UM2 extrusion rates revisited

[gr5 2,167]

I revisited my tests of max extrusion rates. First for the results - mostly look at the dark blue line which is PLA at .2mm layers - you can go twice as fast for .1mm layers:

 

When I tried to print any faster than the blue line the extruder slipped backwards with the exception of the pink dot where it didn't slip but I could clearly see underextrusion visible on the printed part by about 5 to 15% underextruded (I compared with flow purposely set to 95% and 90%).

 

The yellow line was max extrusion rates when PLA is printed with nozzle not touching part. This I later decided was unrealistic because when the nozzle is .2mm above the part the pressures will be higher. At .1mm height you should be able to print almost 2X as fast because you are extruding half the amount of PLA.

 

The light blue line was a test for nylon PA6 - also in air so you have to subtract 20% or so I assume for realistic printing.

 

Here is the test piece I did for the dark blue line in the graph:

Basically I printed a 30mmX30mmX30mm cube with zero shell (hence the strange edges). I would print at a given temp and speed up and slow down the feedrate until I found the fastest speed that didn't slip. At 240C I could go faster than 100mm/sec but I started to see underextrusion even without the slipping hence the pink dot at 110mm/sec which didn't slip but still underextruded. At the slower speeds/temps I never saw this underextrusion without slipping also occurring. I don't understand why this is different at 110mm/sec but something to do with how stepper motors work maybe or the momentum of the faster moving stepper. Implying that higher servo currents might slip less but you probably get underextrusion anyway.

The previous experiment (with the yellow line data only) is here:

http://umforum.ultimaker.com/index.php?/topic/3418-um2-extrusion-rates/

EDIT:

Steve from Ultimaker does material tests and did a more thorough test that concentrated more on different manufacturers but on average got almost identical data - this is very thorough and informative!:

http://gr5.org/PLA_undereextrusion_tests.pdf

 

[illuminarti 18]

Nice test, George!! Thanks :-)

 

[aaronalai 2]

Thanks for the details! It would be interesting to see if others can reproduce the same results. Just making sure, but the x-axis is temp and the y-axis is mm/s right? Also, illuminarti, your calibration cup is a really great idea. I'm going to be revisiting these posts when I get my UM2, they seem like fantastic reference points to test for a well calibrated machine.

 

[illuminarti 18]

Yes, speed on vertical axis, temp on horizontal one.

 

[joergen 2]

Yes, this looks in line with what I have been seeing. the UM2 has lower throughput rates than the UM1, and I would say this is due to the smaller heat reservoir and the smaller heater.

The standard UM2 seems to max out at 8mm^3/sec in the chart above, and my modified UM2 (0.5mm nozzle) is maxing out between 12-13mm^3/sec (260deg ABS). Just as a comparison, my modified UM1 (0.56mm nozzle and 50W heater) was able to push more than 30mm^3/sec.

Max throughput seems to be squarely related to nozzle size and heater size:

0.40mm nozzle = 0.1256mm^2 at 25 Watt = 8mm^3/sec

0.50mm nozzle = 0.1963mm^2 at 25 Watt = 12mm^3/sec

0.56mm nozzle = 0.2463mm^2 at 50 Watt = 30mm^3/sec (about 4x of the standard UM2 throughput)

All of this is of course simplified, leaving the material and the print temperature (aka viscosity) out of the equation. But this should give some decent data for the slicers to calculate the max speed, or increase the speed to the reasonable max.

But looking at this data, it seems that ABS/PLA share a constant factor (extrusion per sec / watt / nozzle size) that is roughly in the vicinity of 2.5 (mm/Ws)

 

[alluvian 0]

Knowing this limitation of the UM2 (slower max extrusion rate), would you advanced users say you are dissapointed with the performance of the UM2 compared to the UM1?

Also, would you suggest that a new user bore out the stock nozzle to something larger?

If so, how big? I seem to see 0.5, 0.65 and 0.8 (for wood) to be fairly common.

 

[gr5 2,167]

It's not clear if the limitation is because the extruder has about half the power or if the tiny heater block takes longer to melt PLA. If the problem is only the former then yes, it would be great to have .5mm nozzles and it would be great if I could easily swap nozzles like I can on the UM1.

For really beautiful prints, the speed limitation is not a problem. You have to print slow anyway to get beautiful prints.

If I had to give up my UM1 or UM2 I would be torn to choose which one to lose. I usually only use the UM2 now but it's nice to be able to hack the UM1. If I ever get a stiff bed for the UM1 that doesn't need adjusting every day I think I might lean to this being a better printer. Not sure. They are both great printers.

 

[gr5 2,167]

.7mm nozzle!

I repeated the 30mm cube test with a 45mm cube to see how fast I could print with .7mm nozzle with .2mm layer height. At 230C I was able to print 18mm^3/sec!!

Here are results:

220C 110mm/sec no skip

230C 130mm/sec no skip (75mm/sec no underextrusion visible)

240C 140mm/sec no skip (100mm/sec no underextrusion visible)

look at how fast the extruder spins at 240C at 18.2mm^3/sec:

 

 

[woofy 12]

How does print quality compare with the 0.4mm nozzle?

 

[IRobertI 520]

Am I reading it right that when printing at 130mm/s you're not skipping but there's still under extrusion and 75mm/s is the "real" speed you're able to print at?

Looks like I might have to find someone local to drill out my hotend sometime in the future. Maybe not the full 0.7 though, I'm thinking 0.5-0.6 for me. Will have to think about it. It's really too bad that you can't just switch nozzles like on the UM1 :/

 

[gr5 2,167]

Am I reading it right that when printing at 130mm/s you're not skipping but there's still under extrusion and 75mm/s is the "real" speed you're able to print at?

 

Yes. It was very minor undextrusion. I don't think it would be visible from the sides. But on a solid layer there was a tiny gap between lines.

By the way at 18mm^3/sec when you get a skip, the underextrusion only lasts for a few mm of travel! It's easy to miss.

 

It's really too bad that you can't just switch nozzles like on the UM1 :/

 

I guess each time you do it it gets easier? If you ask Sander nicely he will almost certainly give you a free one.

 

[gr5 2,167]

How does print quality compare with the 0.4mm nozzle?

 

Good question. I printed one thing so far with a .65mm nozzle on the UM Original and it came out great. In fact I believe that's Illuminarti's primary nozzle on the UM Original (.65mm). Everything in this article I think was .65mm nozzle:

http://www.extrudable.me/2013/08/13/printing-with-laybrick/

 

[joergen 2]

If more people are interested, I can get a bunch of UM2 nozzles from UMHQ and drill them to custom size/specs (I am located in NYC).

 

[illuminarti 18]

Yes, indeed, all of those LayBrick prints were with a 0.65mm nozzle.

I don't think it's too surprising that the UM2 also experiences some degree of non-skipping under-extrusion at high volumes per second. The knurled bolt drive mechanism is pretty much identical to the UM1 at the point where the filament meets the printer - similar tooth pattern on one side, and spring loaded pressure wheel on the other. So, I'd expect to see the same sort of back-pressure related slipping that http://www.extrudable.me/2013/04/18/exploring-extrusion-variability-and-limits/

 

[medicalmodeller 5]

So I have a 2 week old UM2. I have printed out 8 large gears (2 hours each) and a model of a upper end of humerus without a problem. The latter took about 7 hours to print and was/is perfect.

Then I tried to print a heel bone which was a big file 80+Mb when in stl. I reduced the size to print 50% of true size. It was meant to take 6 hours to print. I have tried repeatedly to get a complete print but the filament stops flowing after 2 hours and the print head carries on printing air. I have now done an Atomic unblock on 3 or four occasions. Each time the nozzle blocks after 2 hours or so

I have Cura set to 50mm/sec and all other settings to default

I have done the extrusion test with the print head set to 240

Here is the result

It seems that extrusion is not the problem - or is it?

This is so frustrating. I feel like sending the printer back it is this difficult to consistently get a good print

TP

 

[gr5 2,167]

"printing in air" clogs can be caused by many things.

Some people have feeder steppers that are too hot and the filament gets very soft at the feeder wheel especially if there are lots of retractions. I don't have this problem. But you can recognize it because you get a long section of flattened filament.

There could be house dust. Or tangles in the filament. I recommend putting the filament on the floor so that it goes straight up into the feeder and not at an angle.

It's possible to continue a "failed" print. I posted about this many times so you can google that on the forums. The key thing is to not let the glass bed cool down while you learn what to do. Otherwise the part "pops" off the glass.

You may be printing too hot nozzle and the PLA gets gummed up after a long time (I kind of doubt this). Or maybe you are printing too fast.

What is your layer height, print speed, nozzle temp?

 

[gr5 2,167]

Oh and also sometimes the filament gets stuck in the bowden. Please measure the filament and also maybe check if it flows smoothly through the bowden (difficult to do unless you detach from the feeder).

 

[gr5 2,167]

Steve from Ultimaker does material tests and did a more thorough test that concentrated more on different manufacturers but on average got almost identical data - this is very thorough and informative!:

http://gr5.org/PLA_u...usion_tests.pdf

 

[Dim3nsioneer 557]

Very informative indeed. Do you know if the three 'suppliers' were actually three different brands (at least two of them being not UM) or three companies which all deliver filament to UM which is then sold as UM-Filament?

 

[gr5 2,167]

I learned this past week that Ultimaker uses multiple suppliers for their existing filament. I assume the three different suppliers consist of existing and potential suppliers.

Steve wants to get the best material for each color that UM sells. "best" is complicated when you include things like strength, flexibility, print flow, glass temp, layer adhesion, shininess, color consistency, etc.

 

[gr5 2,167]

By the way Steve has many more PDFs like this one testing other properties but some of those mention the suppliers by name so he has to "scrub" them before they can be published here.

 

[IRobertI 520]

And he's already struggling to find enough time in a day to do the tests he wants so it's probably better to let him "finish" (can you ever "finish" something like this?) his work instead of censoring reports IMHO

 

[ultiarjan 1,223]

Please just publish... find-replace takes only a few minutes.....

[gr5 2,167]

I don't have a pdf editor unfortunately.

UltiArjan - did you also ask me about antishake video software? I forget who asked me where to get it - I owe someone my cheat sheet on doing anti-shake.

 

[ultiarjan 1,223]

ha, I meant for steve to do the find>replace but I do have a pdf editor ......

We did talk about the video antishake in a group of people, I remember mentioning the microsoft tool for timelaps. Why don't you just share what you use in the gossip point section ....