Top Layers not touching - UM2

[varsanyy 0]

now i've successsed in uploading the photo

 

[shurik 87]

Based on the photo, you've got a little bit of underextrusion or it seems so. You may try to raise the temperature a little bit, combined with slower speed and the cleaned nozzle. It will make the bridges you have trickier, but this is the trade-off you should try to balance.

 

[gr5 2,178]

I'm not certain but it looks like the extruder is skipping back and causing extrusion in a few areas. You might want to consider sanding, filling, and then painting.

Anyway, if it *is* skipping backwards there are many causes of underextrusion. The quick fix is to print slower and hotter. I would try 240C at 40mm/sec.

Your current settings of 210C, 55mm/sec at .2mm is right at the limit of what a um2 can do. You are right on the blue line of the graph (first image in below link). I recommend printing at half that speed of the blue line so for 55mm/sec you should be fine at 240C.

http://umforum.ultimaker.com/index.php?/topic/4127-um2-extrusion-rates-revisited/

 

[3dmaker4u 0]

Just an issue that does not let me sleep

It seems to me really peculiar to use at 240 C a material that is rated 190-210 C, even for workarounds. Actually I understand that 220-230 C are "normal" in UM world

It also seems that UMs, in general, report higher temperatures than other printers. Could be more accurate readings, but that's the point, it means UM prints at considerable higher temperature.

Why is that?

I normally print ColofFabb PLA at 200 C for 40 mm/s, 0.4 mm layer width and 0.2 mm layer height (3.2 cmm/s) with no underextrusion. Hotend is either E3D v6 or a J-Head version.

 

[DidierKlein 729]

Hi guys,

Finally i tested this issue on my printer. I probably have to do another test to compare the same cube.

I download the calibration_cube.gcode linked several pages ago and i modeled a test cube but it's smaller

My test cube was sliced with Cure 14.09:

0.2mm layers

50mm/s

24% infill

0.8mm shell thickness

Both cubes are printed with the same material (Colorfabb Traffic Red) at 220°c.

I have the latest firmware upgrade (Cura 14.11_RC7)

On the left my cube on the the right the gcode downloaded.

 

My test cube looks pretty good to me (not perfect i agree).

 

On the gcode cube we clearly see layers not touching.

 

I will do the test again with the same size cube as the gcode to really compare but i'm really wondering if this is not more of a software issue than hardware?

 

[Nicolinux 288]

This is getting a little crazy because there might be 3 distinct issues going through this thread:

1) Underextruded top layers (the main thread!)

2) Underextrsusion everywhere (the main problem?)

3) Overextruded edges

@nicolinux - regarding #3, your orange "people" photo just above this post looks like #3 is fixed with faster acceleration. The edges look better than the cube you printed earlier where the edges were overextruded. You still clearly have underextrsuion - on the top layer if nowhere else. I don't know what the cause is but another solution might be to just increase the flow by about 25%

Do you have the latest firmware installed? I don't. Maybe that's why I don't have a problem? The latest firmware changed PID stuff for both nozzle and bed.

 

I didn't even see the overextrusion on the outer edges... Seems that it doesn't bother me that much

I don't like increasing flow. Tried it already and it doesn't help with this issue.

I do have the latest stable firmware installed (14.09.0). Also tried older firmwares. But I have this issue since day 1 (UM2 is a few months old by now). At first I didn't care much about it because I was so glad that overall underextrusion was gone. But now that everything else is perfect, I start to look closer.

 

[Nicolinux 288]

now i've successsed in uploading the photo

 

To me it looks like you have "regular" underextrusion. The lines on the top surface are not evenly spaced apart and some of them are broken (or look like a dotted line). Printing hotter is just a workaround but I'd check first the usual suspects for underextrusion:

 

• Nozzle partially clogged (atomic method)

• Filament hard to pull by the feeder (loosen filament, use a filament guide, unspool it a bit)

• Bad filament with uneven diameter that gets stuck in the bowden tube (use better filament)

• Feeder gunked up with bits of filament (build a filament dust filter)

• (Old) UM2 feeder loosing bits of its housing that are transported into the nozzle and block it (print Robert's feeder)

• Deformed teflon piece (disassamble hot end and take a look, maybe drill out to 3mm or order a replacement)

• Nozzle tip deformed maybe due to excessive force applied to it - bed crashed into it or delivery fail (inspect the nozzle and make sure the tip is even and round)

• Unknown issue (had this two times - contact support if everything else fails)

I'd also print the cylinder test by Illuminarti (or did you already?).

 

[Nicolinux 288]

Just an issue that does not let me sleep

It seems to me really peculiar to use at 240 C a material that is rated 190-210 C, even for workarounds. Actually I understand that 220-230 C are "normal" in UM world

It also seems that UMs, in general, report higher temperatures than other printers. Could be more accurate readings, but that's the point, it means UM prints at considerable higher temperature.

Why is that?

I normally print ColofFabb PLA at 200 C for 40 mm/s, 0.4 mm layer width and 0.2 mm layer height (3.2 cmm/s) with no underextrusion. Hotend is either E3D v6 or a J-Head version.

 

Same here. I don't like printing too hot because of stringing and other issues.

I've built a 3DR Delta printer and I am just fine tuning it. But I can print already way colder than with the UM2 (using a E3D hot end).

 

[Nicolinux 288]

Hi guys,

Finally i tested this issue on my printer. I probably have to do another test to compare the same cube.

I download the calibration_cube.gcode linked several pages ago and i modeled a test cube but it's smaller

My test cube was sliced with Cure 14.09:

0.2mm layers

50mm/s

24% infill

0.8mm shell thickness

Both cubes are printed with the same material (Colorfabb Traffic Red) at 220°c.

I have the latest firmware upgrade (Cura 14.11_RC7)

On the left my cube on the the right the gcode downloaded.

 

 

My test cube looks pretty good to me (not perfect i agree).

 

On the gcode cube we clearly see layers not touching.

 

I will do the test again with the same size cube as the gcode to really compare but i'm really wondering if this is not more of a software issue than hardware?

 

I'd say you have the same issue. And it is interesting to see that (whe you rotate the cubes by 180°), the pattern is the same. It gets worse on the top right (where the hot end leaves the object when the print is finished).

 

[DidierKlein 729]

Yes true Nico, however the issue is less visible on the cube is sliced myself. On that one the printer start printing the biggest diagonal in the middle and goes to fill until one corner, that first part is close to perfect.

The it goes back to the middle and fill the rest of the top part and is worse indeed, not horrible (from my point of view).

The biggest one (from the gcode) started good 5 or 6 layers that are close then it started separating slowly, the second half was ugly.

 

[gr5 2,178]

I download the calibration_cube.gcode linked several pages ago a

 

Oh didier - that was a bad choice. That cube was sliced with .5mm shell meaning *all* infill is .5mm wide lines meaning you are printing equivalent of 125% flow which is difficult. It takes quite a bit of extra pressure to squeeze that plastic out the thin gap *next* to the nozzle. This particular issue is complicating any other underextrusion issues. Which is unfortunate because I'm still not convinced the top layer is the only layer that is underextruded. Although I'm leaning towards "yes, it's just the top layer".

 

[3dmaker4u 0]

1. I really appreciate the diagonal infill, but when it comes to debugging it would be much, much better to have the axes movement separated (i.e. 0 infill angle). It's easier to judge in one axis than in a square of two I suggested this once to Nicolinux, could we have the tests performed the same here.

2. Extruding 0.5 mm trace width with a 0.4 mm nozzle at 0.2 mm layer height and 50 mm/s should not be such a big deal (for a printer that easily runs at 6-8 mm/s at nozzle width. Is the skipping philosophy implemented in UM really useful? I mean, does it pay all the issues with underextrusion for the very rare cases when the hobbed barrel may grind the filament? If we are looking back to the topics that were most debated and actually not completely solved, they all relate to the "ability" of the E motor to skip if it "seize" a... danger

 

[gornnr 2]

This may or may not be helpful.

I've had this problem. I suspect my bowden tube has too small of ID for the filaments I've been using. While watching, if this problem was visible I could "assist" the feeder by providing slight upward pressure on the filament. This would eliminate it in my case.

So, one step I started out with was to print a spool holder that could sit on any flat surface. I typically sit it just to the left (facing the front) of the machine as to allow the filament to travel in the direction of the coiled tension, see photo. Additionally, printing furthest away in the front right corner of the printer helps.

I run initial layer at 120% to get a glass smooth solid base with no ridges. From there I may vary the flow from 100 to 110% depending on the filament. If it seems there is an issue getting proper extrusion Ill assist the feeder and up the flow as required.

Black is difficult to see clearly but this is actually pushing into lines (the flow should probably be turned back down).

Have you hand fed filament through your bowden? Move the head to each of the four corners and give it a try. Then turn the filament at 90 and 180 degrees and try. The change in resistance is extreme in my case.

 

[gr5 2,178]

Extruding 0.5 mm trace width with a 0.4 mm nozzle at 0.2 mm layer height and 50 mm/s should not be such a big deal

 

I'm not sure if it's a big deal or not but it might be. I mean it has to squirt .1mm worth of filament sideways between the previous/lower layer and the shoulder of the nozzle. It's a tight fight - and it has to do a 90 degree turn. It *might* be a big deal. To find out one could print a cube with no shell and just infill and fine the limit where the extruder starts skipping back e.g. 60mm/sec at 210C is typical for me and then do the same experiment with .5mm nozzle setting in Cura and see what the max speed is before skipping back at the same temperature.

 

[DidierKlein 729]

Ok silly me i took the gcode that was proven to show the problem and it did as expect but that's more of a slicing problem.

Anyway i have to test a "correct" gcode that was posted here in order to see if i can reproduce it. I found an other one called w1z2.gcode a couple of pages before. I'll try that one.

Meanwhile here's the picture comparing the calibration test.gcode with a similar cube sliced by me

 

I also checked the gcodes with http://gcode.ws/ showing interesting stuff.

 

I compared calibration_test.gcode (orange) with w1z2.gcode (black) and my own sliced gcode (pink).

 

 

There's an obvious difference between orange and the other ones (as expected), the difference between black and pink doesn't show as much but to me the "gap" in black is slightly bigger than the "gap" in pink. I'm not sure if this is relevant or not but it's an interesting point I think.

 

The profile differences are the following:

 

My test cube:

 

layer_height = 0.2

wall_thickness = 0.8

retraction_enable = True

solid_layer_thickness = 0.8

fill_density = 15

nozzle_size = 0.4

print_speed = 50

bottom_thickness = 0.3

layer0_width_factor = 100

object_sink = 0.0

overlap_dual = 0.15

travel_speed = 150.0

bottom_layer_speed = 20

infill_speed = 0.0

inset0_speed = 0.0

insetx_speed = 0.0

 

w2 cube:

 

layer_height = .2

wall_thickness = .8

retraction_enable = False

solid_layer_thickness = 1.2

fill_density = 20

nozzle_size = 0.4

print_speed = 25

bottom_thickness = .2

layer0_width_factor = 100

object_sink = 0.0

overlap_dual = 0.15

travel_speed = 150.0

bottom_layer_speed = 25

infill_speed = 0.0

 

[mevander 0]

I also checked the gcodes with http://gcode.ws/ showing interesting stuff.

 

Ooh, that was an interesting site. I can't claim to understand gcode very well but if I slice a cube and leave all settings except change the layer height from 0.1 mm to 0.2 mm, the site goes from showing me a solid rectangle to a rectangle with separated lines. Should it look like that?

 

[mevander 0]

It seems that once you go over 0.15 mm layer height the lines are shown as separated in the analyzer. I'm guessing this might be on purpose to take into account that more material will "flatten" itself more?

 

[gr5 2,178]

That sounds strange. What slicer are you using?

 

[gr5 2,178]

@didier - I also sliced a tiny cube (5mm on a side) and tried gcode.ws.

I checked the box "emulate extrusion width" but there is a problem - I had sliced for um2 which uses "ultigcode" which means E (extruder) values are in cubic mm instead of linear mm. So the lines were quite fat. So I sliced for UMO and checked the "emulate extrusion width" and it came out closer but showed gaps. So I loaded the traces into excel and checked and the spacing between diagonal lines is properly .4mm and the extrusion amount is perfect for .2mm layer height and .4mm width. Within .3%.

So I don't think gcode.ws is useful for utligcode and for regular (UMO) gcode I don't understand why it shows small gaps between the traces - probably just to make it easier to visualize what is going on.

 

[gr5 2,178]

By the way I tested Cura 14.09.

 

[mevander 0]

That sounds strange. What slicer are you using?

 

Cura 14.09 in OS X. Should I post pictures of the results?

 

[gr5 2,178]

Should I post pictures of the results?

 

No. I tried the same thing and saw the same gap in gcode.ws.

gcode.ws doesn't do a good job showing you the trace widths. Unfortunately. See my post above #345.

 

[Nicolinux 288]

I got sidetracked a bit lately because I am building a new 3D printer (3DR Delta). The new printer - although barely tuned - does not have problems with tight surfaces. I printed this vase with the same settings (but different slicers) on the 3DR and the UM2. The 3DR one is watertight, the UM2 is not. Shell thickness is 0.4 (I used the spiralize option)!

 

[opit78 0]

I gonna need more test but here's what's happened.

Our older printer (August 2014) failed to print anything yesterday morning. Extruder motor was spinning, no skipping or anything but it wouldn't push any material through, not enough of filament anyway (material flow set to 130% all the time lately).

I've printed couple test with very nasty results

 

Then I took whole feeder apart and that's what I have found !

 

I know, friction, because of that the filament were slipping off the bearing and it wasn't pushed any further by the motor.

Unfortunately this part is not available to purchase at Ultimaker's website (why?).

 

Thanks God we do have second UM2 !!

I've printed out iRobert's V6 https://www.youmagine.com/designs/alternative-um2-feeder-version-two and https://www.youmagine.com/designs/um2-iroberti-s-feeder-v6-yoke-and-hub-with-guide part as well

After quick setting up the tension I was able to print with new feeder straight away, no issues so far.

Result - 100% covered top layers from edge to edge, no gaps, no nothing

Box 30x30x5

0.2mm layer hight

220*C hot end

55*C bed temp

50mm/s print speed

15% infill

6x top/bottom layers

material flow 100%

 

[Nicolinux 288]

Wow, nice find. I can't believe it is that easy because someone else with this issue would have used Robert's feeder (which is pretty popular). But I'll try it too. Wanted to change the feeder anyway.