or similar, or include in your start gcode.

The command is M204 with an S parameter.

So, M204 S1500 would change your acceleration rate to 1500mm/s². The default is currently 5000. Used to be 3000.

All of that said, I don't think that you're overshooting due to acceleration. I think it's probably just the slicer coming up with a bad toolpath. I've seen that sometimes. If you'd care to post the gcode, I'd be interested to take a look.

 

[lewis-levin 0]

Thank you for agreeing to look at this. Do you have any suggestions for how to send it? Perhaps private email? The gallery will not allow me to upload gcode or text.

I have pasted a tiny portion here (even that seems to cause this page to hang). To everyone else, I apologize for the long post.

Let me know if this is enough to diagnose the tool movement.

http://pastebin.com/T6PfgGST

 

[IRobertI 520]

Hi Lewis,

I edited your post to make it a bit shorter. You can use a service like http://pastebin.com to share large chunks of code. I added a link with your g-code snippet to your post.

 

[illuminarti 18]

THanks, I'll look at the code...

Btw, if you click the 'more reply options' on the bottom right when replying, then you get to a more fully featured interface that should give you the option to attach files to the post.

 

[illuminarti 18]

No, its not really enough code to see what's going on - just a layer and a bit. Can you email me the full file to gcode@fbrc8.com?

Thanks!

 

[lewis-levin 0]

Will do. In just a few minutes.

 

[lewis-levin 0]

Here is an update. I printed again at 200C, 40 mm/s, speed (nonprinting) down to 80 mm/s, the cube rotated 15 degrees off x alignment, wall thickness .8, and double skin OFF. The vertical ridge is pretty much gone. I can detect the merest trace of the ridge on the curved surface of the chamfered corner, but I know what I am looking for. It really isn't noticeable.

Like noob, I changed too many things so I don't know what really is responsible for the improvement. I think, from watching the print, that is the combination of 15 degrees off axis and wall thickness of .8. Previously, I thought thicker walls would come out nicer so I tried 1.0, 1.1, and 1.2--all of which resulted in 3 (or 6 with double skinning) passes to do perimeter and loops. I think the odd number affected the order of doing the wall of the extruded inner cut (for the letter) and the outer wall. With 2 passes to create the walls, the order was altered and the outer perimeter was done before the inner perimeter.

But, now, I have more vertical striations on the smooth vertical faces, which I read is caused by friction of the head pushing plastic along the surface causing the plastic to "bunch up". I am trying some cubes at different speeds and temperatures to see how to eliminate this. What's interesting is that the striations are very noticeable with black PLA and non-existent with silver PLA, which lays down incredibly smoothly at a wide variety of temperatures (205 to 220).

 

[illuminarti 18]

I've never heard the 'bunching up' theory. Which isn't to say it isn't true (although I suspect it's more likely some sort of mechanical resonance). Where did you read about that?

[Nicolinux 288]

@Lewis: What do you mean with "double skin OFF"? I have the exact same issue and would like to get rid of the seam:

http://umforum.ultimaker.com/index.php?/topic/2786-please-analyze-my-prints

 

[origamimavin 0]

But, now, I have more vertical striations on the smooth vertical faces, which I read is caused by friction of the head pushing plastic along the surface causing the plastic to "bunch up". I am trying some cubes at different speeds and temperatures to see how to eliminate this. What's interesting is that the striations are very noticeable with black PLA and non-existent with silver PLA, which lays down incredibly smoothly at a wide variety of temperatures (205 to 220).

 

Could this possiibly be due to diffeerent filament diameters? Try remeasuring and changing that setting within Cura. I had trouble in the past of having too small of a number set, and it was spitting out more plastic than I wanted and gave some issues. Once properly measured, it's able to spit out the proper amount and bunch up less. Depending on what I'm printing, I actually find that slightly underextruding has better results than overextruding.

 

[origamimavin 0]

@Lewis: What do you mean with "double skin OFF"? I have the exact same issue and would like to get rid of the seam:

http://umforum.ultimaker.com/index.php?/topic/2786-please-analyze-my-prints

 

There's a setting to make the outer layer print at half heights to simulate a smaller layer height. For example, the inner layers are all printed at .2mm, while the outer layer is printed at .1mm heights. I believe tthis is what he's talking about.

 

[Nicolinux 288]

I thought that it might be a setting - but where is it in Cura? thx

 

[illuminarti 18]

It used to be a setting in Skeinforge-based Cura, but it was removed in the SteamEngine builds, because it wasn't terribly useful, and caused problems.

It wasn't truly slicing the outside at half-the height, just taking the one perimeter and printing it twice, exactly the same, but at half the normal height each time. As such it didn't help overhangs/near-horizontal-curves at all, and the implementation tended to cause problems with the print quality.

 

[matt-gajkowski 0]

Any chance there will be a 'true' half-height-skin setting appearing in the foreseeable future? Or equally good, an "infill every __ layers" setting? I mean, most people want the quality of a fine layer height but the speed of a larger one, and this would be perfect for those of us who don't want to compromise either one.

 

[gr5 2,177]

In other words, Cura 13.06 doesn't have the "skin" feature anymore.

1) To get rid of the vertical seam at one corner, print slower so there is less oozing when you pause. Also double your Z acceleration to 200mm/sec/sec so that it hops up to the next layer faster.

2) To get rid of ringing, which is vertical seams that are stronger at a corner and fade out, reduce your XY acceleration by half.

3) To get rid of vertical seams that are part of the CAD model, you have to edit the CAD model.

4) To get rid of the remaining vertical seams you have to play with temperature or flow (see second picture with white cubes):

http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/

I'm wondering if lowering flow instead of changing temperature would have worked better in the above photos.

 

[Nicolinux 288]

That's perfect for the cheat sheet - thanks George.

 

[gr5 2,177]

Any chance there will be a 'true' half-height-skin setting appearing in the foreseeable future? Or equally good, an "infill every __ layers" setting? I mean, most people want the quality of a fine layer height but the speed of a larger one, and this would be perfect for those of us who don't want to compromise either one.

 

This is basically impossible to do. Or damn hard anyway. When you slice 2 consecutive layers, one might have just finished the top of the part.

Or the 2 layers may be significantly moved in X or Y from each other. Where do you put the infill? Should the infill go only inside the inner of the 2 combined layers? So it only touches the edge of one of the shells?

The way the current slicer seems to work is it seems to be relatively stupid about what is going on in the layer above or below. So I think it would be hard to do even this "do the smaller area" infill option.

But I agree it would be great if Cura could make half as many infill passes as shell/skin passes.