Top Layers not touching - UM2

[chrisr 18]

Been thinking a lot and experimenting with the feeder design recently.

Imho you really want just a metal eyelet in contact with the filament where the filament enters the feeder body and nothing else. Preferably no more than 4mm internal diameter to help guide the filament and prevent any chance of it "wandering"

This (metal) eyelet needs to be as close as can be to the hobbed gear and the pressure wheel. As that (in the experiments I've been doing) obviates any need for any additional guides which imho just add extra friction and wear. PLA/ABS/XT on PLA/ABS/XT surface to surface is a recipe for wear.

I'm trialling my own version of Roberts great design, as there's several other features of that design I'm not happy with

 

[opit78 0]

I've ordered these of eBay yesterday and I gonna apply them to iRobert's feeder as soon as I get them.

I'll also add some "sleeve" for bearing with groove/guide for filament

 

[chrisr 18]

The same supplier I got mine from :wink: They must be wondering why there's a rush on their 4mm eyelets :smile:

 

[Nicolinux 288]

So I switched to Robert's feeder but the issue still remains:

Stock feeder:

Robert's feeder:

 

[gornnr 2]

One thing that has just crossed my mind was that my quality always looks great at the start of the print. First layer always has great fill then youll get these sections where it seems that it is underextruding.

I'm wondering if either the Feeder stepper or the Z-axis steppers are possibly getting a bit too hot and missing a step here or there. In the feeder, the heat could cause too little material to be fed or possibly heat propagation down the shaft and softening the filament. In the Z-axis, it could be taking a slightly larger step which causes larger actual layer height.

 

[dirkdirk 5]

Today I had exactly the same problem with a print. It does not always occur but gets worse with increasing layer. So I played around a little.

Result -> problem are the side fans . If the fans are slowed down or disabled the problem disappears. Because it does not occur at all prints so it can only be connected to the different air flow of the part geometry.

It would be nice If someone could confirm this.

[Nicolinux 288]

Hm, I don't know... Yesterday I printed something for a client (3D Hubs) with XT and fans off and this issue was very pronounced. Maybe it is due to the object's geometry and the way Cura slices it in a way that triggers a wrong interpretation of gcode within the UM2 firmware.

 

[dirkdirk 5]

I was able to reproduce the error. So in my case with normal pla the fan is the problem.

I have printed at 100 % fan -> surface was not closed. Fan to 0% -> the surface is completely closed. I tested it on several layers.

[Nicolinux 288]

Could you share the model and the .gcode file?

 

[dirkdirk 5]

Could you share the model and the .gcode file?

 

Sorry, i can't share this specific model. Just print a rectangle and turn the fan off. I can upload one if you like..(but I guess that is created faster than uploaded).

something like this:

Left side fan 100% right side fan 0%, all other settings identical.

 

here i have dropped some paint on the print. you can see that the surface isn't closed at all.

 

on the right side you can just brush away the paint. left side not.

 

[sputnik 0]

i stumbled upon this thread while researching my own underextrusion problems;

my prints looked pretty much the same as yours, perfect in some places but then random places of underextrusion.

larger prints usually failed randomly after 6-10 hours of printing.

the extrusion test piece however always printed flawlessly.

 

in my case the problem was the knurled bolt on the feeder.

sometimes - depending on where the filament spool sits on the spool holder - the filament was going into the feeder at a slight angle (not exactly vertical but tilted towards the front of the printer); then the filament was only partly riding on the knurled part of the knurled bolt which caused random slippage and underextrusion.

the solution was to unmount the feeder and slide the knurled bolt inwards, closer to the stepper motor, for about 2mm, so that it is no longer flush with the feeder stepper motor shaft. perhaps this shaft is too long on my machine.

 

printed the same .gcodes again, no more problems, printer now working flawlessly for 50+ hours

 

[3dnerd 6]

Hi,

I think I have the same problem, but did not knew about this thread.

I opened a new one:

http://umforum.ultimaker.com/index.php?/topic/10299-top-surface-not-closed/

and dirkdirk (thanks) pointed me to this thread. I read now all 19 pages - phew!

I want to try also with the fan, but I don't know hot to switch off the fan only for the top layers.

Is there any setting in Cura?

And is this problem really solved now?

Is Ultimaker working behind the scenes for a solution?

Maybe Cura update that switches off the fan for the top layers automatically?

What about the 'solution' with the front left head skrew?

 

[Nicolinux 288]

Sadly there is no generic solution. For me this problem comes and goes. I haven't been able to determine the cause. Today I printed a flat object with woodFill and the problem was quite visible. A few days ago I printed an (albeit smaller) object and the surface was perfect.

You can use the "TweakAtZ" plugin in order to change the fan speed at a certain height. Switch to the expert view in Cura and access the plugin in the "Plugins" tab (top left).

And another remark: Now that I re-read sputnik's answer I can at least say that the feeder plays a role here. A week ago I changed Robert's feeder to an updated design. After that I had the problem where the filament would slip off the ball bearing (towards the back of the printer). Not only that but I have observed a degradation in print quality.

I exchanged the "yoke" part with another design that keeps the filament in place but the print quality didn't change and I _think_ I saw the "top layers not touching" problem appear more often since then.

So maybe that's the cause. Feeder doesn't move the filament correctly leading to some kind of underextrusion.

A few posts ago gr5 speculated that it is a general underextrusion problem but I stated that the cylinder test prints fine (which otherwise would indicate an underextrusion problem).

 

[illuminarti 18]

There are two types of under-extrusion:

1) The extruder motor stalls and clicks back, causing a momentary interruption in material flow. This is what the extrusion test cylinder tests for;

2) The extruder consistently pushes out less material than is being asked for; until it gets incredibly bad, this doesn't cause any obvious defects in a single wall test like the the extrusion cylinder; the wall gets thinner, but remains intact. The cause of this seems to be related to the filament slipping in the feeder assembly when there is back-pressure resisting the movement of the filament. The knurled sleeve teeth marks get closer together, suggesting that the filament is slipping in between forward movements, rather than the motor not turning as fast as it should. This is what causes the gaps in top surfaces, and was also what I documentedhttp://www.extrudable.me/2013/04/18/exploring-extrusion-variability-and-limits/.

Here's a simple test you can do;

1) Remove the filament, then remove the bowden tube from the top of the feeder.

2) Push some clean PLA filament into the feeder, until it emerges a few cm above the top of the feeder

3) Measure 30cm below the bottom of the feeder, and mark the filament

4) Use pronterface to advance the filament 250mm at 1mm/s (this corresponds to a moderately fast print at a bit over 6 cubic mm/s)

5) Measure the distance between the bottom of the feeder and the mark made in step 3, to calculate how far the filament moved when asked to move 250mm.

6) Repeat steps 3, 4 and 5 at least twice

7) Since you are pushing filament into the air, with no resistance, you should find that the filament has moved about 250mm, as expected. Maybe even a bit further, if your experience is anything like what I've seen.

8) Remove and cut off the filament that has been through the feeder once already

9) Reinstall the bowden tube, and heat up the nozzle to 230º; insert the filament, and feed it through until it starts to extrude from the nozzle.

10) Repeat steps 3,4, and 5 three times and see how far the filament moves now that it is encountering significant resistance. Note that you still should not be experiencing extruder click backs, if your printer is set up ok.

11) Take out the filament, remove the bowden tube from the feeder, and repeat the resistance-free test one more time.

In my experience, on several printers that exhibited signs of under-extruded top surfaces, the distance moved with no resistance was at or above the 250mm that was expected.

However, when resistance was added by having to actually extrude through the nozzle, the distance moved immediately fell to about 200mm, and in each subsequent test got worse and worse, down to around 175mm, iirc. That's about one-third less extrusion than expected, which would totally explain the gaps between the lines in the top surfaces. Furthermore, trying to increase the flow rate just increases the volume per second, and hence the resistance pressure, which is why increasing the flow rate isn't sufficient to solve the problem usually.

I don't know why the results got worse each time I repeated the test, but as soon as I removed the Bowden tube and hence the resistance, the extrusion distance returned to normal. So it wasn't some sort of degradation in the motor or electronics - or at least not one that was visible when there was no back pressure to overcome.

So, what does this tell us? The problem isn't related to gcode, or anything like that; it's simply that - on some printers, some of the time - back-pressure-related under-extrusion gets so severe that it leaves visible artifacts in the prints.

As to why, I have no idea.

 

[Nicolinux 288]

I don't know why the results got worse each time I repeated the test, but as soon as I removed the Bowden tube and hence the resistance, the extrusion distance returned to normal. So it wasn't some sort of degradation in the motor or electronics - or at least not one that was visible when there was no back pressure to overcome.

 

As usual, thanks for the great insight Simon.

Wait a second. What if the stupid bowden tube is too narrow? Maybe manufacturing tollerance, bad batch - who knows?

 

[illuminarti 18]

As usual, thanks for the great insight Simon.

Wait a second. What if the stupid bowden tube is too narrow? Maybe manufacturing tollerance, bad batch - who knows?

 

Yes, in at least one case, replacing the Bowden Tube made a significant improvement to the throughput, although it didn't totally fix the problem.

 

[gornnr 2]

I agree with illuminarti, I've had a feeling for some time that bowden/nozzle resistance has been an issue. Here is the path & info of what has been done to cure this.

Resistance Failure Modes:

-Bowden to Filament difference too small

-Temperature too low for extrusion rate

-Bowden Curve to Filament Curve difference

-Filament dragging on deformed insulator

-Filament catching on hardened filament pulled up/created during retraction

What I've tried to eliminate resistance:

-Changing where filament spool is located/feeds from

-Doing EXTREME Atomic pull cleaning (become a pro at this, its a serious must)

-Various Bowdens, Larger ID (Slightly larger and MUCH larger than filament), Different material, etc.

-Changing from 3mm to 1.75mm filament

-Moving the Feeder, Spool, and Bowden routing

Here is the current configuration

Conclusion:

Provided the nozzle is VERY clean and temperature is set correctly for the material and extrusion rate then the best improvement was moving from 3mm to 1.75 filament. Changing the filament path may help 3mm filaments more but at this time the setup above has only been used for 1.75mm filaments.

Positive

1.75mm flexes more than 3mm

-Less prone to breaking/snapping at higher curves

-Follow bowden curve easier & handles changing curve of bowden during printing

-Contacts nozzle extruder at end of extrude, molten for less time, lower plastic breakdown

-Faster Material Move, again decreasing time material is molten

-Cheaper & more selection than 3mm in the US

Negative

-Requires Bowden capable of 2mm ID with 6mm OD (none premade, NOTHING bonds/glues PTFE - Teflon)

-Requires feeder changes (follower to Drive wheel spacing)

So, with this current setup if there is any layers that are not fully uniform its now ONLY a function of extrusion rate (temp vs speed x layer height)

TL;DR - Remove feed resistance then its only a function of setting the right temp for the extrusion rate for a given material.

 

[3dnerd 6]

Hi guys,

thanks so far. I think I will use TweakAtZ first to turn off the fan at the top layers and see what happens. Then I will raise the temperature from 230° to 235° and may slow down print speed.

I wonder if those problems will be solved by Ultimaker team at least for the UM3? :wink:

(And hopefully we will get a better feeder in UM3)

 

[esteban-pacheco 33]

Hello Everyone, im glad i've found this tread today, i thought i was one of the few having this headache.

I have somethimes this Kind of Problem and ist very annoying, sometimes the top layers Closes really smoth, sometimes dont. The object's Shells sometimes fuse togehter (if im printing lets say and regular shape, for example an square) but if im printing something "organic" where the outlines are not straight, morelike "curvy" then the outlines doesnt fuse evenly, in some parts yes, and in somere Areas dont. I've been testing with my ultimaker in the last couples of weeks, to try to find the answer to this "erratic issue" but it is imposible to find (at least for me) a solution, because of the same ist erratic nature.

i tried EVERYTHING--- i even got a new bowdentube, new PTFE coupler, nut hotend issulator, oiled the axis, my shortbelts are really thight, im printing PLA (colorfabb) at 215° and 40 mm/s... im doing i think all properly, and still having issues,

i think it can be anything causing this...

HAVE SOMEONE NARROWED THE SEARCH FOR CAUSES A LITTLE BIT?

[esteban-pacheco 33]

and i did all the available print test in order to discard blackslash Problems, or underExtrusion, stringing, etc etc etc...

[gornnr 2]

Rule #1 of underextrusion, Clean your Nozzle.

Rule #2 of underextrusion, Slow down your speed (or turn up the temp)

Just because plastic comes out of the tip doesn't mean you are getting 0.4mm of plastic out of the tip.

[esteban-pacheco 33]

Shhhhhh Still no luck with this issue, it seems the effect is worse using White PLA Colors

[Nicolinux 288]

I have a lil update on this too. I am using the Olsson Block now and I've experienced the problem again. There was the idea about the temp. sensor not working properly. I swapped it but the problem persists.

[esteban-pacheco 33]

i think is a software problem... i get better results in top layers using the new cura (15.06.03), because you can specify the line width of the top and bottom layers..

BUT, i have noticed, theres a problem with combing, and that makes the external surfaces (shells) a little bit strange and crappy

Edited August 3, 2015 by Guest

[Nicolinux 288]

The question is, why do only a few people experience this problem if it is a software issue?