Occasional Banding on Prusa i3

Hi
my Prusa i3 Hephestos has been producing solid quality prints from the start.
Only the surfaces have alway been uneven. It looks like everytime the printer starts drawing the outline there is a little overextrusion.


The threaded rods are not very straight, but wouldnt this cause wobbling instead of random overextrusion?
There are a few things I already tried:
- I replaced the lm8uu bearing on the y- and x-axis with igus bearings
- The printer is now running repetier instead of marlin
- The frame is reenforced with two aluminum rods
- belts are tightened
- the problem occurs with any filament and slicing software (cura, slic3r, mattercontrol, simplify)
- I changed temperature and extrusion multiplier during a print, there was no noticable difference
- in vase mode the surfaces are really smooth
The next thing I am going to do is replacing the threaded rods with trapezoidal rods. But if that doesn't work I am out of ideas.
Do you think the upgrade on the z rods would be worth it or do you have any other suggestions for me?

This can be caused by
- Backlash. If a layer perimeter is printed in reverse on some layers the backlash may be different on those layers. Backlash can be produced by any component that has a springiness/looseness, including bearings, belts, idlers, print plate moving/print plates clamps not strong enough, hot end clamped to the extruder, bowden tube swinging to and fro freely etc. In vase mode the motions are exactly the same so my guess is this is the problem. Readjust extrusion multiplier when you've eliminated the backlash, it may be you need much less.
- Over extrusion. Buildup of over-extrusion will push the perimeters to the outside, where there's no resistance. With over extrusion it greatly depends on the previous layers and perimeters near the infill and the infill itself. This can be exacerbated by blobs on retractions etc.
- Heated bed large temperature swings making it bend up/down, especially with bang/bang temperature control
- Z motor missing steps

The bend z rods will create more regular patterns, though I can see signs of this too. There's a part on thingiverse that can alleviate this by printing studs for under the x motor and x idler with two nuts per rod with springs in between. The x/y motion will be decoupled from the rods as it just slides around under the x motor and idler.
Z decoupler

Edited 1 time(s). Last edit at 09/25/2015 09:49AM by imqqmi.

It can also be caused when using bowden, if after retraction/resume it extrudes less because the bowden system introduces springiness. If there's any up/down motion in the bowden tube, make sure the tube connectors are tight. I put some single core wire under the lip and solved this problem mostly. Make sure retraction distance is around 4-5mm.

Thanks for your answers.
@imqqmi: The Hephestos has a direct extruder, so this should not be the source of the problem
Backlash seems to be the most logical explanation.
It just fixed a few parts that were a bit loose and tighened the belts again. But the test dice came out exactly as the previous one.
There is really not much left for me to improve on the mechanical side. Do you think the backlash ompensation in the repetier firmware is worth a try?

Try running a PID tune on your heatbed. I know that if mine is off so much that the heat goes up/down by one degree, I get horrible prints.
Are you using a heatbed relay or direct power from the control board?

---

Master Tinkerer

I think there are several issues at work here, i still can't rule out backlash and over extrusion based on the pictures above. The top perimeter of the dice are wobbly, not straight, the top infill is blobby and a bit of a mess. You should be seeing clear straight lines from the perimeter and infill.

What about the x,y,z,extruder acceleration setting? Can you show us a picture of the bottom and/infill, and the perimeter of 90 degrees corner? The holes in the dice show some waves just after the corners.

Did you try different hot end temps?

Did you try a different brand filament?

Is the extruder idler too tight or not tight enough? Both can show similar problems.

Some of the lines on the dice occurs on the same hight, that will indicate trouble with the z - movement, like binding.
You could also try to set the layer hight to a multiple of your mm / full step. In theory this should not make any difference but, at times, microstepping don't agree with reality that well.

@Montiey: I don't have a heated bed either. Sorry, I should have described my printer better
@imqqmi: It tried different temperatures (195, 200, 215) without a difference. The white filament is better quality the lines are still visible.

The XY acceleration is set to 1000mm/s² and 100mm/s² for Z. 5000mm/s² for the extruder.
I think the blobby top infill was due to the slic3r settings. It is much better with Matterslice.

Yesterday I replaced the threaded rods with new ones, sadly the prints came out exactly the same as with the slightly bent ones.
Then I activated the backlash compensation settings int the firmware, still no change.
The last thing I did was set the infill overlap from 0,06 to 0,01 and the travel speed from 100 to 50mm/s.
And that somehow did the trick. The test dice was much smoother than the previous ones. Unfortunately I printed them in white filament and I can't get a nice picture of it.
After that I printed the benchy which also has very smooth surfaces now.




Edited 1 time(s). Last edit at 09/27/2015 07:49AM by MaxMKA.

Ok, that could point to over extrusion during sharp turns. You could tweak the jerk xy parameter. Increase the speed say 10mm/s higher. I can definitely see some backlash on the top infill. When you fix that the over extrusion will be improved during sharp corners too as you'll be able to lower the extrusion multiplier. What had the most effect: the overlap from 6% to 1% or the travel speed? At such slow travel speeds I can't imagine it could have much of an effect.

Z acceleration looks on the high side too. I got mine on 20mm/s², and extruder looks high too, I got mine on 2000mm/s². Vxy-jerk at 42mm/s. I've got overlap at 13.5% (slic3r) down from the default 15%.

I don't know what setting did it. The travel speed could have been the solution because it always looked like the extruder was moving a bit too far before printing the perimeter.

I will try to lower the z acceleration, maybe that will help. My xy-jerk settings are at 6mm/s. I thought that smaller values make for better quality, should I increase that value?

Not necessarily, Vxy-jerk is a feature that makes the assumption that the set speed for cornering, ie one motor slows/stops while the other goes on moving the speed is kept at most at Vxy-jerk.

At 6mm/s what happens is that if you have an infill speed of say 30mm/s, it slows down to 6mm/s in the corners. The pressure in the nozzle can't clear in time to print at 6mm/s and extrudes more filament than is needed or wanted causing blobs in the corners. This also happens at the corners of perimeters.

The more constant the motion is the better the print quality as it's much easier and predictable to do with the extruder. I wouldn't make Vxy-jerk higher than your average print speed but not lower than 70%. If you have on average 30mm/s you can use between 20-30mm/s for Vxy-jerk as I understand it. I print at 60mm/s for perimeters and 80mm/s for infill. Vxy-jerk is 42mm/s. At that point most of the blobbing in the corners stopped appearing.

Thanks for the explanation!
The surfaces look much better now:

(yellow: after optimization)
Finally I can do some challenging prints:

That looks great! The ears overhang looks good too, did you use support structures?

No, it printed without any support.
The Hephestos has a pretty effective cooling fan.
Thanks again for your help!

Impressive! Glad to help!

Hi

What did you to get good results with yellow owl?

thank you