Looks exactly like what I would expect with shrinking effects. When you have a long part it shrinks more at the ends than two short parts.
You can compensate for this all in the model by measuring the final sizes accurately and increasing or decreasing only those sides that need it.
This is getting interesting... :blink:
First Q&A:
It might be some sort of cooling/shrinking effect due to the larger volume of plastic making the flat surface at that height. Might also be partial under-extrusion due to the head oozing on travel moves associated with that flat surface.
Are you printing infill faster than perimeters? You might also try disabling combing, and see if that makes a difference.
The infill is printed with the same speed as the perimeters, with 100mm/s (I wrote 50mm/s in my original post by mistake). I haven't yet tried to deactivate combing, but it's on the list (see also below)...
Looks exactly like what I would expect with shrinking effects. When you have a long part it shrinks more at the ends than two short parts.
You can compensate for this all in the model by measuring the final sizes accurately and increasing or decreasing only those sides that need it.
Just to make sure we're talking of the same side supposed to be (more) affected by shrink: If you print a cuboid, let's say x=10mm, y=30mm and z=15mm, which dimension would you expect to shrink (more)? The effect I see let would let the x-dimension decrease, but just in the upper 7.5mm...
And it's getting even better:
- I rotated the part by 180°C around z. The deplacement stays on the same side of the print, but is now on the opposite side concerning the printer (which cancels any theory about cooling from the left side only).
- I mirrored the part along the x axis (the displacement occured on the minus-x-wall). After that, it occured still on the minus-x-wall (still the upper half). The plus-x-wall shows no sign of deplacement.
- I rotated the part by 90°C (once clockwise, once counterclockwise). There was no deplacement at all! I have to mention that the rotated part were at dfferent places on the print bed than the original print.
I'm currently thinking of some strange coincidence of retraction-fault (place-dependent?) and/or some kind of uni-directional backlash effect.
I'll do some more tests such as compare the results when printing at different places on the bed, deactivation of combing, etc. I will also have a closer look at the extruder and Bowden tube as my filament has a thickness between 2.90mm and 2.97mm and therefore quite some friction in the Bowden tube (this might be the time to try the oil trick...) Using some other (thinner) filament might be worth a try too.
Finally there should be a test on the other extruder as my Ultimaker can play in stereo... :smile:
But further ideas are still welcome! Thanks so far!
The Y=30 dimension will shrink the most. That long 30mm length of filament on the walls is pulling, pulling and it pulls the ends inwards towards the center. The Z dimension is always perfect because layers have long ago cooled before the current layer is being applied so you don't notice dimensional errors in Z. Z is always perfect or close to it.
The pulling probably occurs within seconds - probalby within 5 seconds after laying down a layer. The layer above is placed towards the outside of the layer below and quickly shrinks. I could be wrong about the timing - maybe it takes a few minutes.
ABS has a much higher temperature coefficient. In other words it shrinks much more than PLA. People who print with ABS (not me)... Some of them prefer heated chamber so that there is less shrinkage happening until the print is done. Then they open up the chamber and the whole part shrinks at the same time after it is done printing.
It's not clear to me at what temperatures most shrinking occurs (above glass temp? below glass temp?) I wish I knew how to measure that and could better understand how shrinking works. I guess I'll have to try another experiment...
In the meantime this is becoming a case for CSI Ultimaker I guess...
Yesterday, I did a few times the same print at different locations on the print bed. It turned out the effect ist strongest on the right side of the print bed and weaker on the left side. It also gets stronger towards the back of the bed.
I've scaled the model in z to 2/3 in order to check if the displacement also changes height. It stays relative to the model, i.e. it was happening at 2/3 of the original height. This is actually the only thing that is clear to me so far: it must have something to do with the rectraction. The deplacement starts on the first layer where the first retraction with a significant move in between takes place between the larger and the smaller structure of the print. And it seems only to happen when the retraction goes from the small structure to the larger one (underextrusion-theory).
As suggested by Illuminarti, I also deactivated combing and set the minimum travel distance for retraction to 5mm. It looked exactly the same as with coming.
Next test was to leave z dimension as it is and to scale x dimension by 2 and y by 0.5. The size of the displacement stayed about the same (no scaling).
I then mirrored the print along the y axis. What I got was a (smaller) displacement at the small structure and no displacement at the large structure.
Next try was to print the model on the second extruder with the same filament. The result was the very same as for extruder one. Thus, it's nothing related to a single extruder.
The final test was exchanging the filament to Colorfabb which is known to have a diamenter between 2.80mm and 2.85mm (i.e. reasonably thin to fit through the Bowden tube). The displacement was even more pronounced in this print.
When looking a this very distinct pattern of this last print I decided to check what's really different on the layers with displacement. I found that on these layers Cura, first lets print the shell of the small structure, then the shell of the large structure, then the infill of the small structure and finally the infill of the large structure.
Below the height the displacement starts, there is only one shell for both structures. On these few layers above the first displacement zone and below the second displacement zone, it first prints the shell of the larger structure and then the shell of the smaller structure.
I'm beginning to wonder if my Ultimaker does something which really no other does (bad airflow?) or if this is an issue that will occur on every Ultimaker. So if someone else would like to print this model, I would appreciate it. Please just print the part no. 2 exactly in the way Cura aligns it.
I don't really have much to contribute but I will say that I've seen the same effect on my prints and it always happens on a transition between a uniform layer to an "uneven" layer like in your model. I always figured it was due to a combination of slight inaccuracies in the rapid move and shrinkage since a solid layer is often laid down where it happens (solid layers shrink more than infill layers). I never bothered to do anything about it though because it was never as severe as in your model.
Retraction may start where the shrinkage changes but several other things also start there. For example more "top" infill. And an air gap. Obviously shrinkage factors won't pull across an air gap.
I'm pretty sure all FDM printers will do what you see. I'm pretty sure it is 100% caused by shrinkage and not over or under extrusion. But shrinkage is a bit complicated because it is shrinking while printing.
It's not clear to me at what temperatures most shrinking occurs
I since learned that shrinking is very linear at all temps. In other words if you graph PLA density on Y axis versus temp on the X axis it is a relatively straight line. Also PLA in injection molding only shrinks .3% even though the shrinkage from 200C to 20C is over 1%. .3% agrees mostly with a change in temp from around 70C to 20C which means most of the shrinking issues start only when PLA gets to around 70C or colder. Above that temp it's close enough to a liquid to not be a issue.
So a more interesting test for this piece would be to print it in either a heated chamber or at least on a heated bed at 70C.
Also like Robert says, look to see if Cura is putting in some solid layers - solid layers are going to pull much harder than open infill at 20%.
I would love to help print this part because this is an interesting topic for me but I am away from my printers until Monday night.
I don't really have much to contribute but I will say that I've seen the same effect on my prints and it always happens on a transition between a uniform layer to an "uneven" layer like in your model. I always figured it was due to a combination of slight inaccuracies in the rapid move and shrinkage since a solid layer is often laid down where it happens (solid layers shrink more than infill layers). I never bothered to do anything about it though because it was never as severe as in your model.
Thank you for confirming the general existance of this effect! I guess this model shows it much better than other models such as figurines or similar... If it's possible it would be interested to know if in your prints the effects is stronger on the fan side (assuming you only had one from one side...) I have the impression it is strongest on that side...
[...]
I would love to help print this part because this is an interesting topic for me but I am away from my printers until Monday night.
Uuh...you're going through a very hard time then... without any printer... :smile:
No, seriously... as I understand, you have both Ultimaker models, correct? If your original Ultimaker model still has one fan it would be a very interesting comparison between the two...
I'm right now doing some more tests. Just for really, really, really making sure Cura is not responsible, I sliced the part with Slic3r and got the very same result. Doing the infill before the shell (possible in Slic3r, not in Cura) didn't change anything either. The same goes for adding 0.06mm filament after each retraction (equal to about 10mm printed filament at 0.1mm layer height). It just gives more stringing and over-extrusion.
Right now, I'm trying the print with a solid infill every 5 layers in Slicer. After that I think I'll do a 100% infill print.
BTW: There is no difference between 0.8mm and 1.2mm shell thickness...
Some additional information:
- The print with solid infill every 5 layers looks the same as all the others: deplacement begins a the height where part of the model gets a solid infill
- The print with 100% solid infill also has a deplacement. But it begins where the model gets narrower.
You have a UM Original right? No heated bed?
Yes to the first and no to the second (actually not yet, the parts are waiting to be assembled and some of them to be manufactured)...
I printed the same thing in September, with Cura 13.6.4, 0.1mm layer height, on an Ultimaker Original without a heated bed. I don't remember the exact amount of infill, but I think it was about 20%. Temperature was probably 210°C and speed 75mm/s, as these are my standard settings.
I can see a slight change in the surface pattern and something like a seam at the height where the geometry changes,but not a displacement as in your print.
Maybe your problem is caused by backlash. When the geometry of the part changes, the path of the print head has to change too, and with a different path, backlash can have a different effect. I strongly suggest to use belt tensioners for both, the long and the short belts, to get rid of backlash. Another sign for backlash would be, that small holes of around 6mm diameter are not printed round.
I'm with znib on this - after all the tests, I think it's backlash related somehow due to the differing geometry causing a different toolpath that shows the problem.
I may try to print it later as well, if only because it looks like an interesting puzzle.
[...]
Maybe your problem is caused by backlash. When the geometry of the part changes, the path of the print head has to change too, and with a different path, backlash can have a different effect. I strongly suggest to use belt tensioners for both, the long and the short belts, to get rid of backlash. Another sign for backlash would be, that small holes of around 6mm diameter are not printed round.
I thought of that too. Indeed I had some non-round holes in the past and still got sometimes small spaces between shell and infill (only at some spots, not everywhere). However I would expect backlash to show up as well when rotating the print by 90°...
I have belt tensioners for the long belts. They are really tight and make exactly the sound the should when picked (as shown in the Ultimaker video). So far I haven't seen a good solution for the short belts and in the middle to long term I plan to implement a direct drive. I think the short belts are not an ideal design.
I use Jeremie's http://www.thingiverse.com/thing:34785, and haven't had any problems. Direct drive might be marginally better, but I'm not really convinced - it's certainly overkill if you haven't just tried the simpler alternative first, I think. Definitely worth trying - short belt tension is critical to good print results.
I use the same short belt tensioners as illuminarti. First I only installed belt tensioners for the long belts, but that wasn't enough. Without any belt tensioners, holes were printed oval. With the long belt tensioners they were a bit rounder, but now they looked a bit like squares with round corners. Since I also installed the short belt tensioners they are almost completely round (round to the eye, with a caliper I can messure that the diameter still varies by about 0.1mm from the widest to the smallest).
Although I can only 'see' the non-roundness of the holes with calipers it's certainly worth a try. And it will feel good to print something else again than just the same puzzle piece over and over again... :roll:
Thanks for the moment! I'll give you feedback as soon as the belt tensioners are in place...
... short belt tension is critical to good print results.
Maybe Ultimaker gets rid of the short belts for the third generation printer...? Does the UM2 have short belt tensioners included?
:angry: This is not fair...
After having installed belt tensioners for both (x and y) short belts, I get wonderful prints in really high resolution - with the same crappy step!!!
As I had to remove some cables I even decided to separate the x motor cable from the extruder cables in order to eliminate the extremly unlikely possibility of cross-talk between the two... as you can see, I'm slowly running out of ideas...
While this print was running I listened to the printer. I had the impression (very biased!) that the z step sounded differently at exactly the height where the troubles begin. It was kind of smeared, longer than the usual 'zac'... Letting the z stage make 0.1mm steps brought no real news. The steps do not sound the same for one rotation. But none of them really sounded as strange as during the print.
Any other ideas?
I still maintain this is a cooling/shrinking issue. Try printing it hollow. Don't worry about support - just do it hollow. I will print this part with a cold and hot bed on my UM2 tomorrow.
I had the impression (very biased!) that the z step sounded differently at exactly the height where the troubles begin. It was kind of smeared, longer than the usual 'zac'...
Interesting.. so maybe your problem is z-stage related? What happens if you set "Cut off object bottom" in Cura to about 4mm, does the displacement happen closer to the top of the part, or still where the geometry changes?
Which version of Cura are you using? You mentioned you used Slicer to rule out a Cura bug, but if I'm not mistaken, old versions of Cura (older than 13.6 AFAIK) used the same slicing engine like Slicer. You see, I'm running you of ideas too. :grin:
This is the result of the hollow version:
The difference is that there is only one step instead of three (one right, one left and another one right).
I listened to the z stage again. No real difference audible. That one was false alarm.
@znib: I use Cura 13.12. And I have a Marlin version compiled on marlinbuilder.robotfuzz.com. Therefore I also did a quick check with the standard firmware uploaded by Cura today: same result.
Hi Dim3nsoineer,
My first guess would be to modify the model, perhaps enlarging or making it smaller (putting it on a box?) of maybe tilt it in another direction. If the position of the line should lower for a heightened print, its has got to be Z-axis related, If the line persists at a higher print, I would relook to the slicer (even thought you tried multiple slicers).
OR
I have read that you have printed in different directions with always the same side being pointing out. Having had a thread recently on warping issues, could it be that this is the first line to be laid down, and then it being to hot when the second line in the same direction is being laid down, causing some sort of a warp outwards?
I hope it is food for thought,
Cheers,
Lennart
Here is some update. In order to check / rule out the electronics, I swaped the x and the y axis, i.e. the x motor was driven by the y stage and vice versa.
The original deplacement along the long edge is significantly smaller (green arrow). However I got a new displacement in x-direction (red arrow). Thus, it's rather not the electronics as the major displacement is still in physical x direction (actually in direction of the fan).
I also measured the temperatures of both x and y motor with an infrared thermometer. While the x motor was something like 48°C warm, the value for the y motor was 55°C or even a bit higher. A quick look at the stepper drivers showed that they have indeed a different setting. Wasn't there once a page in the Ultimaker wiki showing in which direction you have to turn to increase the current? Can't find it anymore...
Wasn't there once a page in the Ultimaker wiki showing in which direction you have to turn to increase the current? Can't find it anymore...
You mean this one? http://wiki.ultimaker.com/Electronics_build_guide
ArunC posted a topic in UltiMaker Cura,
.thumb.jpeg.0b7a05eafc09add17b8338efde5852e9.jpeg)
Dustin posted a topic in Firmware,
Recommended Posts
Top Posters In This Topic
28
23
11
6
Popular Days
Dec 28
12
Dec 31
11
Jan 3
10
Jan 6
10
Top Posters In This Topic
Dim3nsioneer 28 posts
gr5 23 posts
illuminarti 11 posts
znib 6 posts
Popular Days
Dec 28 2013
12 posts
Dec 31 2013
11 posts
Jan 3 2014
10 posts
Jan 6 2014
10 posts
illuminarti 18
It might be some sort of cooling/shrinking effect due to the larger volume of plastic making the flat surface at that height. Might also be partial under-extrusion due to the head oozing on travel moves associated with that flat surface.
Are you printing infill faster than perimeters? You might also try disabling combing, and see if that makes a difference.
Link to post
Share on other sites