UM2 Underextrusion at Low Volumetric Flow Rate

[kaledly 1]

Hi all, this is my first post and it's a long one. I'm relatively new to working with UM machines and I've got two of them at work that I'm having some extrusion issues with. Both machines are UM2e's with Olson Blocks installed and 0.4mm E3D nozzles.

 

Basically, I'm seeing a lot of gaps between lines in the infill and between perimeters which is an obvious sign of under extrusion. If I try and print at what I would consider normal temperatures for PLA (190-210C) the extruder will click and you can tell that it's obviously missing steps. With PLA, I have to crank the temperature up to 225-235C to stop the extruder from skipping at which point print quality goes down hill. Similarly, with PETG I have to print between 245-255C, which is also quite hot compared to normal.

 

I've rebuilt the hotend on both machines with new nozzles, PTFE couplers and bowden tubes and I've not seen any improvement in extrusion quality. The only thing that seems to work are unreasonably high temperatures or really low speeds. Looking at the volumetric flow numbers, I'm well within range of what I've read the machine should be able to do, so I'm a bit stumped as to why I can't achieve good flow. Nozzle temperature varies by a few degrees during printing (+/- 2C from target), but what I'm seeing in the microscopy pics below is that heat transfer just isn't sufficient to provide consistent extrusion. 

 

See below for some PETG test prints that I did for an example of the issue.

 

Left, Cura defaults for 0.2mm layer height.

Middle, Cura defualts, but slowed inner perimeters to 30 mm/s to match outer perimeter

Right, Cura defaults, but all speeds set to 20 mm/s

 

 

Left side print from top photo - Matterhackers black PETG - Cura defaults for 0.2mm layer height, 0.4mm line width, 30mm/s outer walls and top infill, 60 mm/s for inner walls. Note the gaps between perimeters and top infill lines. Top infill and outside perimeters printed at 30 mm/s have a volumetric flow of 2.4 mm^3/s, inner perimeters are double that at 4.8mm/s. Both should be within range of what a UM2 can do from what I've read (10 mm^3/s max, <5 mm^3/s preferred). 

 

 

Middle print from top photo - Matterhackers black PETG - Cura defaults for 0.2mm layer height, 0.4mm line width. Inner walls slowed to 30 mm/s. Volumetric flow rate of 2.4 mm^3/s

 

 

Right side print from top photo - Matterhackers black PETG - Cura defaults for 0.2mm layer height, 0.4mm line width. All speeds set to 20 mm/s. Volumetric flow rate of 1.6 mm^3/s

 

 

 

I think both machines are running the stock 25W heater cartridge. I have the 35W versions on hand and can install, but since I'm not seeing a temperature drop while printing faster (30-60 mms/) I don't think that's going to change anything since the lower wattage units seem to be able to keep up with power demand. The only other thing I can think of to fix the issue is upgrading the extruder to supply more feed pressure to get a more consistent extrusion, but I don't really want to drop $400 on the UM2+ upgrade kit to find out that that doesn't work either. 

 

Any ideas to fix this, or do I just have to print at super slow speeds with these machines?

 

[gr5 2,230]

I totally agree with your power assessment.  If the temperature is within 5C of the goal temp then you don't need a more powerful heater.

 

I think you just have typical UM2.  At 0.2 layer you shouldn't be going over 30mm/sec and 60mm/sec is the absolute fastest speed at 210C for a typicall, well running UM2 with the old feeder.  Faster than 60mm/sec and it will start clicking.

 

I do recommend you upgrade the feeders but it's not mandatory.  You have several options:

1) Get the $400 kit

2) Get a knock off copy kit from china at under $100.  on aliexpress.

3) You can do a meduza upgrade for $19 (and around an hours work typically for each printer)

 

I sell the Meduza kit in my store thegr5store.com

 

It involves printing some parts.  It's definitely more work than doing the "plus" kit.  All the instructions are on my website at gr5.org/med/

 

Here is my table of how fast a UM2 can print:

Here are top recommended speeds for .2mm layers (twice as fast for .1mm layers) and .4mm nozzle:
20mm/sec at 200C
30mm/sec at 210C
40mm/sec at 225C
50mm/sec at 240C
The printer can do double these speeds but with huge difficulty and usually with a loss in part quality due to underextrusion. 

 

You also might want to keep the petg on one arbitrary printer and the pla on the other.  The nylon degrades faster at 230C versus 210C and petg (and abs) can put up with the degradation better (I have no idea why).

 

[kaledly 1]

Thanks for the advice gr5. 

 

I did some more testing on one of my machines yesterday. I basically performed a modified melt flow index type test to find out how much material the extruder was able to push through the nozzle before it started to kick back and skip steps. TLDR, with a 0.4mm nozzle I'm only able to push about 2-3 mm^3/s through my machine before seeing issues. 

 

My test protocol was as follows:

 

Machine Setup:

• UM2 Extended
• Stock black extruder
• Olsson Block w/ 0.4mm E3D nozzle
• Test temperature range 190-260'C in 10'C increments.
• ~4m long free piece of filament removed from the spool.
• Machine heated to 240C when new filament was loaded and 60mm of filament was extruded to clear residue from previous test. 
• Pronterface was used to send extrusion commands and control machine temp.

 

Testing:

1. Heat machine to test temperature and wait 1 min for equilibrium. 

2. Extrude 20 mm of filament at specified rate (starting at 10mm/s). 

• If extruder clicks reduce extrusion speed by 5 mm/s and retest. 
• If extruder does not click, increase extrusion speed by 5 mm/s and retest. 

 

On to the results!

 

 

 

A couple of interesting points:

1. I had one old role of Gizomodorks black PLA that had really high results compared to all other spools of PLA I tested, so I reset the machine and verified the numbers, which were repeatable. Interestingly, this role old/brittle and was saturated with water to the point where it visibly steamed while coming out of the nozzle. I may try and stick in the oven to dry it and retest to see if the higher flow rates are due to moisture content. 
2. The Matterhackers PETG showing signs of under extrusion when printing walls is what kicked off this whole thing and it actually had really good maximum flow results compared to any of the PLA samples I tested. I saw under extrusion on inside perimeters at 60 mm/s print speed which only equates to about 5 mm^3/s of required flow, and the required flow value is less than half of what my test result showed at 245C. Based on that metric it looks like under extrusion starts to show up way earlier than when the extruder will kick back due to excess feed pressure. 
3. My data appears to be consistent with gr5's recommended print speeds for PLA (actually they may even be optimistic in my case). 

So, it looks like I need to upgrade the extruder on my machines and significantly lower my expectations regarding print speeds. 

Edited June 21, 2019 by kaledly

[gr5 2,230]

Nice graph!!  Thanks you for this.  I have done similar graphs on an older version of this forum and the text is still here on this forum but the graphs got lost.  Some points:

 

1) Please limit time that your block is > 220C.  The teflon degrades faster as temperatures rise.  260C is just not good for that teflon.  It lasts somewhere around 500-1000 hours at 210C and much less at 260C.  The teflon is touching the heater block so 1mm or so of that teflon is at that temperature.  And that's the part of the teflon that gets soft (part looks fine if you remove it) and collapses (squeezes) onto the filament under the pressure from the spring and the bowden at the top of the white teflon (you can see just by looking, without needing to take anything apart).

 

2) I've never seen wet PLA!  I've seen wet PVA, wet Nylon, but never wet PLA.  They steam and sizzle and pop and the filament comes out snowy -- full of air pockets.  Like a sponge.  If I leave Nylon out for just a few days it gets ruined and needs to be dried again.  Try putting PLA on a heated bed or in a cup of water and notice that it starts getting soft (like clay) around 52C.  If you dry PLA on the heated bed under a blanket or in an oven do not let the temperature exceed 50C!!!  At that low temperature it will take (I'm guessing) 10 hours to dry out just the outer 20 meters or so of filament.  Alternatively buy some 50 gram packs of color-changing dessicant and put the PLA in with those packs and recharge them when the change color until the spool goes a week without the packs changing color.  I've left out PLA for a few years and never seen it get wet.

 

3) I tested ABS and nylon for max flow but never PETG!  I had no idea it was less viscous.  Good to know.

 

4) That sudden increase in flow at 260C is scary.  I think there is some chemical change that may be a bad thing at that temperature. I really don't know.  But at least realize that for ABS, if you get it too hot it caramelizes and clogs the nozzle.  Unfortunately that "too hot" temperature is very close to normal printing temperature for ABS.  ABS just sucks in many ways as a filament - stay away from it.

 

[gr5 2,230]

Yeah something is wrong with your printer.  It's printing at about half the volume you should be able to achieve.  If you go into the menu on the machine you can see how many hours is on this printer.  How old is it?  How many hours of printing?

 

Most likely the teflon needs replacing again already - probably due to being used at higher temperatures.  There could be overheating issues or current-to-the-extruder-stepper issues but I doubt it.  Also your bowden probably should be replaced.  Bowden's get worn after a while.  I sell regular and also "super slip" teflon bowdens.  It really helps a lot to have the teflon bowden but personally I only have it on one of 7 printers.  It also is more opaque so you can't see as much what is going on.

 

Here is a complete list for what can cause a UM2 to underextrude.

 

CAUSES FOR UNDEREXTRUSION UM2 AND HOW TO TEST FOR THEM AND REMEDY THEM

One can quickly check the feeder (it takes seconds) so I would always do that first.  The feeder on UM2+ and UM2 can push with 10 pounds (5kg) force easily so first thing is to lift that lever (if um2+, if um2 regular do move material) and insert filament so it is only part way into the bowden (e.g. half way).  Then move the filament with "move material" command in menu so the feeder is energized and then pull down on the filament under the feeder.  You should be able to pull with 5 to 10 pounds force without it slipping.  Then have it move the material up while you pull down.  5 pounds force (2kg) is enough to make decent prints at moderate speeds listed below but you need 10 pounds (5kg) to print fast (e.g. 0.2mm layers and 60mm/sec)

As far as underextrusion causes - there's just so damn many.  none of the issues seem to cause more than 20% of problems so you need to know the top 5 issues to cover 75% of the possibilities and 1/4 people still won't have the right issue.  Some of the top issues:
1) Print slower and hotter!  Here are top recommended speeds for .2mm layers (twice as fast for .1mm layers) and .4mm nozzle:
20mm/sec at 200C
30mm/sec at 210C
40mm/sec at 225C
50mm/sec at 240C
The printer can do double these speeds but with huge difficulty and usually with a loss in part quality due to underextrusion.  Different colors print best at quite different temperatures and due to imperfect temp sensors, some printers print 10C cool so use these values as an initial starting guideline and if you are still underextruding try raising the temp.  But don't go over 240C with PLA.

2) Shell width confusion.  Shell width must be a multiple of nozzle size (in cura 15.X.  In cura 2.X it doesn't matter as much but still makes a difference).  For example if nozzle size is .4mm and shell width is 1mm cura will make the printer do 2 passes with .5mm line width which is possible but requires you to slow down much more to make a .5mm line out of a .4mm nozzle.  If you really want this then set nozzle size to .5mm so it's clear what you are asking Cura to do for you.

3) Isolator - this is most common if you've printed extra hot (>240C) for a few hours or regular temps (220C) for 500 hours.  It gets soft and compresses the filament under pressure.  It's the white part touching the heater block.  It's very hard to test when not under full pressure (spring and bowden) so sometimes it's best to just replace it.  Also if you notice parts of it are very soft (the blacker end where it touches higher heat) then it's too old and needs replacing.

4) Curved filament at end of spool - if you are past half way on spool, try a fresh spool as a test.

5) curved angle feeding into feeder - put the filament on the floor -makes a MASSIVE difference.

6) UM2 only: Head too tight?  Bizarrely MANY people loosen the 4 screws on the head by just a bit maybe 1/2 mm and suddenly they can print just fine!  Has to do with pressure on the white teflon isolator.
6b) UM2 only: Bowden pushing too hard - for the same reason you don't want the bowden pushing too hard on the isolator.
6c) Um2 only: Spring pushing too hard.  Although you want a gap you want as small as possible a gap between teflon isolator and steel isolator nut such that the spring is compressed as little as possible.

7) clogged nozzle - the number one problem of course - even if it seems clear.  There can be build up on the inside of the nozzle that only burning with a flame can turn to ash and remove.  Sometimes a grain of sand gets in there but that's more obvious (it just won't print).  Atomic method (cold pull) helps but occasionally you need to remove the entire heater block/nozzle assembly and use flame.  I found soaking with acetone does not help with caramelized pla. Even overnight.  Maybe it works on ABS though. Simpler cold pull: https://www.youtube.com/watch?v=u07m3HTNyEg

😎 Temp Sensor bad - even the good ones vary by +/- 5C and bad ones can be any amount off - they usually read high and a working sensor can (rarely) fail high slowly over time.  Meaning the sensor thinks you are at 220C but actually you are at 170C.  At 170C the plastic is so viscous it can barely get out of the nozzle.  You can verify your temp sensor using this simple video at youtube - on you tube search for this: mrZbX-SfftU

9) feeder spring issues - too tight, too loose.  On the black UM2 feeder you want the tension such that you can clearly see the diamond pattern biting into the filament.  You want to see at least 2 columns of diamonds.  4 columns is too much.  On the white UM2 plus and UM3 feeders you usually want the tension set in the center.

10) Other feeder issues, one of the nuts holding UM2 and UM3 together often interferes with the feeder motor tilting it enough so that it still works but not very well.  Other things that tilt the feeder motor, sleeve misaligned so it doesn't get a good grip.  Gunk clogging the mechanism in there.

11) Filament diameter too big - 3mm is too much.  3mm filament is usually 2.85mm nominal or sometimes 2.9mm +/- .05.  But some manufacturers (especially in china) make true 3.0mm filament with a tolerance of .1mm which is useless in an Ultimaker.  It will print for a few meters and then clog so tight in the bowden you will have to remove the bowden from both ends to get the filament out.  Throw that filament in the trash!  It will save you weeks of pain

11b) Something wedged in with the filament.  I was setting up 5 printers at once and ran filament change on all of them.  One was slowly moving the filament through the tube and was almost to the head when I pushed the button and it sped up and ground the filament badly.  I didn't think it was a problem and went ahead and printed something but there was a ground up spot followed by a flap of filament that  got jammed in the bowden tube.  Having the "plus" upgrade or using the IRobertI feeder helps you feel this with your hand by sliding the filament through the bowden a bit to see if it is stuck.

12) Hot weather.  If air is above 30C or even possibly 25C, the air temperature combined with the extruder temperature can soften the filament inside the feeder such that it is getting squeezed flat as it passes through the feeder - this is obvious as you can see the problem in the bowden.  The fix is to add a desk fan blowing on the back of the printer.  Not an issue on the UM3 or UM2 "plus" series.

13) Crimped bowden.  At least one person had an issue where the bowden was crimped a bit too much at the feeder end although the printer worked fine when new it eventually got worse and had underextrusion on random layers.  it's easy to pull the bowden out of the feeder end and examine it.

14) Worn Bowden.  After a lot of printing (or a little printing with abrasive filaments) the bowden resistance can be significant. It's easy to test by removing it completely from the machine and inserting some filament through it while one person holds it in the U shape.  Preferably i nsert filament that has the pattern from the feeder.

15) Small nozzle.  Rumor has it some of the .4mm nozzles are closer to .35mm.  Not sure if this is actually true.  I'm a bit skeptical but try a .6mm nozzle maybe.

16) CF filament.  The knurled sleeve in the extruder can get ground down smooth - particularly from carbon fill.  4 spools of CF will destroy not just nozzles but the knurled sleeve also.  Look at it visually where the filament touches the "pyramids".  Make sure the pyramids are sharp.

17) Hot feeder driver.  I've seen a more recent problem in the forums (>=2015) where people's stepper drivers get too hot - this is mostly a problem with the Z axis but also with the feeder.  The high temps means the driver appears to shut down for a well under a second - maybe there is a temp sensor built into the driver chip?  The solution from Ultimaker is that they lowered all the currents to their stepper drivers in the newer firmware.  Another solution is to remove the cover and use desk fan to get a tiny bit of air movement under there.  TinkerMarlin lets you set the currents from the menu system or you can send a gcode  to lower the current.  Ultimaker lowered the default currents in July of 2015 from 1300ma to 1200ma for X,Y,Z but left extruder at 1250. Other people (I think the support team of a major reseller but I forget) recommend X,Y,Z go down to 1000mA.

M907 E1250

Above sets the extruder max current to the default - 1250mA.  So try 900mA.  This will only change until next power cycle so if you like your new value and want to save it use M500.  You can just put these into an otherwise empty gcode file and "print" this and it will change.  Or get tinkergnome marlin!  You will wonder how you lived without it: https://github.com/TinkerGnome/Ultimaker2Marlin/releases

M907 E900
M500

18) third fan broken.  This tends to cause complete non-extrusion part way through a print.  In the rear of the head for UM2 and the front of the head for UM3.  Without this fan several things can go wrong.  It can take a while as usually you also need several retractions to carry the heat upwards.  There are a few failure mechanisms and I don't understand them all.  One of them is probably that the molten PLA spreads out above the teflon and sticks to the metal in a core or fills the gap at the base of the bowden in UM2.  Later it cools enough to keep the filament from moving up or down.

19) Spiralize/vase mode.  This is a rarely used feature of Cura but you might have left it on by accident?  In this mode the wall of your part is printed in a single pass.  So if you have a .4mm nozzle and the wall is .8mm thick it will try to over extrude by 2X.  This is difficult to do and may instead lead to underextrusion.

20) non-standard or bent fan shroud.  Sometimes people print some fan shroud off of thingiverse or youmagine out of PLA or ABS.  Some of these are great but most of them are crap.  One needs to do good air flow modeling.  Also if it's PLA it will slump and direct air differently.  Air directed at the block or nozzle can cause severe underextrusion and also sometimes HEATER ERROR.  Put the original shroud back on or just turn off the fan to prove that the fan is the problem.

21) Firmware settings - for example UM2+ firmware on UM2 or vice versa will cause 2X over extrusion or 2X underextrusion.  Downgrading or changing firmware can mess up steps/mm and other settings - so if you updated firmware and then problems started then do a "reset to factory settigns" which corrects all the steps/mm values.

22) too many retractions (this causes complete failure) - if you have too many retractions on the same piece of filament you can grind it to dust.  10 is usually safe.  20 is in the danger zone.  50 should guarantee failure.  You can tell cura to limit retractions to 10 per a given spot of filament.  Do this by setting "maximum retration count" to 10 and "minimum extrusion distance" to your retraction distance (4.5mm for UM2 and 6.5 for UM3 and 8mm for S5).

23) Brittle filament.  Espciallty with older PLA but even brand new pla can do this.  If you unspool some (for example if it's in the bowden) for many hours (e.g. 10 hours) it can get extremely brittle and it can snap off into multiple pieces in the bowden.  It's not obvious if you don't look for this.  Then it starts printing just fine and at some point one of those pieces reaches the print head and gets hung up somewhere and the printer suddenly stops extruding for now apparent reason.  This usually happens within the first meter of filament - once you get to printing the filament that was recently on the spool it should be fine from then on.

24) The "plus" feeder can have an issue where the filament doesn't sit properly for one print and it permanently damages the arm inside the feeder as shown by this photo - the hole is ground down asymetrically: http://gr5.org/plus_feeder_issue.jpg

25) Other feeder issues.  You can test the feeder by putting the filament only part way down the bowden and with the feeder electrically turned on (or moving) pull very hard on the filament until it slips.  You should be able to pull with about 5kg or 10 pounds of force before it slips.  4kg is acceptable.  2kg is a problem.  If the stepper motor isn't engaged you can try going into the move menu.  After a minute or so power is removed again from the feeder stepper.
 

[kaledly 1]

Thanks for the tips. The printers both have 2-3k hours on them, but I've performed maintenance on both hot ends recently (<100 hours ago) by installing new E3D nozzles and TFM couplers instead of the older PTFE style ones. I didn't replace the Bowden tube itself, but I did trim it down a 1/4-1/2" on each end and re-bevel the edges. I have a new Bowden tube from UM on hand so I can try installing that and seeing if it makes a difference.   

 

On your list, the only thing I think I have an issue with is the isolator spring pressure. I've got a 1-2mm gap between the isolator nut and the bottom side of the coupler. I'm not sure I understand the reasoning behind minimizing this gap. The only way I can see to reduce this gap is to unthread the Olsson block a bit to reduce the gap, but if I do that, wouldn't I run into an issue with molten filament leaking between the isolator nut and the block?

[gr5 2,230]

The isolator nut (the round nut) is designed initially for 2 nozzle printing to let you get the two nozzles the same height within 0.01mm (that's 10X thinner than 0.1mm - not a typo).

 

So, no, it shouldn't leak. The thing that stops the leak is the gap between teflon and block.

 

Please do the "pull" test and see how many pounds of force your feeder can pull.  Test it once while "cold" and test it again after stepper is well warmed up (it's not really the stepper that has the heat issue - it's the stepper driver).  Explained above issue #1 above.  If pull test fails it could be #10,#17, or something else.

 

Then if that passes concentrate on: 3,4,5,14,15

 

It really could be #3 - that you already wrecked your teflon somewhat.  Just 3 hours at 240C might be enough.  I don't know.  One test for #3 is to remove bowden and nozzle, clear block of filament and let cool.  Then slide filament through.  Your budget is 10 pounds force - how much of that is used up by that friction?  It's okay to drill out the teflon in situ.  do it slowly and carefully.  This will give you a few more days of printing while waiting for replacement.

 

#15 - did you have this problem with 2 different nozzles?  Maybe the newer nozzle is bad?  This is less likely but worth considering.

 

[gr5 2,230]

Note that if you lower the olsson block too much it hits the metal fan shroud and you have a whole new set of problems (block can't get up to temp once fans come one).

[kaledly 1]

Thanks again for the tips gr5, I really appreciate you taking the time to help. I'm out of time to experiment today, but I'll take a look at it again next week (the printers are work machines) and see if I can make any progress.