Why does my print have tiny zebra stripes?

[DrR1pper 4]

So with the zebra stripes related post out out of the way, here are my results from my test designed to observe how print temperature and print speed might affect the visibility of layer lines/banding on all my um2+ prints.

The test object consisted of a 75mm tall, 30x15mm rectangular tower. On both printers, printed with only 1 bottom layer, two perimeter wall and hollow for the m200 and only a few percentage grid infill for the um2+ prints (as i printed the um2+ prints first only after to remember that the lowest infill setting available for the m200 is equivalent to something like 30% for the um2 but since i don't think this matters much i did not bother to repeat the um2+ prints with truly zero infill). A layer height of 0.20mm on the um2+ and 0.19mm on the m200. Print speed was "Normal" for the m200 and 40mm/s and 20mm/s for the two um2+ prints. In both um2+ prints, the tower was divided into 5 printing temperature zones along the z axis starting with 220*C for the first 15mm, then 210, 200, 190 and lastly 180*C for the last 15mm. Temperature control on the m200 was not possible but also not a variable i needed to test as the m200 does not have a layer line/banding visibility issues with the unchangeable presets in z-suite anyway.

Here are two pictures of all four test print objects first taken with 50% lit ambient room lighting. From left to right, z-abs, z-ultrat, PrimaValue PLA 40mm/s, PrimaValue PLA 20mm/s. UM2+ prints were sliced with s3d and prints speed was set to constant (i.e. same infill, outter/inner perimeter speeds).

I also created a video to try and demonstrate the layer line issue better.

 

I forgot to conclude in the video that print temp and print speed appeared to have no affect on the visibility of layer lines/banding on my um2+.

Here is another video demonstrating the visible play in my z axis. Looks like it's around 1mm. Conversely the m200 has zero visible play here (but not shown in the video). Not sure if this could be an indication of the z axis being the problem or not. When the bed raises and lowers, the lead screw seems rather straight. I do remember it spinning off center when i first received it but it seems to have straighten up on it's own over time. Odd but not sure if relevant.

 

And here is one of the um2+ prints next to the z axis lead screw. Is it just me or does it look like the distance between each of the larger layer banding lines on the print matches the pitch of the lead screw?

Lastly, i printed this object for a friend last night with 0.3mm layer height. I'm including these because they highlight the randomness of the visible layer banding lines when in the right lighting (full ambient room lighting with camera flash in the second and third photo down)

and during printing...

Notice in the last picture vs the second last, there is a shadow being cast onto a small section at the top of the print which makes the layer lines looks very neat compared to when there is lighting shining upon it (and almost everywhere else on the printed object) then the layer lines/banding is very visible.

Finally, photos of each with a single direct camera flash and very low ambient room lighting does a very good job of reducing the visibility of those layer lines on the um2 print compared to the m200 print (last). (I snapped off the top of the first print...not relevant but just so you know it's the same object as above).

But put them all side by side in the same non camera flash and well light room lighting and the layer lines/banding on the um2 prints really stick out.

Sorry for that wall of text, photos and videos. I just wanted to be really thorough and try my best to leave no stone unturned. The question i wish to ask is whether these banding lines are in fact quite normal/typical of ultimaker 2/2+ prints? I've done some looking around online of the print quality on other um2's and from what i can see, i'm somewhat inclined to believe it is typical (however much i wish it not to be the case). I've actually come to really like (dare i say love) this printer for many reasons. I think it looks really freaking cool, i love how quickly the hot end heats up and cools down when needed, large build volume, oh and it looks really cool (did i say that already?). But to me it doesn't print as nice a surface finish as compared to the m200 which is a bit of a big deal to me. However, I've still really come to love it in many ways for many reasons, even if it leaves a lot of stringing in my prints also.

Anyways, I'd really like to hear an honest take on whether what i am seeking from the um2+ is in fact unattainable or not because if this is not how the um2+ prints should turn out, then i wish to rectify it ofc.

Big thanks in advance for any help/light you guys might be able to shed on the matter.

Edited June 20, 2016 by Guest

[gr5 2,172]

And here is one of the um2+ prints next to the z axis lead screw. Is it just me or does it look like the distance between each of the larger layer banding lines on the print matches the pitch of the lead screw?

The Z screw is triple helix. So it goes down 3 "bumps" for every rotation. So it seems most likely that this is a coincidence. For example if the Z screw was warped in the shape of a "C" I would expect patterns to repeat once every 3 bumps on the z screw - not every bump. But maybe I didn't think of something.

[DrR1pper 4]

Oh right. Thanks @gr5. So you're saying if a repeating layer banding problem were to be caused by the z screw, it should be at least 3 times the distance between each of the banding lines seen on my prints?

Any ideas what might be causing it then? Or is this typical of um2+ prints for whatever reasons?

Thanks again!

Edited June 20, 2016 by Guest

[neotko 1,417]

I bet that's the bed bangbang mode, specially with a circumference that allows the heat to scape faster. Try to enclose the printer so it can keep a more stable heat or use a firmware that allows to activate the pid on the bed.

[DrR1pper 4]

I bet that's the bed bangbang mode, specially with a circumference that allows the heat to scape faster. Try to enclose the printer so it can keep a more stable heat or use a firmware that allows to activate the pid on the bed.

hmm. Interesting theory @neotko. So if it's caused by bangbang mode, then the frequency of the banding lines should match the frequency of the oscillations from bangbang mode, right?

I can test this theory by simply printing without bed temp, right?

Edited June 20, 2016 by Guest
wording error

[neotko 1,417]

Indeed the best way to test it it's by not using it.

[DrR1pper 4]

Just started the same gcode now with bed temp manually set to 0*C.

[DrR1pper 4]

It's printing right now, but i totally forgot, i made two prints from the um2+ remember. One at 40mm/s and the other at 20mm/s and in both examples the banding frequency was the same.

Also, just checked on print and looks like no change to banding pattern and frequency. :(

I'll do the print with double print area to test if it might be something with the extruder gear, but i noticed the other day that there is 1mm of momentary play when changing filament travel direction.

But thanks for the suggestion of possible cause.

Edited June 20, 2016 by Guest

[DrR1pper 4]

Very interesting developement. I modified the tower to double the perimeter length (by increasing the x and y dimensions 41.4%) which should roughly double the print time per layer and used all the same slicer settings as before and the layer banding doubled on that print in frequency. Take a look....

I'm thinking this is caused by inconsistent extrusion rates now caused by some sort of inconsistent feed per revolution of the either the knurled feeder gear, extruder gears or extruder motor itself (or if i should be so lucky, all the above, lol).

Thank you @neotko for suggesting bangbang mode as a possible cause as that spurred me on to discover this.

[DrR1pper 4]

Has anyone ever experienced this before and know how to fix it please?

[SandervG 1,521]

Seems like the topic has slightly shifted from Zebra-stripes to banding as it appears?

Have you tried keeping your bed springs tight? When leveling tighten them first and use any room to loosen them to level the bed.

Have you tried different positions on the bed, or a different slicer?

It is kinda weird that the banding doubled on your bigger print..

Have you tried different nozzle sizes to see how it behaves then?

[DrR1pper 4]

Seems like the topic has slightly shifted from Zebra-stripes to banding as it appears?

I'm really sorry about that. Do you think it best that i start another thread?

 

Have you tried keeping your bed springs tight? When leveling tighten them first and use any room to loosen them to level the bed.

I think my bed springs are tight already. When i first calibrated the bed, i made sure to thoroughly tighten them and apart from re-leveling the bed a few times before i got the height just right at the beginning (with each time making sure that i had to tighten the springs if i needed to change the level rather than loosen then), I've not needed to re-level the bed since.

 

Have you tried different positions on the bed, or a different slicer

I have tried this before but not with this model. I can do that though, no problem. I'm not so sure if i noticed any significant changes in the appearance of minor banding lines though by doing this. More testing would be required.

 

It is kinda weird that the banding doubled on your bigger print..

I agree it's kinda weird. The best theory i can come up with based on the evidence thus far is that it might be due to inconsistent extrusion, with it cycling between under to over extrusion (in perhaps a sinusoidal manner) and then repeating itself over the course of a print.

The two facts that printing the same test object with the same slicer settings except halving the speed did nothing to change the frequency of the banding lines and secondly that enlarging the test object just enough so that each print layer requires twice the time to print (for the same print speed and slicing settings) caused a doubling in frequency of the banding lines, the only possible culprit i can think of is that it has something to do with the extrusion rate being inconsistent. Be that due to hardware or software, i don't know for sure but i am more inclined to think it'll be hardware given the two aforementioned observations.

 

Have you tried different nozzle sizes to see how it behaves then?

I have not but this is a good idea/test to perform as it would either break my theory or reinforce it's probability of being the right theory. If i use the 0.8mm nozzle on the same test object and if my theory is correct, i should expect to see the banding frequency halve compared to the 0.4mm nozzle print. I will test this soon. Thank you for the idea @SandervG!

Edited June 21, 2016 by Guest

[gr5 2,172]

Yes. new topic please. Put a summary at the top of what you are worried about at the moment (photos of the latest banding) and then post a link to it from this topic so we can all go over there and subscribe to it (or not). I think your over/under extrusion theory has merit and you should check the nozzle temperature for a few minutes while it's printing to see if it is oscillating. You may need to adjust your PID values on your nozzle.

[JJB 2]

 

Hi guys!

I wanna add some information regarding the Zebra Stripes.

I was having this problem and I didnt know what it was until I came across this thread.

I did the DRV8825 hack posted HERE and I can say it's 80% gone.

Heres a picture where you can see the problem to the right and the new print after the hack to the left. (I stopped the print because the result was already clear)

I know it looks like the problem is gone but after inspecting the new print you can clearly see that the stripes are there, the agle changed and they are more subtle:

It is important to mention that I only did the DRV8825 hack to the extruder driver.

 

Hi JJB,

This driver is in the "Pololu" family, and have the same feature of configuration (as we name it) as the one used in Ultimakers late family 3D printers (as far as I'll know). Actually your "hack" is just the same thing as I did with my printer, that is using "PoLolu A4988".

You said it improved your print about 80%, yes that was missing step, -but then there is some sign of it still, right. When things improve like this, we might wonder; is there still some missing steps? No there is not! So what could it be then, the answer is "production tolerances". It is maybe a cheap (Chinese?) stepper motor.

The truth is; those steppers might not stop at the "exact" same position when the current is reduced. This is actually true for every stepper motor, if we go "deep" enough into it...

This is why we still may have sign of those stripes, if we look close enough into it.

You can have a better stepper motor, or you can gear down the speed by 2 (sure using belt for this), in order to improve the print further.

All this said, however, most of the steppers today is made in China.

So, if you like to have “real good” quality, the stepper motors used here for X/Y play an important role.

Thanks.

Torgeir.

 

By "hack" I mean this:

I dont use chinese drivers. Everything is original.

I did the "hack" and the zebra stripes are almost gone.

Changing from drv8825 to a4988 (originals) also "solved" the problem.

Changing from drv8825 to a4988 (chinese) made it worse.

[axjurado 0]

What drivers are you using?

I have been working on this issue for the past few days. My board uses A4988 Allegros. There is a resistance there that needs to be shorted to get rid of the issue. If you look into the documentation it would be Rsoc. Just short it down and the driver will stop missing steps and creating the ripples. Noise will increase though.

[Titus 170]

Ok, time to jump on the band wagon!

Had a talk with @SandervG about some upgrades and hacks, cura feature topics etc. He pointed this topic out(which I had completely missed). Before I start some tests myself, I'd like to tag some people here. Reading through the topic I noticed it was mainly UM2 electronics and drivers, though people had the same on UMO right.

I really wonder how the guys with the SilentStepStick drivers are doing, as I recall they were doing the microstepping differently: https://ultimaker.com/en/community/11571-step-by-step-installation-of-silentstepstick-drivers-on-umo?page=1

@drayson @ataraxis could you guys browse this topic, print a benchy and perhaps even compare old to new drivers, as they are drop in replacements.

[gr5 2,172]

Hi Titus - just make sure what you are trying to fix is this 16th step issue and not one of MANY other issues that cause vertical lines. The lines are diagonal on benchy only because the side of benchy is *almost* but not quite lined up in the XZ plane but is instead tilted slightly around the Z axis (normal rotation on the bed) and even more strangely the wall is tilted out of vertical! only because of both of those tilts do the lines show up as diagonal. Walls lined up with the printer won't show any lines at all.

[chrii 5]

any news from the Decay topic from UM side or related to the UM board ?

Chri

[lancelet 0]

I'm also interested to know if there are any updates here.

I've been working on getting very high quality prints of a gently-curved object acquired from a photogrammetric scan. It's reached the point where the "zebra stripes" are far more visually distracting than the layer artifacts (which are barely visible at all at 0.06mm layer height with a 0.25mm nozzle).

I'm considering buying a "knock-off" board to try out the resistor mod. So any updates on people's successes (or failures) a few months down the track would be very useful.

Edit: below is a close-up of the zebra-stripes I'm getting. (This particular print also has one dodgy layer. Please ignore that; I have others without the dodgy layer. This one just happens to show the stripes particularly well.)

Edited October 24, 2016 by Guest
Added image showing zebra stripes.

[DrR1pper 4]

By any chance have you tried the mod/hack yet @lancelet? Also, very nice looking print. What was the print speed?

Edited November 6, 2016 by Guest

[cloakfiend 995]

im guessing that a larger 0.4mm nozzle would help, but reduce the surface quality? I simply cant avoid this either and have simply learned to live with it, but I dont print with a .25mm nozzle.

[DrR1pper 4]

Is it just me or does it look like the UM3 has addressed the zebra stripes issue?

Edited November 7, 2016 by Guest

[Labern 774]

Is it just me or does it look like the UM3 has addressed the zebra stripes issue?

 

Nope. Not yet. It's explained a bit more in this UM3 thread

https://ultimaker.com/en/community/23401-inside-the-ultimaker-3-day-4-electronics

[gr5 2,172]

 

@lancelet - do you have a friend who is good at soldering? I recommend you go for it. That indeed looks to me like the missing step issue aka "zebra stripes". You can always undo the change. Just use powerful jewelers or reading glasses and take your time and use the right equipment.

[Torgeir 242]

Hi Folks,

As I've finally made all thing to work in the “Chinese extruder head” with an Olsson heath block and Laberns heath duct, thanks Labern, I'd like to shear some of the final result.

Since I've modified my printer sometime ago, I've never seen any sign of the zebra stripes anymore.

This first picture is a few prints of the “Calibration Block for SLA printers” with some notes. For the note in this picture, correction; it is possible to see through 3 holes...

The second one is a picture of the “MakeRook prints”, yes the print that won the 3Dexpo price made by GuyS in the Team Ultimaker.

In this picture there is three "Make:Rook" prints.

The one in the middle is the first I printed and I thought, wow, how small this thing is.. Until I realized this was the version made for the SLA, well its a charming little thing that’s missing the on top text.

The little one is only 26 mm and the normal one for PLA is 50 mm.

The left print is made at center of the bed, while the right is made at the very far down right corner that’s have less vibration due to “in fill” hit on retraction.

Sure all this rocks.

Good Luck.

Thanks.

Torgeir.