Almost always missing layers / underextruding

[Nicolinux 288]

Nicolinux, did you ever use any other type of filament than PLA in you UM2?

Also, were you ever able to reproduce the underextrusion in the exact same spots of the print consistently?

We had a lot of sporadic underextrusion after using the beautiful XT. We had run an all-nighter print and filled the fan shroud/hotend with XT because the object loosened from the platform (we didn't use glue that time) and just squirted glue for 10+ hours.

XT is another type of base plastic and to make a "perfect" transition (purge/cleaning) from say ABS to PLA or XT to PLA might be difficult. (I will say that we didn't try to put a lot of time to clean it as we didn't have the time to make a complete disassembly)

After replacing the nozzle to a completely new one it was really smooth again.

That said, we are still getting occasional underextrusion on long prints but that might be also due to friction from the filament guide etc.

 

Never printed with something other than Ultimaker PLA. But I noticed that there are differences. The silver one looks a bit grainy when it is hot - compared to other colors like orange or white.

But since I'll take the head apart anyway, I will look for a partial plug.

 

[chrisp 10]

Nicolinux,

 

Sounds good. Would be an easy fix if one could actually access the teethed gear without taking the extruder apart

I have taken apart the extruder about 5 times and I can tell you that it would definitely be worth checking it to see if there is play. I thought that might be an issue about 1 week ago on my machine, so I took my set of allen wrenches and removed the extruder and motor. After marking the exact position of the gear which moves the filament, I then unscrewed the set screw and removed the gear. (This is much easier than taking apart the hot-end, but it's very difficult to get everything back together because you need a third hand). You have to take off the metal cover inside that covers the motor, I think there was just (1) screw. Then take out the (4) screws on the side of the extruder, but be careful because the motor can fall off from the inside. All of the parts in the extruder are extremely easy to take out and put back together, just try not to mess with the adjustment screw, which has a nut on it. Once the motor is free from the case, you can take the allen key which came with the printer and remove the gear. When you put it back on, make sure it is extremely tight. When the bowden tube goes back on, it has to be seated tight against the inside ridge of the extruder, You have to hold the motor in place, extruder cover on, bowden tube tight against ridge, and put screws in, all at the same time. It's possible, but difficult.

If you need help, I have a teamspeak server and we can link up and chat about it.

 

[Nicolinux 288]

Hey Chrisp, thanks for the offer. I will disassemble the head first and if it doesn't help, I'll tackle the extruder next.

 

[illuminarti 18]

So I finally had to take my Bowden tube off last night to fish out a piece of filament that snapped off during the filament change.

Putting the tube back on the hot end, I noticed that it slides pretty easy through the retaining ring and into the coupler. Once it's seated in the coupler, then you need to be careful to keep it pressed down tightly while sliding the locking ring carefully up over the stationary tube, in order to attach the red clip. You can feel the teeth on the ring scraping the tube as you do this, but I think it's the only way to ensure the tube is fully seated. If not, the tube can move up with the ring,when the ring moves up to the locked position, and you end up locking the tube in place several mm above where it is supposed to seat in the white part. So that's one thing to check.

Next, I wanted to print something to see what the throughput of the extruder is. My first print was a simple 6cm diameter, single-wall cyclinder. I printed it with Colorfabb PHA/PLA at 230º, with 0.25mm layers (and a 0.4mm wide extrusion) so that every 10mm/s of speed equates to 10 x 0.25 x 0.4 = 1mm³/s second of extrusion. And I edited the gcode in 1cm bands to increase the speed from 30mm/s to 40mm/s, all the way up to 100mm/s - i.e., 3mm³/s all the way up to 10mm³/s. And it printed all of the print very, very nicely. The z-scar got more pronounced as the speed increased, and there was a touch more ringing on the lettering. But away from those parts, the layers were nice and smooth and even at all speeds, with no sign of any under-extrusion, and certainly no click-backs from the extruder.

 

So that provides a basis, and corroborates what Ian saw in his prints. When there is no retraction required, the printer can handle at least 10mm³/s second without too much problem. (Of course it depends on the filament, and the temp - but I'm encouraged that while this was a highish temp, it wasn't a 'crazy hot' temperature, and it suggests that there's still some room to go even faster).

 

Next to repeat the test with two cylinders side by side, and see if retraction plays a part. If anyone wants to try to repeat the test with their own printer and filaments, I'm attaching the single cylinder gcode.

 

[gadgetfreak 7]

Beautiful work Illuminarti!

Can you repeat with plain PLA?

[gadgetfreak 7]

I can do testing with this model in lots of different plastics on two UM2s but not likely until after the weekend...

[illuminarti 18]

Yes, I can definitely repeat this with more plastics this weekend, and I'm especially interested to see what happens when i print two side by side, with retraction added.

This shot of the back of the cylinder probably gives a better impression of just how nicely these printed:

The fans were on full, after the first few layers, as per Cura's normal ramp-up schedule. It was a fairly easy print because there were no overhangs (apart from the layer after the dividing lines, which are inset 0.2mm) but I was pleasantly surprised to see that there were no 'minimum cooling time' issues even in the top section, where it was laying down a layer every 1.9 seconds. I thought I might see a little bit of softening and buckling of the walls at that speed, but there wasn't any.

Also noticed that there is a faint vertical pattern that I think corresponds to belt tooth pitch, which is only noticeable if you really look for it, with the light catching it just right - and it's slightly more pronounced in the 90mm/s (9mm³/s) band - probably the result of some mechanical resonance.

 

[Nicolinux 288]

Compare this with my print...

Interestingly enough just before that I printed another object (without retraction) that came out just fine (under the given circumstances).

But ok ok, I know. Enough bitching, off I go to take this thing apart

 

[illuminarti 18]

Hmm. I think the correct interpretation of your test result is: "Something ain't right with your printer." :-)

But still, it's nice to be able to quantify it a bit more :-)

 

[gr5 2,172]

Illuminati - your z seam gets more pronounced as you move up to higher speeds. Any idea why? I guess with higher speeds you have higher nozzle pressures and you get more leaking during the Z movement?

Well hell that is nice 100mm/sec at .25mm layers! I'm going to try that baby! Maybe at a few different temps.

 

[illuminarti 18]

Yes, I think it's just the nozzle leaking more at higher pressures during the relatively long time it takes to move such a thick layer. I'm not sure what the speed and acceleration settings are on the UM2 Z-axis - there's probably some room for fine tuning it, as we did with the UM1; it sounds fairly slow.

 

[illuminarti 18]

I did more tests...

First, I used the same filament as before and printed two speed test columns at the same time, with retraction on the moves between them.

They printed perfectly fine:

 

I made a timelapse video of the print and put it on YouTube:

 

 

 

I also grabbed a shot of the end of the filament when I removed it; it has a slight step at the end, but seems quite small and compact - and perhaps tellingly, it didn't get pulled out into a narrow neck, as if the wider tip had gotten snagged on something on the way out:

 

 

[illuminarti 18]

Next I tried PLA - the Ultimaker Blue PLA that came with the printer, specifically. And my first attempt started to fail at 7mm³/s:

Then I noticed that the filament was really tightly coiled. I'd pulled it loose off the spool, as I usually do, but because the spool was more than half used up, the radius of curvature on the filament was pretty tight (it fell into coils which were about 14cm diameter):

 

That made it really springy, and I wondered if that was adding extra resistance in the Bowden or extruder drive, and causing problems.

 

So, I cut off a couple of meters of filament and straightened it out by hand, so that it only had a very gentle curve left on it:

 

 

This time the print went absolutely fine:

 

 

So I did some more tests. I switched back to the coiled filament and it printed much better this time, with only 1 failed extrusion line in the 10mm³ range. I took the filament back out and fed it back in again, and this time paid attention to how it fed into the extruder... I consciously arranged the filament so that fed into the extruder in such a way that the natural curvature of the filament was being fought all the way through the drive mechanism and the arc of the bowden tube. And lo and behold, once again it failed pretty much as the first PLA version had, starting at 7mm³/s, and gradually getting worse as the print got faster.

 

Finally, I switched back to the straightened filament, and it printed perfectly once again.

 

So my conclusion on this is that (when everything else is working ok - perhaps most importantly the seating of the Bowden in the teflon coupler, and the coupler itself being clean and unimpeded) then the UM2 can happily print at at least 10mm³/s in both PLA and PLA/PHA.

 

However, if the filament was tightly coiled and badly oriented so as to add resistance to the feed process, then I definitely did see underextrusion. I'm pretty sure, also, that in all the cases I saw under-extrusion it was because the extruder stepper was skipping steps and slipping back.

 

 

 

[illuminarti 18]

Finally, I'm attaching the gcode for my '2-column with retraction' version of the throughput test, in case any one wants to try it.

 

[takei-naodar 0]

what temp did you print on?

 

[illuminarti 18]

All these tests were at 230ºC.

 

[takei-naodar 0]

Hmmm from your results i would take both feeder and spool need to be repositioned to improve quality and quantity.

Maybe on the side?

 

[gadgetfreak 7]

Excellent testing and very useful information. I guess that in order to have a good extrusion, the feeding mechanism should be even stronger/better than currently in UM2. I wonder which one is considered the best on the market?

And how to measure it?

It needs to handle all kinds of filament, both hard and flexible (though the UM2 feeding mechanism does already handle rubbery filament pretty well, we've tried NinjaFlex as well as FlexPolyester 45D and 40D).

Now with these models you designed it should be possible also to establish a temperature baseline and see what temperature offset is needed for each color. At least within a manufacturers line, for example Colorfabbs PLA/PHA line.

[Nicolinux 288]

Ok, I disassembled the head. Here are my findings.

Before I started to take the head apart, I took the bowden tube out and followed this procedure to remove a potential clog (http://umforum.ultimaker.com/index.php?/topic/3561-ultimaker-2-extruder-nozzle-blocked/?p=28940). The filament tip looks alright when I took it out at 90°:

Next I started to take the head apart. It isn't very hard. I followed braddock's advice and taped the black plastic parts together so they stayed at the top. Then I unscrewed the fan holder and moved it out of the way to the back:

I had to unscrew two more screws to split the lower part of the head from the upper one. Then I took the teflon piece out. It was in perfect shape:

When I took upon the nozzle, I noticed that the exit was somhow obscured. I then ran a fine wire through it and I guess something changed because it then looked much better:

After I put everything together I ran illuminarti's test (no retraction). Did not went well until I straightened the filament and held in my hand the whole time...

 

When it came to the retraction test, it failed at the end. I got lazy and put the filament on a broom leaning on the side. When I wasn't around for two minutes the filament must have tightened to the point where it started to underextrude :/ I let it finish the print and some areas recovered well.

Conclusion

I guess it is the extruder. Somehow it is extremely sensitive to any resistance from the filament. This wasn't always the case since I have printed some objects before without any problems.

So I need to take the extruder apart. But what am I looking for? The small white switches are already at the top (I guess at maximum strength).

Maybe the extruder does not get enough current - but how do I check that? This thing isn't documented at all.

 

[Nicolinux 288]

Forgot to mention. When you turn the metal piece beneath the teflon coupler, it moves the nozzle and the teflon coupler up against the bowden tube. If the bowden tube is held firmly in place, both sides are pressed against each other. I have now 0 play with the bowden tube.

 

[woofy 12]

I guess I have a major problem then :sad:

printed with ultimaker blue at 220.

There's a hi-res here.

 

 

[chrisp 10]

Nicolinux

 

When I took upon the nozzle, I noticed that the exit was somhow obscured. I then ran a fine wire through it and I guess something changed because it then looked much better:

Did you clean out the nozzle really good, or did you just stick a wire through it? I use a straightened paper clip and twirl it around and almost always a burnt clump comes out that is stuck to the inside of the nozzle.

I also have a much faster/better way to clean out minor clogs now that only takes about 10 minutes.

Take off the red clip to the bowden tube, push down on gray fitting holding the bowden tube in place, then lift bowden tube a couple inches out of teflon fitting.

Then loosen all 4 long screws holding the hot-end together. only take the back 2 all the way out, then lift the front 2 enough to drop the aluminum with hot-end attached, like in pick. The 2 screws left will hold the carriage in place while you work on it, no tape or supports needed.

You can heat up hot-end (180 was sufficient) while holding one of the side fans so you can use the other hand to clean out nozzle.

By doing this you can skip taking the side fans apart and releasing the tension between the teflon fitting to the milled fitting. (This only works if you clog is below the teflon fitting)

 

 

 

[illuminarti 18]

Forgot to mention. When you turn the metal piece beneath the teflon coupler, it moves the nozzle and the teflon coupler up against the bowden tube. If the bowden tube is held firmly in place, both sides are pressed against each other. I have now ß play with the bowden tube.

 

Good to know - I kinda figured that was probably what it was for (or could be used for at least). But it makes sense to have a way to get the last bit of slack out of the bowden.

 

[chrisp 10]

(Not to sound like a nag or know it all), but you guys should really try a spool roll with bearings!

My tension/friction problems with the extruder are not an issue anymore.

 

[illuminarti 18]

Nicolinux - With the white thing at the top, that's minimum tension. When your extrusion test was failing, was the extruder clicking back (indicating that it could grip the filament, but couldn't push it fast enough), or was it seeming to turn normally, but not enough plastic came out (suggesting that it was simply slipping past the filament, and grinding it a bit).

If the latter, tightening the extruder tension a bit (moving the white indicator block down) might help. If the former, I'm not really sure... but it wouldn't hurt to try.