Core-1 under extruded if Core-2 is used in the same layer

[JD-Art 1]

 

Hi. I have this weird problem where Core-1 under extruded if Core-2 is used in the same layer. (it looks like under extrusion!) It's not a problem that's been there before. But I'm not entirely sure when it started, because I have been solo printing on core-1 for some time.

 

(Edit: As mentioned below, sorry for not being more clear about this but the printer in question is a Ultimaker s5)

 

 

 

 

 

Core-1 PLA, Core-2 PLA (Core-2 PLA as support)

 

 

Core-1 PLA, Core-2 PVA (PVA suport only on "Support Interface" )

 

 

Core-1 PLA, Core-2 PVA (Cura 4.13)

 

Core-1 PLA, Core-2 PVA (Cura 3.5)

 

 

What i have tried:

Rod alignment: OK
Core alignment: OK

 

Cura 3.5 : No difference (I know for sure that this Cura version and profile have printet before without the problem)
Cura 4.13 : No difference
Cura Arachne2 : No difference

 

Print size has no effect, and distance between the prints also doesn't affect the problem. Prime tower don't help either.

 

 

Any ideas ?

 

 

Edited April 4, 2022 by JD-Art

[JD-Art 1]

No one has a comment for this ?

[JD-Art 1]

Hi I do still need some help with this problem. Do i need to describe it differently or is it in the wrong category of the forum?

[GregValiant 1,251]

Is it a UM printer(?) or another manufacturer?  If you use the File / Save Project command and post the 3mf file someone should take a look.  

[JD-Art 1]

Hi GregValiant

It's a Ultimaker s5, I have set it as a tag. but yes in hindsight i should have specified that in the text also.

 

Here's the test print. I'm not sure it will be of much help though. it is just three squares on top of each other to force a layer with support/PVA

 

UMS5_test-print-JD040422.3mf

[GregValiant 1,251]

I'm not a UM guy but I thought maybe something might jump out of your settings - but nothing did.

@Smithy, @gr5, or @Torgeir may have takes on this.

 

(In three of the 4 images that portion of the right model that is adjacent to the support almost looks like it is also going down at 2X layer height.)

[gr5 2,172]

Thanks for posting the 3mf project file.  I looked it over and sliced it in cura (had to uncheck the preference "automatically drop models to the build plate") and sliced it and it all makes sense (models and slicing agree with your pictures that have a thick pva infill layer).

 

So the quick answer is you need to check enable prime tower setting.  I'm not sure it is underextruded but while it is printing with the pva core the left core leaks a bit.  After it is done with the right core and goes back to the left it is a bit emptied out and takes a bit of time to get back up to pressure.  A prime tower's purpose is to get the pressure back up so your left core is primed with 1/10th second of when it starts printing with PLA again.

 

I examined the gcodes in the web page gcode.ws and looked at extrusion amounts and the PLA extrusion is consistent throughout.  So nothing wrong with the gcodes.

 

However I noticed that the support overextrudes by 2X.  I see you have layer height overall to 0.15 but the support layer height is 0.3mm.  Somehow this confuses cura and it extrudes PVA as though layer height is 0.3mm (even though the support layer is still 0.15mm).

 

So I would fix that as well.

[JD-Art 1]

Hi @gr5, I think we might be talking past one another, or maybe I'm just not reading your post right.

My problem is not so much the PVA but the part on the tall tower that's printed beside the PVA (yes the PVA is ugly to i know)

 

 

I have tried with the prime tower and on large prints it's the same behavior over the whole layer. so dont think its a cold or leaking nozzle.

"I can be wrong", but the problem in my opinion is to uniform, the layers are almost perfectly wrong. If it was a cold or dripping nozzle it would be more random right?"

 

 

 

 

Fresh print with "my" profile, yes it's not perfect but it gives me the best prints most of the times. (the pva part godt knocked off so didden finish, but you can still see the problem part on the right tower)

UMS5_test-print-JDAccuracy-JD050422.3mf

 

 

My PVA is really shitty today. I don't know if the air is too moist or something. but that's a different problem.

Though the print is ugly you can still clearly see the problem om the right tower. The line at the top "arrow" is just from a shift in the print and not an issue

 

Fresh print with the normal Cura Visual profile (common ground to make sure its not some random setting i have turned on or off by accident)

UMS5_test-CuraVisual-JD050422.3mf

 

 

Changes:
drop models to build plate: off
Prime tower: on (standard settings)
Support: on
   support extruder: extruder 2
  horizontal expansion: 0.0
support infill layer thickness: 0.15

 

 

 

@gr5 Will you try to set the print profile as you think the settings should be, then I will print it, to see if it makes a difference.
If you have access to a different slicer then Cure I'm willing to try that as well.

 

 

 

[gr5 2,172]

So if you play through on the preview screen you can see there isn't much difference in what is happening on those pva layers.  Whether pva or not it does:

 

left tower, then right tower, then moves up a level, continues right tower, then left tower then moves up a level (then repeats the left tower).  So it does two layers in a row then switches to the other tower and does 2 in a row.

 

Also when it starts the right tower it always starts in the same corner and starts with the inner walls, then the outer wall, then the infill.  When repeating a layer it does the same thing.

 

head play or alternatively belt play

 

Okay - I just deleted this section.  If it was this the pattern would only show on front/back and not the sides so am going to skip to my next theory because I can see in one of your photos that the pattern goes all the way around.

 

 

 

z-axis play

 

Another theory is that something is really off with the Z axis.  When you only use one core, during most of the print, the bed only goes down.  Never up.  So backlash is less of an issue.  But with two cores it has to move the bed down (away from the core) while it does the switch and then brings the bed back up.  So if there is any play in the Z axis it will fall farther than normal  (or less than normal) when it moves the bed down to switch layers.

 

Normally z-axis play is caused by dirt on the Z screw but that usually gives you very irregular amounts of play so you get thicker and then thinner layers in a random pattern.  Your pattern is very regular.

 

I don't know how to diagnose and fix this.  I would recommend replacing both the z screw and the z nut.  And while it's apart I would check how well the 2 Z bearings slide on the 2 Z rods  (does it get stuck?  Does it slide smoothly?).  A bad Z nut can definitely cause a LOT of z-axis play.

 

I guess the best way to diagnose this would be to put a brick or hand-weight (something around 10 pounds/5 kilos) on the back of the bed.  If the pattern gets much better than it's z-axis play.

 

 

By the way PVA must be keep crazy dry.  Buy some color changing dessicant (get a liter) and put 1/4 of that in some container with tiny holes and put that and your spool in the same huge ziplock.  Never leave the PVA on the printer over night.  If you do, you can recharge the filament on the print bed at 70C (but slightly above the bed spaced with some PVA as the spool itself can warp)  with a towel over it overnight (or preferably for 2 days).  8 hours only dries the outer bit of the spool but if you only left it out in air for a day then you probably only "wrecked" the outer bit of the spool.

[JD-Art 1]

Hi @gr5 Thanks for the analysis.

 

19 hours ago, gr5 said:

head play or alternatively belt play

 

 

Okay - I just deleted this section.  If it was this the pattern would only show on front/back and not the sides so am going to skip to my next theory because I can see in one of your photos that the pattern goes all the way around.

 

I have tested the rod alignment and belt play. And as you mention if this was the problem, the problem would properly be more in one direction and not so uniform over the whole layer.

 

 

19 hours ago, gr5 said:

z-axis play

 

I guess the best way to diagnose this would be to put a brick or hand-weight (something around 10 pounds/5 kilos) on the back of the bed.  If the pattern gets much better than it's z-axis play.

 

I don't have a 5kg weight that I can easily put on the build plate. but I have made this set up with a 10mm Z-hop on travel to mimic the changing core movement. If it was do to a bad z-axis this test print should have the "ring/stripe" effect all the way up right?

UMS5_Z-Hop.3mf

 

 

The print in this way is okay, there's some small imperfections in the layers but they are now random and more likely to be do to leaking nozzle or variation in the spool diameter.

 

So I dont think its the z-axes. if you still think i should try and test it with the weight i'll do that.

 

 

19 hours ago, gr5 said:

By the way PVA must be keep crazy dry.

I know 🙂 Already have a Polybox. But for some reason It just acted up more than usual yesterday

 

Edited April 6, 2022 by JD-Art

[gr5 2,172]

Wow - great experiment.

 

Okay so it's not the Z axis.

 

Edit the gcode and try adding a T1 and T0 every other layer?  I'm not certain but I think that will switch to the other core and then immediately back to the first core.  Examine your working gcode used for the pva first to make sure it does indeed use T1 and T0.

 

To get to the gcode you can either tell cura to save as gcode or you can rename the .ufp file to .zip and then open that and it's not too hard to find the gcode file inside there.  Really it's probably best (easiest) to just export as gcode file instead of ufp.

 

I guess another issue is temperature.  It cools the non-active head while the other head is in use.  But it waits until the head is back to temp so I don't see how this could be it but you could test T1 and T0 alone, and then if that looks fine add in the temperature changes (to the left head only) to see if that does it.

[JD-Art 1]

20 hours ago, gr5 said:

Edit the gcode and try adding a T1 and T0 every other layer?  I'm not certain but I think that will switch to the other core and then immediately back to the first core.  Examine your working gcode used for the pva first to make sure it does indeed use T1 and T0.

 

My knowledge to gcode is limetet to what i can follow a totorial on. So i dont realy understand what you wont me to try?

 

20 hours ago, gr5 said:

I guess another issue is temperature.  It cools the non-active head while the other head is in use.  But it waits until the head is back to temp so I don't see how this could be it but you could test T1 and T0 alone, and then if that looks fine add in the temperature changes (to the left head only) to see if that does it.

 

I made a temperature tower test and it looks fine, i also tried printing the model with a delay of 20 sek on every layer and ditto make a difference idher. But i'm not sure if this is the right way to test what your saying.

 

 

I have tried cutting one of the models in half, and it actually looks like the proplem section are "skipping" every second layer. even though they have the same layer height in Cura. Is this possible, that the printer somehow skips a layer. and how can i test this?

 

 

 

I have tried switching AA and BB core so that prime core AA now is in the extruder 2 position and BB in extruder 1.

 

 

This print is more or less identical with the first. and the problem is in the same place and to the same degree. So I don't know what the conclusion of this test is. Maybe that if it was a mechanical or extruder problem they should have been different.

 

Since there's a difference in what Cura shows and what gets printed can it be a firmware problem?

 

 

 

 

 

 

[GregValiant 1,251]

@JD-Art said "I have tried cutting one of the models in half, and it actually looks like the problem section are "skipping" every second layer."  (BTW - Nice job slicing the model.)

I said "...the right model that is adjacent to the support almost looks like it is also going down at 2X layer height."

 

When I sliced your 3mf file - the gcode it produced did not show that .30 layer height in the right side print.  All the layers on the right print were at .15.  The support Outer Wall layers on the left print were at .15 and the support infill on the left print was at .30.  That's how it appeared to be set up in Cura, and when I read the gcode into AutoCad that's how the gcode file prints and it is exactly the same as the Cura preview shows.

 

"Since there's a difference in what Cura shows and what gets printed can it be a firmware problem?"

Once again, I'm not a UM guy but that would seem to be a good question.  Are you printing from a UFP file?  Can you print a plain gcode file on the S5?  Is there anything between the slicing and plastic coming out of the extruder (Octoprint, Digital Factory, something that could have an effect on the printer output but down stream from Cura).  If you can print a plain gcode file it may give a further clue.

I can print and emulate gcode files but not UFP files.  When I open a UFP file with Cura the preview is correct.

 

Some of those images sure look like the printer is skipping the outer walls every other layer.

 

I made up this 3mf file from your 3mf file.  The only difference is the model itself (which is my own "support calibration" shape.  It's all one piece rather than one instance of a model suspended over a second instance of the same model.

GV UMS5_test.3mf

Edited April 7, 2022 by GregValiant

[gr5 2,172]

I'm pretty sure it's not the firmware.  The firmware is extremely reliable.

 

So your part is definitely not skipping every other layer.  It would be much more severe.  Those holes are "underextruded" layers because the Z moved too far.

 

Could you verify that each ridge is 0.3mm apart?  If they are .6mm apart that would tell us that something is very much wrong.

 

I'm going to look at the gcode again.

 

Did you ever remove the 0.3mm layer height in the support section?  That still bothers me.  a lot!  It's not good to have different layer heights in the same layers and if nothing else it seems to be doubly extruding the pva layers.  I don't think this is your issue but it might be.

 

I'll get back to you later.  Too busy right now.

 

 

 

[JD-Art 1]

 

Took some time fixing this first!!

 

 

On 4/7/2022 at 2:27 PM, GregValiant said:

Are you printing from a UFP file?  Can you print a plain gcode file on the S5?  Is there anything between the slicing and plastic coming out of the extruder (Octoprint, Digital Factory, something that could have an effect on the printer output but down stream from Cura).  If you can print a plain gcode file it may give a further clue.

I can print and emulate gcode files but not UFP files.  When I open a UFP file with Cura the preview is correct.

I slice in Cura and use print over the network. but i will try and print from an usb to see if that makes a difference.

As far as i know you can't print gcode directly unless you put the printer into developer mode. and i'm not quite ready to go there.

 

Your T-model: Its only the top 2mm that's printed okay. but thats also the only part of the model that don't have a part with suport in the layer.

 

 

22 hours ago, gr5 said:

Could you verify that each ridge is 0.3mm apart?  If they are .6mm apart that would tell us that something is very much wrong.

 

If i set a caliper ruler* to 0.30 and look at it under a microscope i would say that the top ridgets are 0.3 apart and 0.15 on the lower part. But that is also a problem either it's skipping a layer or there is a layer so under extruded in between that it can't be seen under a microscope

*dont know if thats the right name for the tool

 

22 hours ago, gr5 said:

Did you ever remove the 0.3mm layer height in the support section?  That still bothers me.  a lot!  It's not good to have different layer heights in the same layers and if nothing else it seems to be doubly extruding the pva layers.  I don't think this is your issue but it might be.

 

Yes, in my print file (S5 test-print-Accuracy-JD050422.3mf) they are both set to 0.15.

The 0.15 and 0.30 actually comes when you use the default support settings in Cura!

 

 

 

Test print with a layer height of 0.2 same result - "from USB"

On the 0.2mm test you can see a small layer between the "big" ones so it's not skipping layers but under extruding. at least on the 0.2mm test

 

 

 

Edited April 8, 2022 by JD-Art
text part missing

[gr5 2,172]

On 4/8/2022 at 9:12 AM, JD-Art said:

 

On the 0.2mm test you can see a small layer between the "big" ones so it's not skipping layers but under extruding. at least on the 0.2mm test

Yes, some of your older picutres were really excellent and I could see it was extruding on every layer but some layers extrude a LOT more than others.

 

Did you fix that 0.3mm parameter yet?  To see if somehow (no idea how) that is the cause of the problem?  But it's just so weird to have 0.3mm for support layer height and 0.15 for regular layer height.  I don't see how that is possible.  And when I looked at the gcode it was overextruding the PVA if I remember right.  Can you please please confirm?  This is a pretty serious bug and I want to know if it is somehow related to your every-other-layer issue.

 

This is the 3rd post where I asked you to fix that.

 

 

This still looks like a Z axis play issue.  So I went back to your "hop" experiment and realize the experiment was a bad test.  I assumed it would do 2 layers of one tower then go over and do 2 of the other.  But it doesn't!  It is doing only one layer and then hopping over to the other tower and doing only one layer and then hopping back.  So it's not the same.

 

Instead print only a single tower with no hopping enabled.  Manually move the Z up ever 5th layer or so.  Search through the gcode for the letter "z" and also "layer".  There should be only one Z per layer.  Look at the layer numbers and on the 5th and 10th and 15th layers add:

G1 Z30

 

On the line before the other Z command that moves to the new layer.  Or pick an Z value at least 3mm higher than the next layer.

 

I'm really sure now.  My earlier guess was correct.  Z-play issues.  Did you ever see how dirty the Z screw looked?  Particularly near the top?  Anyway I wish you had tried that brick experiment.  You can use rocks instead.  Place them after the print is doing the first layer.

 

z-play can be caused by friction in the z bearings, or the z rods or the z screw or the z nut.  Or dirt in the z screw.  The Z nut is the most common problem but sometimes the 2 z rods are not parallel and things are binding and it takes a lot of friction to move the bed up or down.  You can test this by lifting the bed by hand yourself when the power is off.  You have to fight the stepper but you can do it.  It's a similar force as lifting the entire printer.  Lift from the bed near the back of the printer so you aren't tilting the bed a lot.

 

 

[gr5 2,172]

Anyway, if you are able to prove my theory (by adding z hops or 3kg of rocks) and if your printer is new you can then call your reseller (if it's still under warranty) with a successful diagnosis.

[JD-Art 1]

16 hours ago, gr5 said:

Did you fix that 0.3mm parameter yet? 

 

This is the 3rd post where I asked you to fix that.

 

Yes. I have changed that, sorry if that was not clear in previous posts.

I do, however, have a remark on this. when i pick Curas visual-normal-0.15 profile and click yes to support i get this mismatch by default. And as far as i know i can't override the default settings. So is this a mistake in the Cura profile!

 

 

16 hours ago, gr5 said:

Did you ever see how dirty the Z screw looked?  Particularly near the top?

 

The screw is "clean" it's not spotless because there grease on it.

 

 

 

16 hours ago, gr5 said:

 The Z nut is the most common problem but sometimes the 2 z rods are not parallel and things are binding and it takes a lot of friction to move the bed up or down.  You can test this by lifting the bed by hand yourself when the power is off.

 

The bed moves freely. in addition to the resistance from steppers. there's no uneven movement.

 

 

17 hours ago, gr5 said:

Anyway, if you are able to prove my theory (by adding z hops or 3kg of rocks)

 

4 * 75 g = 3 kg weight on the build plate

UMS5_Visual-120422.3mf This is the default Cura settings for Visual 0.15 profile with support. the only thing i changed is setting support infill layer thickness to 0.15 instead of 0.3 and support horizontal expansion to 0.0

 

 

The first layers were pretty bad and the model had a hard time sticking to the buildplade. There were some new and ominous sounds from the printer. But as you see the problem is still there. The weight on the bed ditten make any difference.

 

Tower to the right is with the 3kg weight on the buildplade, and the one on the left is witout.

 

 

 

 

[gr5 2,172]

Please try to print the attached file.  It's just your cube, 10mm on a side, but I added 3 moves in Z direction that are similar to what the moves look like when you do PVA support.  All it does is move the Z up 2mm (and then back down by 2mm) at the start of 3 of the layers.  All 3 layers are in the first 5mm of the cube (the first half).  So there should be 3 innie/outie lines where I did the 3 z moves.

 

There is another oddity but one thing at a time.  You can stop the print after it has printed the lower 5mm if you want.  It's a 15 minute print but you only have to do the first half.

temp1.gcode

[JD-Art 1]

@gr5 Sorry for the waiting time, but i have been down with influenza.

 

I have tried printing your g-code. And the print looks pretty good there's not many deviations or layer fails, theres some problem at the top where the top layers start but beside that.

 

On 4/13/2022 at 4:41 AM, gr5 said:

All 3 layers are in the first 5mm of the cube (the first half).  So there should be 3 innie/outie lines where I did the 3 z moves.

 

I don't see them, if it was the z-axes. They should have been visible the same way as when i print with pva, right?

 

 

Front

 

Right

 

Back

 

Left

[gr5 2,172]

Checking that gcode...

 

It starts out at 10mm at the first layer so the 0.2mm layer gets a virtual z hop as the med moves upward just starting the first layer.

 

Then one at 0.35mm (the very next layer) so those 2 shouldn't be visible as they are the same.

 

 

Then one at: 1.55mm 2mm 2.3mm

 

I think at least 2 of those 3 are visible?  Can you measure up to see if those lines near the base are at those 3 locations?

 

Also does the filament stick out as much at these 2 or 3 locations as on your earlier models?

 

Layer height is 0.15mm so I never did it 2 layers in a row.

 

I still think the problem is Z axis related but I don't feel this completely proves it.  Maybe I should have done every other layer for like 10 layers in a row?  It takes about 10 seconds for each hop edit by hand.  So I could do 10 layers in just a few minutes I suppose...

 

 

[gr5 2,172]

Or maybe all those lines are below 1.5mm?  You tell me.

[JD-Art 1]

There might be a small layer imperfection at 0.5 mm but it might as well be due to squicing at the bottom.

Beside that its pretty uniform all the way to the top.

 

if the 0.35mm is the one i see as "0.5mm" thers a smal changes there.

But i cant se any diffence at 1.55, 2,0 and 2,3 mm

 

[gr5 2,172]

Wow.  wow.  That does look good from 1 to 3mm.  What the heck is going on near the top?  I didn't do anything near the top.

 

Anyway: okay, it doesn't look like a Z issue.

 

All I can think of is temperature.   If you know the IP address of your printer you can connect to it with any web browser and it has a webpage.  There is a link to the temperature graph.  You can watch the temperature while it prints.  Disable everything except things related to the PLA nozzle.  You will know when it gets to the pva section because the temp will go up and down a lot.  You can tell when it is printing by listening and looking at the graph to see how far off the temp is when it starts the layer and how long until it gets back to the correct temperature.

 

That's all I can think of.

[JD-Art 1]

@gr5 Can I get you to import the models from my .3mf file and only the modles and try to slice them at what either settings you think is good for a model with PVA suport and save it as a g-code not .3mf

Becaus then if the print from you g-code is good, i know that its either my Cura or my printers interpretation of the .3mf fil (not sure if that is possibole, but im running out off ideers)

If the print is bad, then its most likely a problem on my printer, either mecanical or firmware. And not a setting proplem.

 

I will try and monitor the temperature graph, but not entirely sure what to look for.

 

Towers.3mf