Temp drops during prints on UM3

[waizman 0]

Hi!

I have an UM3, converted to accept 1.75mm filament (PTFE tubes in the printcore, nothing more was needed) and I've noticed some of my prints failed with no apparent reason so i decided to pinpoint the problem. 
 

I've managed to improve the issue by reducing the "Minimum Extrusion Distance Window" to 3mm, and the "Max Retraction Count" to 10, as well as printing with higher (230c) temp.
Other settings - Speed 80mm/s, nozzle 0.4mm, layer height 0.15mm (= Flow 4.8mm^3/s, I try to not go over 5.6mm^3/s).
 

Now it seems my prints mainly fail due to temperature drops - by looking the temp graph, I see that sometimes the nozzle temp drops below 190c and that's when I get a lot of underexcretions and fails. Why? I don't know...

I believe it is caused by the fact that it takes some time to heat up the fresh filament and the heater is just not fast enough.
I've noticed it tend to happen in areas with more retractions like the top of the steering wheel in 3DBenchy (photo attached).
 

Any advice on how to better retain the chosen temperature? 
Latest firmware (5.3) and Cura (5.6.0) versions.
 

Thanks in advance

Edited March 17 by waizman

[waizman 0]

Some more info:

- This happens almost always on Extruder 1 (Left one). Extruder 2 works well with the same exact settings or only different by the temperature setting (210c).

- The temperature in the printer lcd is always correct (same as set in Cura).

- Tried 2-3 printcores on Extruder 1 and nothing changed.

- issue happens whether Extruder 2 is disabled or enabled in Cura.

- Filament is Generic PLA+ (Purple). Also have eSun PLA+ and Ultimaker CPE (PETG) to test with.

- It seems the non-printing printcore never drops below 40-45c.

 

Added more images - temperature graph and its corresponding print (3DBenchy trimmed to the problematic area).

[gr5 2,176]

Wow.  Yeah that's extremely unusual.  These nozzles should have plenty of power.  This makes no sense.

 

Why is extruder 2 heating up at all?  Are you positive it isn't printing support or something with extruder 2?

 

If a nozzle gets 100% power and still keeps cooling for more than 30 seconds, the printer will halt with an error code ("heater error").  Why isn't it halting with an error code?  Maybe it's not getting 100% power?  Why not?  It just doesn't make sense.

 

I'd try swapping with another core.  Another AA 0.4.  Your printer should have come with a  BB 0.4 and two AA 0.4 cores.

 

Also try going into the TUNE menu on the printer and go to the nozzle temp and it should show you what it thinks the temp is.  YOu can adjust it upwards.  You can also try slowing down the printer by 2X to see if the lower amount of flow results in less watage needed.  You can also slow down the fans, sometimes the air from the fan gets diverted to the nozzle.  Either bouncing off the heated bed or if there is no silicone sheild.  Did you remove the silicone shield on the bottom of your print head?  Maybe show a picture of that.  They cost something like 3 for 5 dollars/euros.

 

[waizman 0]

Answered in red font:

7 hours ago, gr5 said:

Wow.  Yeah that's extremely unusual.  These nozzles should have plenty of power.  This makes no sense.

 

Why is extruder 2 heating up at all?  Are you positive it isn't printing support or something with extruder 2?

Nothing, it just got heated up for bed leveling and then should be inactive.

As mentioned, I also tried disabling it via Cura but got same results.

 

If a nozzle gets 100% power and still keeps cooling for more than 30 seconds, the printer will halt with an error code ("heater error").  Why isn't it halting with an error code?  Maybe it's not getting 100% power?  Why not?  It just doesn't make sense.

 

I'd try swapping with another core.  Another AA 0.4.  Your printer should have come with a  BB 0.4 and two AA 0.4 cores.

Tried that. Same results.
Also some printcores have original nozzles and some AliExpress nozzles - same results.

 

Also try going into the TUNE menu on the printer and go to the nozzle temp and it should show you what it thinks the temp is.  YOu can adjust it upwards. 
Tried that. That's why I now print in 230c instead of 210c usually.

As mentioned in previous post - the temperature in the printer lcd is always correct (same as set in Cura).
 

You can also try slowing down the printer by 2X to see if the lower amount of flow results in less watage needed.  You can also slow down the fans, sometimes the air from the fan gets diverted to the nozzle.  Either bouncing off the heated bed or if there is no silicone sheild.
I usually print at ~70mm/s @ 0.2mm layer/0.4mm nozzle. Tried lowering to 0.15mm layer and 50mm/s speed. 
Layer height helped but still fails. Speed doesn't seem to have much effect.

Another thing I tried and managed to finish one 3DBenchy - changed the Flow Equalization Ratio from 100% (Speed) to 0% (Flow).

Did you remove the silicone shield on the bottom of your print head?  Maybe show a picture of that.  They cost something like 3 for 5 dollars/euros.
I installed a new one about a month ago. It's in good condition. You want me to remove it?

The old one was well worn, and it took me a long time to find a replacement here in Israel 😕

Added pics for you.

 

Also did a "heater test" in the maintenance menu - got these results:

I've noticed extruder 2 heats up faster then 1...

 

Thanks ❤️

Edited March 18 by waizman
Added heater test

[gr5 2,176]

I don't like the way the temperature jumps to 700C briefly.  Both in your latest graph and the earlier graph. Something is wrong there.  There is a loose connection somewhere between the pt100 sensor and the bottom of your printer so it could be in the printcore (most likely), where the printcore connects to the print head, in the printhead connector, or where it connects to the PCB under the printer.

 

It's common on the UM3 for the printhead cable to pull up and eventually out of the connector.  On the top of the print head, remove the two very long (maybe 6cm) bolts.  Nothing will fall apart.  Then on the back half you can pop off the top cover - the back half of the top cover of the print head.  Then you will see where the electric cable goes into the print head and to a connector.  Push firmly down on the connector noting if it moved at all.  Use a flat head screwdriver or similar so you can get a good look at it while you press down and see if it moves.  If it does move, when you put it back together, put some paper or masking tape around the cable as a shim (google "shim wiki").

 

If the cable is fine then you should really throw away that printcore. It's very difficult to make those sensors reliable.  You can't solder the pt100 to the wiring because solder melts 200C to 250C (depending on the type of solder).  So instead the wiring is crimped to the pt100 and with lots of heating and cooling the crimp can fail.  This is somewhat common (maybe 5% of all temp sensors fail within a year of printing - mostly a guess based on my printers).

 

Do you still have all 3 of your original printcores?  If not, it's time to buy one.  Or you can use the BB 0.4 for PLA.  It will work just fine.  Consider using the BB as a test to see if this fixes your issue and if it does then we have definitely diagnosed the problem.

 

 

[gr5 2,176]

 

[waizman 0]

41 minutes ago, gr5 said:

I don't like the way the temperature jumps to 700C briefly.  Both in your latest graph and the earlier graph. Something is wrong there.  There is a loose connection somewhere between the pt100 sensor and the bottom of your printer so it could be in the printcore (most likely), where the printcore connects to the print head, in the printhead connector, or where it connects to the PCB under the printer.

 

It's common on the UM3 for the printhead cable to pull up and eventually out of the connector.  On the top of the print head, remove the two very long (maybe 6cm) bolts.  Nothing will fall apart.  Then on the back half you can pop off the top cover - the back half of the top cover of the print head.  Then you will see where the electric cable goes into the print head and to a connector.  Push firmly down on the connector noting if it moved at all.  Use a flat head screwdriver or similar so you can get a good look at it while you press down and see if it moves.  If it does move, when you put it back together, put some paper or masking tape around the cable as a shim (google "shim wiki").

 

If the cable is fine then you should really throw away that printcore. It's very difficult to make those sensors reliable.  You can't solder the pt100 to the wiring because solder melts 200C to 250C (depending on the type of solder).  So instead the wiring is crimped to the pt100 and with lots of heating and cooling the crimp can fail.  This is somewhat common (maybe 5% of all temp sensors fail within a year of printing - mostly a guess based on my printers).

 

Do you still have all 3 of your original printcores?  If not, it's time to buy one.  Or you can use the BB 0.4 for PLA.  It will work just fine.  Consider using the BB as a test to see if this fixes your issue and if it does then we have definitely diagnosed the problem.

 

 

Thanks for your answer.

I am VERY proficient in disassembling things, don't worry 😉

I have 6 print cores and I've tried 3 different ones in my fixing attempt - all behave the same.

 

I don't like that 700c spike either, but I don't know how to fix it. I also saw it was very common in other people's graphs, so I thought it was not a big deal...

 

Tested the resistance in the back chip connector as well - Heater is ~25ohms, sensor is ~110ohms. Looks fine and same as other people.

 

Also unscrewed the screw between the printcores 1/4 turn as mentioned here:

Print core is taking too long to heat up - UltiMaker 3D printers - UltiMaker Community of 3D Printing Experts

No change.

 

Tested the filament location and holder - no change.

 

I'm now testing another filament from another brand. I have a feeling it's the purple one that's acting weird... Even though it doesn't make sense... 😄

 

[gr5 2,176]

1 hour ago, waizman said:

I also saw it was very common in other people's graphs, so I thought it was not a big deal...

oh!  Maybe it's not a problem then?

 

But the whole thing is strange.  There is a specific test that is in (I thought) all Ultimaker printers where if a printcore is getting 100% power for 30 seconds and the temp doesn't go up by 1C then it fails.  I guess that means it is not getting 100% power.  The PID values may need some adjustment.  You could increase the P value.  This is just strange.  I've never heard of anyone adjusting their PID values for an ultimaker brand printcore.

 

I'm still leaning to getting a new printcore.  Did you check the cable in the printhead yet?

 

I could help you program the printcore PID I suppose.  I'm not a fan of that.  Most printcores have zero values in the PID eeprom and uses default PID settings but I can show you how to program PID values into the printcore.  First step is to put the printer into developer mode and then ssh into user/pass ultimaker/ultimaker

 

If you can get that far I'll lookup how to program the PID.

 

[waizman 0]

As I said, I tested 3 printcores and they all act similarly...

Some of them are already re-programed (BB to AA, of course I changed the nozzle to match) as instructed here:

Print core BB-0.4 wrongly recognized as a AA-0.8 - Improve your 3D prints - UltiMaker Community of 3D Printing Experts

 

The 700c spike is momentarily and as far as I recall it happens on all 3 printcores. So I think it's not them.

I will probably disassemble the whole head tomorrow and make sure everything is tight.

---

So I changed the purple filament I was using and the print was successful. Not the best (stringing-wise mostly), but it was a 5-YEAR-OLD filament I had laying around. Photos attached. It's a 56 minute Benchy BTW. Not bad.
 

Some things I've noticed:
- The graph looks much better to me, overall it's much more stable.
- Retractions cause the small temp spikes in the heater (Brown) line.

- There is a steady decline in the Brown line trend.
- Orange temp sensor line is rock solid.

- The previous two points mean the heating block steadily needs less energy to keep the temperature? 

- The 700c spike is momentary but it causes the heater to drop to ~50c and then overshoot the Orange sensor line. Not a big deal probably but if it is fixable - why not.

 

Now what I don't understand is WTF is up with the purple filament? It's new and shouldn't perform worse than a 5 y/o filament that got no care at all...

My primary goal is to make the system failproof - maximum relatability and quality.
Once that have been achieved - find the maximum speeds possible while maintaining reliability and quality.

Edited March 18 by waizman
My cat sent it before I finished writing ;)

[gr5 2,176]

oh!  I totally misunderstood the graph.  Okay so the orange graph is very flat.  That's fine then.  I was trying to figure out how the temperature was dipping.  But it wasn't.  It's the brown graph that is dipping - that's how much power is applied.  For some reason it needs less power as you go.  Probably extruding slower - or actually as you get farther from the glass bed, there is less bouncing up of the moving air around the nozzle so less nozzle cooling.  That's probably it.

 

Okay so there's nothing wrong with the temperature profiles.  A red herring.  Well... not much wrong.  Something is loose I suppose and it measures infinite resistance and interprets that as 700 ohms, then kills power to nozzle, then temp retruns and it has to compensate for a bit for the lack of power but briefly enough that nozzle doesn't deviate from goal temp by much.

 

So your print failure might have nothing to do with temperature.

[gr5 2,176]

Unlike some filaments, PLA tends to be able to last forever.  It doesn't need to be kept dry.  Although if you unspool "old" (even just 1 year old sometimes) PLA into the bowden and leave it in the bowden for 24 hours, many PLA formulations will crack.  If retract the PLA back onto the spool before turning off the printer it's usually fine.  "old" PLA does not like to be straightened out for more than a few hours.

 

So that could be it.  You can kind of tell because sometimes you can see the PLA has cracks in it when looking through the bowden.  And when there is a retraction, sometimes you see where the crack is closest to the feeder which could be anywhere along the bowden.

 

One fix is that if you leave filament in the bowden for more than 8 hours, take it out and cut off the last meter and throw that away.  You can test the filament by bending it and it should not break.  It should just bend in a knot.  The longer you leave it unspooled, the more brittle the filament gets.

 

Then the breaks in the filament reach the print head and sometimes they pass through but other times they get stuck in the printhead and your print fails.  Once you get past the first meter (typically an hour or so) you are usually fine.

 

I imagine 1.75mm filament isn't as bad because bending it is much less stressful for the plastic.

 

 

[gr5 2,176]

So wait - you are using 1.75mm pla through smaller bowdens.  Does the bowden go down inside of the print core?  I guess that's fine.  I can't think of any reason it would get hung up. 

 

How do you keep the smaller bowden from retracting into the feeder on retractions and chewing up the bowden in the feeder?

[waizman 0]

Thanks for your answers.

 

Quote

as you get farther from the glass bed, there is less bouncing up of the moving air around the nozzle so less nozzle cooling.  That's probably it.

Makes sense!

 

Quote

So your print failure might have nothing to do with temperature.

Dunno. I was the only thing I saw changing while the print failed. Needs more testing.

 

Quote

So that could be it.  You can kind of tell because sometimes you can see the PLA has cracks in it when looking through the bowden.  And when there is a retraction, sometimes you see where the crack is closest to the feeder which could be anywhere along the bowden.

The purple filament is pretty new. Maybe a week since I opened it. I live in a pretty dry place (Eilat) and the humidity level usually stay under 30%.

It is generic, but the other colors from the same brand work great.

During prints it doesn't break and no cracks are visible. It's not "crunchy" either and cannot be broken by bending it.

Very weird filament...

 

Quote

So wait - you are using 1.75mm pla through smaller bowdens.  Does the bowden go down inside of the print core?  I guess that's fine.  I can't think of any reason it would get hung up. 

 

How do you keep the smaller bowden from retracting into the feeder on retractions and chewing up the bowden in the feeder?

I don't. I use the original 6.35mm bowden tubes. BTW I  tried replacing them with new ones and found out mine were too small (6mm) 🙈
Also tried to insert a 3OD/2ID bowden in the original one and it got sucked in by the retractions as you mentioned.

I was considering gluing it inside the original one using superglue, but I don't know if it will give me any improvements and I don't wanna ruin the bowden...

Any thoughts?

[gr5 2,176]

3 hours ago, waizman said:

I live in a pretty dry place (Eilat) and the humidity level usually stay under 30%.

In my experience, petg gets a little worse as it absorbs water.  Nylon and PVA get MUCH worse as they absorb water.  PLA  and ABS don't care. I don't keep my PLA dry.  It can be in a humid environment for years and no big deal.  I have 7 year old PLA and ABS and they don't seem to need any drying or anything.  Unlike petg.

 

3 hours ago, waizman said:

Any thoughts?

Yeah, print 2.85mm aka "3mm" filament.😜

 

I don't understand how you are printing with 1.75 then.  You did zero modifications?  3dsolex.com used to sell a conversion kit.  I'm not sure that they still do.  It involved a different printcore and a bowden-in-a-bowden like you describe (3OD/2ID).  And a 3d printed "plug" that you put in the feeder end to keep the inner from getting sucked into the feeder.

 

This solution only worked in the left core because the right core goes up and down so the inner bowden needs to stretch/shrink.  I know someone who solved this with a 3OD/2ID spring that could compress when the right core went up.

 

 

[waizman 0]

I inserted a 3OD/2ID to the printcore from the top all the way to the nozzle.

Tried a few sections of PTFE tube but the transitions between them always caused a jam.
So took it all apart, removed any PTFE tube that was there, and just put a single one from top to bottom.

The extruders and bowden got no mod. I'm testing a tube-within-a-tube now (4OD/2ID inside a 6OD/4ID tube).

Quote

Yeah, print 2.85mm aka "3mm" filament.😜

NEVERRRRRRRR
It's just so random to have a specific diameter just for Ultimaker... It's a pain in the a$$ to get, and it's always very basic colors and types. Here in Israel at least. Tried 1.75mm nozzles and heatsinks for the printcore - they all work worse then the original 3mm stuff. So a simple 1.75mm conversion it is!

Both sides are converted btw. 2nd extruder actually work better than the 1st.

 

Quote

In my experience, petg gets a little worse as it absorbs water.  Nylon and PVA get MUCH worse as they absorb water.  PLA  and ABS don't care. I don't keep my PLA dry.  It can be in a humid environment for years and no big deal.  I have 7 year old PLA and ABS and they don't seem to need any drying or anything.  Unlike petg.

I agree.

[gr5 2,176]

53 minutes ago, waizman said:

So took it all apart, removed any PTFE tube that was there, and just put a single one from top to bottom.

What keeps the thinner tube from moving upwards during retractions?  Is it in there nice and tight somehow?  Oh - is the aluminum heatsink holding it tight?  Or is it even passing through the heat break?  Is the heat break holding it from retracting upwards?

 

A little history...

 

People have tried to make all ptfe or all metal hot ends.  It's difficult for PLA because it sticks to all known metals and ptfe gets soft at 220C and even 200C after 500 hours at temp and loses it's mechanical ability to not slump/crush the filament when under a little pressure.

 

For metal hot ends there is some region where the pla is only barely above melting point and when it freezes it can stick like hell to the metal around it and then your print fails part way through.  Especially with retractions you are pulling heat upwards when you retract (potentially) although an ideal retraction should not pull anything out of the print core - it should simply release pressure.

 

Anyway Ultimaker's solution is the heat break where the PLA will indeed get stuck in the heat break occasionally but it is only 2mm long and the feeder can just break the plastic there.  The teflon is just above the heat break where the temp rarely gets above 120C such that the teflon can last years.  Sometimes the temp gets high enough at the teflon for the PLA to stick to it but nothing really sticks to teflon.

 

Back to your issue

 

If instead of teflon you have a PFA tube (I suspect your thin "bowden" is made from PFA), the pla can stick to it occasionally (especially if you have lots of retractions and especially if the retractions are pulling the filament upwards in the print core (which they shouldn't)).  You can buy both PFA and teflon tubing.  PFA is transparent.  Teflon will probably work better for your situation but you'll probably have to replace it quite often - maybe every 100 hours of printing.  Plus it's so damn slippery you might want to 3d print something to keep the teflon from getting retracted.

 

The design of a printcore is complex and there were over 100 iterations before the testing people were happy.  Some of the problems were not noticeable unless you printed over 100 prints with the printcore.  They would have 10 successful prints and thought they had the final design.  And then more issues came up.

 

 

 

[gr5 2,176]

Actually another "solution" might be to reduce retraction amount.  for you in particular I think it's important not to let the filament EVER go upwards in the tube as it can then stick to that inner tube of yours.  Once at the end of the print is okay as that may cause a clog at the very start of a print but having a clog half way through the print is much worse.

[waizman 0]

Sorry for the delayed reply, I was out of town and far from the printer and PC.

 

On 3/19/2024 at 4:10 PM, gr5 said:

What keeps the thinner tube from moving upwards during retractions?  Is it in there nice and tight somehow?  Oh - is the aluminum heatsink holding it tight?  Or is it even passing through the heat break?  Is the heat break holding it from retracting upwards?

It just doesn't. Maybe due to friction? It is a tight fit mostly.

 

On 3/19/2024 at 4:10 PM, gr5 said:

Back to your issue

 

If instead of teflon you have a PFA tube (I suspect your thin "bowden" is made from PFA), the pla can stick to it occasionally (especially if you have lots of retractions and especially if the retractions are pulling the filament upwards in the print core (which they shouldn't)).  You can buy both PFA and teflon tubing.  PFA is transparent.  Teflon will probably work better for your situation but you'll probably have to replace it quite often - maybe every 100 hours of printing.  Plus it's so damn slippery you might want to 3d print something to keep the teflon from getting retracted.

 

I use a white PTFE tube as an insert. Not a clear PFA one. At least that what AliExpress says it is...

 

On 3/19/2024 at 4:10 PM, gr5 said:

The design of a printcore is complex and there were over 100 iterations before the testing people were happy.  Some of the problems were not noticeable unless you printed over 100 prints with the printcore.  They would have 10 successful prints and thought they had the final design.  And then more issues came up.

 

I dunno what changed so much, it's a pretty simple design, and I changed some parts of it with 3rd part AliExpress parts that are not 100% and it operates the same as far as I can see.

 

As for the heat break you've mentioned - where is it? That's the area just before the nozzle?

I got so many problems there, So now I'm just using a longer tube and push it all the way inside the nozzle.

Tried 1.75mm nozzles too but the 3mm are somehow better...

Added some pics to show you how I converted the printcore.

It's orange because the original black parts got broken and I found that it is not getting too warm so it is now printed in CPE and works great. And the tube got replaced with a new one, don't worry 😉

 

[waizman 0]

Some more updates - I figured out that my problem was with the 1st feeder original lever.

It was not puttying enough pressure on the filament or something. Which is weird because the 2nd one is great.

As previously mentioned -

Quote

Both sides are converted btw. 2nd extruder actually work better than the 1st.

The 2nd feeder have a "Lazy Lever" (UM2+ Lazy Lever by sneakypoo - Thingiverse) and the 1st one had an Original one.

This is because when trying to change the 1st feeder's lever to a lazy one, it failed to print almost every time.

 

Something got me to suspect the lever again and I printed this one (1.75MM CONDUCTIVE FILAMENT FEEDER for ULTIMAKER by osdoyi - Thingiverse) instead. And it now works PERFECTLY!

Three issues though -

The filament is VERY hard to insert into the feeder. Almost no play with the new angle, even when on lowest pressure setting.

There's no more 2.85mm option as the 1.75mm barely goes in now.

The wedges (WedgeBot/TheWedge) no longer work.

 

So I sent a message to the Lazy Lever designer asking for a mod that will make the lever more like the other one but with the lazy part. Unfortunately I'm no great modeler and don't know how to do it myself.

 

So long story short - now it prints great and FAST! With all the friction in the world holding the filament tight, I can now print at speeds of 150-200mm @ 0.2 layer height. Before 70-80 was the max. 

[gr5 2,176]

This is great.  I may try this myself.  I have some 1.75mm filament to use up and it's so simple.

 

You aren't really using the heat break feature.  It keeps the ptfe around 100C and cooler but for you it is right in the nozzle at ~200C and so it will degrade faster.  You should replace it often.  Maybe every 500 hours of printing?  Maybe 100 hours.  The PTFE will degrade at 200C and even faster at 220C.

 

Heat break below.  It's steel but it's not much thicker than aluminum foil and the most common place to break a print core.  Even during just regular printing, sometimes the nozzle hits the print hard enough to break the heat break.

[waizman 0]

Thanks!

Any way to use the heat break feature without having this threaded part being 1.75mm itself?

 

You only use 2.85mm?