Almost always missing layers / underextruding

[Nicolinux 288]

Hi,

this is bugging me to no end. The UM2 prints fine for hours and then suddenly decides to heavily underextrude. It is very annoying when the print is several hours in and nearly finished and then it is messed up (visually). I don't trust it right any more but I also can't babysit it the whole day.

Here is an example:

I followed Simon's advice and placed the filament spool on the floor and unrolled a bit of filament. Could this be the problem? When the extruder has to pull on the filament and it won't move fast enough? I am not sure, because I didn't see the filament tangled up (or I did see it but untangled it in time before it became problematic)

 

Could this be a partial plug?

Thanks,

Stefan

 

[maurice 1]

To put the filament on the floor doesn't look wise to me. Because the filament doesn't 'unroll', more and more torque is created inside the wire (on long prints), and that can influence the feeding. Mind you, I don't say that is the cause of your problem, but it could be. I've built a separate roll-holder for the larger (3 kg) rolls I use.

 

[gr5 2,167]

Illuminarti puts his filament on a lazy susan. I tried one from rubbermaid - it has ball bearings - very smooth:

http://www.hardwarestore.com/lazy-susan-606736.aspx?utm_source=pla&utm_medium=cpc&utm_term=&utm_campaign=rkg&device=c&network=g&matchtype=

The lazy susan rotates as the filament feeds in.

 

[braddock 1]

another option, one I've used today, is I actually cut a length of PLA based on the indicated length to be used on the U2 display. It hangs freely with no restriction. Of course, this is not ideal, and means you have to reload for the next print, but I'm just experimenting.

 

[gadgetfreak 7]

I agree this is really annoying.

If this is would be due to friction from the filament unrolling, has anyone tried to tighten the feeder mechanism a bit?

Perhaps replace the filament guide with some type of rolling bearing to decrease that friction (although I believe that friction to be negligible I can't back that up with test data).

Also, for longer prints it would be useful to record a video stream of both the print and the roll/feeder side, so that exactly when the underextrusion occurs it could be analyzed whether the roll/feeder seem to be the culprit or not...

 

[Nicolinux 288]

@gr5: Funny name for a rotary table Oh, I have a spare ball bearing from the UM1, maybe I could print one.

@braddock: I'd rather not cut the filament. What if the print fails? My prints aren't too long right now so I'd have relatively short pieces of filament lying around.

@gadgetfreak: I am already using OctoPrint just for the webcam streaming feature. But the resolution isn't good enough to spot underextrusion. But maybe I can spare my iPhone for some hours and use it in combination with the webam to watch both pieces.

However something tells me it isn't this problem. If you look at the first and second picture, the underextrusion (or whatever it is) was very pronounced on the both arms. Why not the center area? Is there a weird relation to extrusion and x/y movement close to the outer edges?

 

[gadgetfreak 7]

That's funny with the arms.

Have you tried another color? For red filament I usually need 10-15 degrees higher than for a specific green for example, for it not to underextrude at a certain speed/layer height.

My feeling is that it's not temp-related though.

If you print the exact same model, to you get underextrusion at the same location in the model? If so, it's most likely related to the slicer. If not, it's probably mechanical....

 

[Nicolinux 288]

Hm, ok. This was my longest print yet (5 hours). I don't want to repeat it just for the heck of wasting filament, but since the first model didn't come out right, I might do it anyway.

I did print another one before, only flatter hence a much shorter print. The underextrusion was there only at the top and only on the center area (and very much less pronounced).

Sadly I don't have the gcode file anymore :/ But I think I can reconstruct the print settings. I will also make sure to load the new gcode file in Repetier Host. Their print preview ist pretty accurate.

If the problem is mechanical I'll will try to reproduce the underextrusion at the same height. If that's the problem, then it is definitely the z-screw.

Another question - how do you guys clean the nozzle on the UM2? I had the luck to not ever experience a plug on the UM1 (but then again, I didn't have it for a long time).

 

[chrisp 10]

Another question - how do you guys clean the nozzle on the UM2? I had the luck to not ever experience a plug on the UM1 (but then again, I didn't have it for a long time).

I have another post under Clogged UM2 (http://umforum.ultimaker.com/index.php?/topic/3749-clogged-um2/&do=findComment&comment=31094) on how to clean the nozzle. I have done it 3 times in the past two weeks.

I also had this problem with under extrusion and I think it might have been due to the feeding mechanism/extruder. I did adjust the extruder to have more tension because it seemed to eat up my plastic too easily.

It's hard to see it in this pic, but I had under extrusion when printing this extruder every 20-30 layers:

This print took 23 hours with 100% infill (3 parts)

Once I had adjusted the extruder, I printed this motor gear which was almost perfect:

This print took 6.5 hours with 100% infill

 

[gr5 2,167]

Sadly I don't have the gcode file anymore :/ But I think I can reconstruct the print settings. I will also make sure to load the new gcode file in Repetier Host. Their print preview ist pretty accurate.

 

I always save my gcode files and also the "profile" in a big "ultimaker" folder so I can always go back later and see what I did. Also I have a notebook where I write down the date, time, what I printed, and any settings not in the profile such as temperature. That way I can go back later. I look over this notebook much more often then one might think (e.g. last night I wanted to remember what temp I like to print nylon at).

 

[illuminarti 18]

On the under extrusion on the arms... I'm wondering if that's related to combing...

If combing is enabled, then when moving within the print, the head oozes, because it doesn't retract. So then when you start printing, the perimeters you might get under-extrusion. That might be some of what you see on the arms, if its done a long move internally before starting the perimeters.

 

[illuminarti 18]

On my blog post from last year about http://www.extrudable.me/2013/04/18/exploring-extrusion-variability-and-limits/, Erik vd Zalm has been commenting recently that he's been testing a a filament movement feedback sensor that can measure the amount of filament moved, compared to the requested amount... obviously this opens up a lot of possibilities for dynamic adjustment of extruder feed rates to ensure that the actual extrusion meets the intended amount, and to compensate for problems.

More relevantly to this issue however, he did mention that he noticed that the feed rate on his printer could be noticeably affected by the force required to unspool the filament. So I do think it's a definite Thing.

 

[gr5 2,167]

filament movement feedback sensor

 

Very cool. And Marlin should slow down the overall feedrate if it can't achieve desired filament movement such that it just stops printing if it can't feed at all and automatically slows down if you requested too cold of a filament temperature for the current print speed.

 

[braddock 1]

Unspooling force seemed to be affecting one of my prints last night.

 

On my blog post from last year about UM1 extrusion rates, Erik vd Zalm has been commenting recently that he's been testing a a filament movement feedback sensor that can measure the amount of filament moved, compared to the requested amount... obviously this opens up a lot of possibilities for dynamic adjustment of extruder feed rates to ensure that the actual extrusion meets the intended amount, and to compensate for problems.

More relevantly to this issue however, he did mention that he noticed that the feed rate on his printer could be noticeably affected by the force required to unspool the filament. So I do think it's a definite Thing.

 

[Nicolinux 288]

@ChrispCreator: How did you adjust your extruder? Do you have a photo?

@gr5: Yes I normally do save .gcode files but I was hacking at Cura to implement some kind of autoupload to my wiki after slicing and this disturbed my normal workflow

@illuminarti: Combing was enabled but even then - why at the top, why not earlier?

@braddock: Look what I found at IKEA, Lazy Ingrid

 

[braddock 1]

Cool! I found last night though, that the way the PLA was spooled on the reel was sort of crossed over itself, so even that would cause additional load.

 

@ChrispCreator: How did you adjust your extruder? Do you have a photo?

@gr5: Yes I normally do save .gcode files but I was hacking at Cura to implement some kind of autoupload to my wiki after slicing and this disturbed my normal workflow

@illuminarti: Combing was enabled but even then - why at the top, why not earlier?

@braddock: Look what I found at IKEA, Lazy Ingrid

 

[IRobertI 520]

@illuminarti: Combing was enabled but even then - why at the top, why not earlier?

 

It looks like the layers where it happened more frequently were smaller in area. Since the perimeters are laid down from inside to outside it could be that on the larger layers the inner perimeters are longer and thus the head has filled up with plastic again by the time it gets to the final outer perimeter. On the smaller layers you have a much shorter amount of perimeters to essentially re-prime the nozzle.

Just rambling thoughts, I have no data to back that up.

 

[Nicolinux 288]

Ok, could be. I have printed a few things since then and this problem didn't re-appear. But these were short prints (under 5 hours). I'll do a +10 hr print soon - let's see how this will work out.

 

[aviphysics 9]

I have been wondering if it could partially be a heat sink issue. Basically, when doing a solid fill at the top, there is more material surrounding the head to carry heat away then when doing the edges.

Sort of related, I have noticed that when the print head fills a large square area along the diagonal, it will get good fill when it is going rapidly back and forth near the corners, but as it approaches the middle, I have to use a higher temperature to get the same results. I think what is going on in this case is that when it is going and back and forth quickly, the adjacent material doesn't have time to fully cool off, but near the center, the surrounding material has more time to cool between passes and therefore takes away more heat from the hot end.

Does that make sense?

 

[illuminarti 18]

I think the issue is that at the corners the head is moving slower on average, because of the acceleration and deceleration involved. So the extruder isn't working as hard.

The heat energy loss from the head due to the extruded filament isn't really all that great; it's a small amount of plastic compared to a lot of metal - especially on the UM1.

[Nicolinux 288]

I am starting to believe it is a filament issue. Just printed the Iris Box (http://www.thingiverse.com/thing:8402) where the base is about 65mm tall. No sign of underextrusion. Just a perfectly smooth surface. This time I printed with pearl white filament.

I do intend to print a large object with my red filament. We will see how that goes.

 

[aviphysics 9]

You don't think the temperature of the plastic the filament is landing on matters?

 

I think the issue is that at the corners the head is moving slower on average, because of the acceleration and deceleration involved. So the extruder isn't working as hard.

The heat energy loss from the head due to the extruded filament isn't really all that great; it's a small amount of plastic compared to a lot of metal - especially on the UM1.

 

[IRobertI 520]

@aviphysics: Regarding your other post about the extrusion being thinner on the longer extrusions compared to the quick movements near a corner. That's more likely to be a case of the head not having time to get up to full speed while doing very short zig-zags. On the longer stretches it has time to fully accelerate to top speed and thus the volume of extruded plastic increases. On the short ziggies it also stays over the same area longer and can radiate more heat onto the surface to make it "mushier". This is especially problematic on sharp points.

As for the surrounding plastic sucking heat away from the nozzle. I doubt that's much of an issue. If anything what heat remains in the plastic should radiate back to the nozzle and apply a cooling effect that is far smaller than when the head is out in the open over the edges where the fans can have a go at it. Also the head should never really in direct contact with old layers since it is a layer height higher and has plastic coming out of it acting as a barrier.

 

[illuminarti 18]

You don't think the temperature of the plastic the filament is landing on matters?

 

It matters a little bit for the quality of overhangs, or thin vertical parts - if you're printing quickly, and the lower layers don't have time to cool, then you can get poor quality vertical/overhanging surfaces because the plastic stays above it's glass temperature for too long, and flows out of the desired shape.

But I don't think it makes much difference in terms of laying down one line of plastic to the next within a layer, no, which was what the original comment referred to, and the fact that the heater is more than capable of providing sufficient heat to keep the head hot in normal circumstances. A different challenge is getting the extruded plastic itself up to temperature in the first place, when printing fast, because the plastic spends very little time in the hot zone before being extruded, and so doesn't get up to thermal equilibrium with the rest of the hot end.

 

[aviphysics 9]

@aviphysics: Regarding your other post about the extrusion being thinner on the longer extrusions compared to the quick movements near a corner. That's more likely to be a case of the head not having time to get up to full speed while doing very short zig-zags. On the longer stretches it has time to fully accelerate to top speed and thus the volume of extruded plastic increases. On the short ziggies it also stays over the same area longer and can radiate more heat onto the surface to make it "mushier". This is especially problematic on sharp points.

As for the surrounding plastic sucking heat away from the nozzle. I doubt that's much of an issue. If anything what heat remains in the plastic should radiate back to the nozzle and apply a cooling effect that is far smaller than when the head is out in the open over the edges where the fans can have a go at it. Also the head should never really in direct contact with old layers since it is a layer height higher and has plastic coming out of it acting as a barrier.

 

To clarify, I am not really talking about the head dropping in temperature, but the plastic extruded from the head cooling more quickly once it comes out, so it doesn't spread as easily.