Why does my print have tiny zebra stripes?

[DidierKlein 729]

It was on a Wanhao printer

One wild guess: the new Cura infill patterns might help also no?

[SandervG 1,521]

Is it infill or is it motor-resolution/micro-stepping?

Edited June 18, 2015 by Guest

[Wisar 86]

Is it infill or is it motor-resolution/micro-stepping?

 

I don't see infill as having anything to do with it!  It seems that the solution is to rotate the piece being printed such that the slicers lines look straighter across the steps.

It is still a bit confusing to me as to what is specific about the Ultimaker that brings out this problem.   There are a LOT of prints of this model out there in the wild without stripes...including the ones from my Wanhao Duplicator 4S.   Obviously it is something about the hardware / firmware interface as the phenomenon was also evident when I sliced using Simplify3D for the Ultimaker 2...and I suspect that Marlin on some other hardware might be producing a shippy without stripes as well.   Too bad I just sold my Prusa machine with Marlin last weekend!

I will stipulate that I can now print a ship without the zebra stripes...am not completely sure who gets the best answer.   MartijnvG did all the work and documented the final solution so I think his last post gets the check...but...nallath was the first to suggest angling.

Edited June 18, 2015 by Guest

[SandervG 1,521]

I will stipulate that I can now print a ship without the zebra stripes...am not completely sure who gets the best answer.   MartijnvG did all the work and documented the final solution so I think his last post gets the check...but...nallath was the first to suggest angling.

 

Whichever you believe would help a future reader!

And without a doubt a big thank you 2 @Nallath and @MartijnvG

[MartijnvG 16]

So it was somehow caused by the infill after all. Thanks for the multiple prints it's very helpfull to the community!

 

No @didierKlein it was not the infill. even withouth infill it still happens. it is the accuracy of the microstepping when a plane is close to either the x or y axis.

we need to get rid off the idea it was the infill.

Edited June 18, 2015 by Guest

[DidierKlein 729]

@martijnvg i misread your drawing then thanks for the clear explanation.

I've already had the feeling that you get better surface quality when you rotate the model, now i know why. This is something that can be tested also with fairphone cases (i print them often with 3dhubs).

If 15° is the magic number i'll have to try that, i usually go for 45°

[SandervG 1,521]

@martijnvg i misread your drawing then :)thanks for the clear explanation.

I've already had the feeling that you get better surface quality when you rotate the model, now i know why. This is something that can be tested also with fairphone cases (i print them often with 3dhubs).

If 15° is the magic number i'll have to try that, i usually go for 45°

 

I guess 15º depends on what angle it is uploaded to start with?

It would be better to have a certain angle depending on the gantry.

It would be even betterderder to solve it, like @IRobertI suggested. Either hardware or firmware related? Could it be as simple as upgrading to motors with more microsteps?

[gr5 2,167]

Not so fast Martijn!  This is not solved yet!

I looked at the gcodes for the lines that get those diagonal lines on benchy.  To the left of the arch (near the top of the arch) a typical movement in X is 7mm and in Y is .363mm.  The steps/mm for the UM2 is 80 for X and Y axes.  This means one microstep is .0125mm (1/80).  That means there were 29 steps on the left side of the arch.

So there should be 29 lines when benchy is not rotated yet the photo shows 2 lines.

looking even more carefully it looks like the Y movement between 2 "zebra" lines is about .18mm which is about 15 or 16 microsteps.  Wait!  Aha!  Are these steppers set to 16 microsteps?  If so then those lines appear to be at one FULL step.  That shouldn't be visible, right?

Also I confirmed that the slope of the walls is tilted horizontally outward as you move towards the bow, and outwards as you move up.  This COMPLETELY agrees with the theory that this is a step issue as the diagonal lines are like isolines on a contour map.  Also the outward-toward-bow angle is stronger than the outward-toward-the-top angle which makes sense as the diagonal zebra stripes are not exactly 45 degrees but are more vertical.  This completely agrees with Martin's step theory.  The only detail is that it is happening every 16 steps!  Not every micro step.  This could be a servo issue or a driver issue.

The UMO has similar mircostepping - 80 microsteps per mm (actually 78) so I'm curious to know if it has the same zebra stripes.

Edited June 18, 2015 by Guest

[joelmongeon 39]

Here is my last Benchy. Rotated at 45 degrees and the stripes are gone.

2 shells, infill at 15%, 20 mm/s @ 180 degrees.

I also switched from my black colorfabb pla to my go to Toybuilderlabs cool grey pla. It just always seems to print better for me.

Interesting that it seems like a microstepping issue. I'll be following this thread with interest!

[joelmongeon 39]

Maybe on a related note, a few months ago I posted some prints on the tips and tricks Google group.

One of the contributors noted the moiré patterns on the print and had the following to say:

"Severe XY stepper clipping, showing up as a distinct moire pattern (eg look in first pic above eyebrows and at forehead and above clavicle) -- probably improper stepper driver current tuning"

No idea if this is correct or how to diagnose and fix if it is.

[valcrow 144]

Thanks for your investigations everyone! I agree this is certainly a stepper thing as my test object has no infill. I disagree that turning the object 15-45 degrees is a 'fix' for it though. It's kind of a workaround. But it is actually still visible at those 15/45 degrees but they manifest themselves more as a denser ripple in the surface than zebra stripes.

 

The UMO has similar mircostepping - 80 microsteps per mm (actually 78) so I'm curious to know if it has the same zebra stripes.

 

I tested the curvey shape with the UMO, UMO+, Ext and UM2 and they all do it.

*edit* I found the pieces!

UMO - black

UM2 - White

They have the same zebra locations except for the center where the UMO is closer (red lines) and the UM2 is further apart. (green) curious, those are the only two differences.

I think gr5 is onto something, a gentle curve shouldn't make the steps THAT noticable. But I dunno anything about gcode or steppers...

Edited June 18, 2015 by Guest

[cloakfiend 995]

Surely the gentler the curve the more noticable the marks, think of it as the jaggies on a diagonal line with no smoothing. Or a low res elipse and look at where it looks lame most. everytime the line (purple line) breaks you get a mark...? i think. Less in the middle and more on the edges. Its a resolution thing maybe?

Edited June 19, 2015 by Guest

[Labern 774]

So I took some close up shots of some of the Zebra stripes and a slowww mooww video of the same face as it was printing. I found the showing the pulleys was the best place to film as it shows the movement better. its not very good quality video but when watched closely you can see the movements.

As you can see it is actually only producing the zebra stripes every second layer. and they are little blobs where the line is a bit wider.

 

[Labern 774]

So I thought that maybe the blobs were only showing on the outside when it was traveling in one direction as it previously would change direction every layer.

So i tried spiralize.

As the model has 2 outer contours it still changed direction but you can definatly see its every second layer

So after Valcrow modified his model for me I tried again. with it only traveling in one direction now you can see its every layer.

would be interesting to see if having it only traveling in clockwise direction for the outer contour would show the same defects?

[nallath 1,120]

It could still be micropstepping as the distance from step 1-2 is not the same as the distance from 6-7 (when using 16x microstepping)

[gr5 2,167]

It could still be micropstepping as the distance from step 1-2 is not the same as the distance from 6-7 (when using 16x microstepping)

 

Yes. Of course. One of the 16 microsteps is larger than the other 15. A DAC has the largest error when stepping from mostly ones to mostly zeroes so 15-to-0 would be the step with the most potential error. But this is unlikely the issue. It's easy to make an accurate 4 bit DAC. Instead I think it has more to do with magnets and stepper motor construction and such.

For example when power is off, the motor has intrinsic stability points - you can feel it as you turn it that it prefers to be at certain positions (steps). As you are microstepping you might be fighting that (going up hill) for 8 steps and then "downhill" for the next 8 steps. maybe. I'm not a stepper motor expert. Anyway that transition to downhill may be where the most movement occurs. Maybe a good stepper driver is supposed to compensate for that? Or be tuned? I really don't know.

[gr5 2,167]

By the way, my understanding of 4 wire stepper motors like on the UM printers is that you have 2 wires going to one coil and 2 wires to the other.  You send sine waves down both pairs of wires and to perform a 1/16 microstep you shift one of those sine waves by 1/16 of a phase.

Edited June 19, 2015 by Guest

[nallath 1,120]

I dont know what causes it, but the distance between the steps is a very nice sine curve.

[MartijnvG 16]

hmm i did some searching, and it seems the issue is indeed in the fact that the 1th or last step in the microstepping seems to be un-event to the rest of the steps.

i found an old thread about it. http://hydraraptor.blogspot.nl/2012/04/stepstuck.html we are using the A4988 chip i think so maybe this solution might be a nice way to fix it.

[gr5 2,167]

Well that article goes over my head. I'm an electrical engineer and am very comfortable with many subjects but I just don't get stepper motors. Yet. Maybe there is a more basic article that explains how they work. The wikipedia article is insufficient.

[valcrow 144]

hmm i did some searching, and it seems the issue is indeed in the fact that the 1th or last step in the microstepping seems to be un-event to the rest of the steps.

i found an old thread about it. http://hydraraptor.blogspot.nl/2012/04/stepstuck.html we are using the A4988 chip i think so maybe this solution might be a nice way to fix it.

 

Nice! I think... sooooooo. for us non-electric people this is fixable in firmware? software?

[Wisar 86]

Nice! I think... sooooooo. for us non-electric people this is fixable in firmware? software?

 

That is the question! The workaround is fine and I am happy to have a non-zebra shippy but it is only a calibration object. Sooner or later someone will not be able to just pivot things...and given they should not have to in the first place, it does seem a permanent fix would be nice!

[MartijnvG 16]

 

Nice! I think... sooooooo. for us non-electric people this is fixable in firmware? software?

 

That is the question!  The workaround is fine and I am happy to have a non-zebra shippy but it is only a calibration object.  Sooner or later someone will not be able to just pivot things...and given they should not have to in the first place, it does seem a permanent fix would be nice!

 

I think @Gr5 hit the nail on the head that the problem is in microstepping somewhere. now we need to figure out if its in the boards, firmware, motors or a combination of it. So it does have our attention. the question is how deep the problem lies.

so only a quick fix for now. but i'll try to keep you guys posted.

[Wisar 86]

So it does have our attention.

 

Great. Thanks for the update.

I wonder if there are other printers that suffer from this or is it just the Ultimaker? As I had mentioned before my Wanhao Duplicator does'nt. It might be interesting to ask the Creative Tools folks if they have seen the same artifact elsewhere. If they did you might have a shorter list of possible combinations anyway.

In any case, good luck!

[joelmongeon 39]

There is a thread going on right now dealing with the same thing with a custom 3d printer.

Seems very technical and over my head: https://groups.google.com/forum/m/#!topic/3dprintertipstricksreviews/AYLkqbn5-k0