Did you use values from the Califlower or Calilantern? I’m trying to figure out what values to use with goskew.
Opened a brand new X1C today to find the same issue. 115mm length off by half a mm. 100mm gives 99.75. Skew test is 114.56 vs 114.72. 10mm cylinders undersized by 0.2mm. Most dimensions small by about 1%, sometimes more, but never the correct size. I printed at all wall speeds 60mm/s and nothing changed.
This is the included Bambu filament that was calibrated using a flow cube, 2x 0.42 walls gave 0.84. I also let the lidar autocalibrate and got even smaller parts. Orca slicer, tried the fitment test and the 6mm hex was 5.9mm
I’m glad to see this conversation has gotten back on track but it’s insane that $1500 flagship printers are coming out of the box with worse dimensional accuracy than my ender 3 from 5 years ago and there’s no fix.
3 Likes
Honest question. If you’re so unhappy with the dimensional accuracy of these printers why on earth would you buy another one?
3 Likes
It would be nice if the printer could do an auto calibration for shrinkage using all the built in sensors it has on the X1C… Their engineers should be able to program the machines to calibrate using some kind of multi ring calibration print to measure the movement increments and then compare that to the data captured by the lasers, lidar, and nozzle camera for each ring of the print. Larger the ring, the more shrinkage where it could measure the difference and automatically compensate per filament… But what do I know, I am just some idiot on the internet.
If they cant do that for us, they should at least implement the shrinkage feature like OrcaSlicer has. Having to manually scale the size of the model or change the hole and contour compensation settings in BambuStudio every time I print something is not going to work for me. I bought the X1C so I could have a workhorse printer that I can trust while I have my other printers to tinker with and I not expect them to be at the level of the X1C, but I find myself using my custom CoreXY just to print parts that I need to be accurate.
1 Like
Please read the entire thread. The root cause is not in the printer but in the natural shrinking of the filament when it cools down. Those shrinkage numbers are perfectly in the expected range. In my experience, PLA typically shrinks by around 0,3%, PETG by 0,4 … 0,5%.
Of course one can mistune the printer to compensate shrinking of one material and achieve perfect dimensions with e.g. PLA. You would probably get good enough results with PETG too. As soon as you use material with a different shrinking rate like ABS, Nylon, … you again have “wrong dimensions”.
So the proper way is to let the printer move the correct distance (which doesn’t need calibration) and compensate the filament shrinkage in the filament profile.
Regarding flagship printers: In my opinion this is not a high spec printer but a consumer printer with an extremely rich feature set at a budget price. At work we have an Ultimaker S5 that was like 6000€ (AMS not included) with far less bells and whistles that is really frustrating in many regards compared to the X1C.
3 Likes
P1 and X1 printers have some problem with dimensions, a simple model shows the problem (a cube with a hole at the bottom and an axis at the top)
the dimensions of the cube will be correct, but the hole and axis will be too small. I encountered this problem when printing a bearing socket and an axle that worked with this bearing. Compensation unfortunately changes the geometry of the entire model, only a compromise can be found between the size of the holes and axes and the rest of the model by adjusting the shrinkage parameter (orca slicer).
As the predecessors wrote, the same model printed on bedslinger printers prints dimensionally correctly (the size of the cube, hole and axis is proportional and can be adjusted linearly)
3 Likes
I would appreciate if anyone doubting my experience would print a calibration cross of some form with zero compensations (no shrinkage in filament settings, no resizing the model, no XY contour compensation) in PLA and check with a certified micrometer or calipers. Anything with multiple X and Y dimensions. Is it true to size?
I bought another Bambu for the AMS, there’s high demand for multicolor prints. Additionally, I purchase from a vendor with a good return policy. If I can’t find a mechanical solution to this it goes back and I’m out nothing but time.
Honest question: What is wrong with expecting a product to perform as advertised and up to par with its cheaper competition? I’ve found dozens of threads about this issue here, on reddit, and in other printing communities. None have solutions besides “just compensate in the slicer”. I don’t understand this “buy the printer and work around any elementary flaws or get out” mentality.
I did and still disagree. When using PLA PBT alloys like Polymaker PLA Pro on my Sovol the parts print dimensionally accurate and stay that way. There’s a 2 month old calibration cross on my desk that has twentieth mm accuracy, the same as it measured the day it was printed. This is with the same filament, the model was printed at 100% size with zero shrinkage compensation. Either the Sovols, Prusas, and old Ender I’ve used are all mistuned and all print big and I never noticed AND the people not compensating with their Bambu’s and getting 100% scale parts have a faulty printer… or some Bambus print small.
I print filled nylon all the time. I’ve got 10kg sitting in my print room. Of course some materials shrink more than others, but there’s material shrinkage and there’s inaccurate machines. Both can be happening here. It could also be a slicer issue.
Regarding flagship printers: In my opinion this is not a high spec printer but a consumer printer with an extremely rich feature set at a budget price.
I agree. It can be both a flagship model and a consumer grade printer. Either way, it should be printing at least as accurately as an unenclosed bedslinger. The accuracy I’m asking for is not unreasonable.
1 Like
Anyone up for a challenege? How about you post such a model to makersworld and provide a link here. Preferably include the original fusion model as well in case people want to take measurements off the exact model itself. Then let’s see how close people can come to printing it as intended.
1 Like
I’ve never put anything on Makerworld. I uploaded a 3mf, stl, step, and f3d. Looks like if you click “download stl” you get all 4…
I’d also like to point to this graph from the CNC Kitchen video.
@Alex_vG
If material shrinkage is the culprit, why are all these printers giving different sizes? Especially the 2 X1C’s compared that have a nearly 100% accuracy gap. If it’s just material shrinkage then at maximum one of those printers is printing accurately.
————
EDIT: I decided to follow the prevailing advice from the “It’s not the printer” crowd. I adjusted my filament shrinkage setting. Stock settings but with wall speed turned down to 50mm/s. 3 walls, Inner outer inner print order, precise walls on. Walls are extruding at exactly the prescribed thickness. All measurements with Mitutoyo certified calipers done multiple times sitting flat on a desk.
Linear Dimension vs Result
10 / 9.93L, 9.89R - ~0.9% error
25 / 24.99L, 25.00R - 0% error
50 / 50.05L, 50.00R - 0.05% error
100 / 100.06L, 100.05R - 0.055% error
Linear dimensions are pretty good! Not sure why the 10mm parts were so small, lets check out the round parts: (calipers are not the correct tools for measuring holes)
Radial dimensions
Hole: 15 / 14.80 in X, 14.78 in Y - 1.4% error
Cylinder: 15 / 14.83 in X, 14.78 in Y - 1.3% error
15mm HSS rod did not fit in hole
That is over an order of magnitude difference in size error.
What does this prove? Not much with a sample size of one, but a brand new X1C, freshly calibrated prior to the print, sitting on the floor, with Bambu filament tuned for dimensional accuracy did not produce an accurate print. It is not filament shrinkage. The advice of “change filament settings to compensate” is not the fix.
2 Likes
Are these all your measurements on those different machines? Were they taken from the model you posted? What is it that you mean by them having a “nearly 100% accuracy gap”? Were they all printed using the same roll of filament? I hope it’s obvious that different filaments might shrink by different amounts. I’m not by any means a metrologist, but I’m guessing it’s the difference in x and y error that might be the bigger concern because, from what little I’ve read, x and y should shrink by the same amount. Right? Just eyeballing it for the two X1C’s, the difference looks to be 0.05% in the one case and about 0.1% in the other. I’m don’t know how to evaluate that, other than asking: How much difference in error is within tolerance?
@Tdoening Is the model posted by Zenith1 a match for what you were referring to?
Sorry for the lack of clarity. The graph is from a CNC Kitchen video about skew posted earlier in this thread. It has nothing to do with the model and is not my data. I was responding to a post blaming dimensional inaccuracy on shrinkage.
100% accuracy gap = one X1C is twice as inaccurate as the other
I looked for some references on material shrinking. Surprisingly, only very few filament manufacturers specify that shrinkage in their technical datasheets. One of them is european Extrudr. for their NX2 PLA the state 0,3% shrinkage in the datasheet:
This website starts with a summary about typical shrinkage rates of different materials: https://filament2print.com/gb/blog/136_warping-contractions-3D-printing-parts.html
So in my opinion, if a printer prints parts that are closer than 0,1% to the design, then either the manufacturer or the user has baked some shrinkage compensation into the firmware.
I have bought the original Califlower with accompanying XLS from Vector3D some months ago. I still have one printout in ColorFabb PLA regrind at 150% scale. These are the measured dimensions:
Outer X measurement is 149,44mm (average of three measurements).
Inner X measurements is 149,59mm.
The average of all X values is 149,51mm instead of 150mm.
Outer Y measurement is 149,56mm.
Inner Y measurement is 149,69mm.
The average of all Y values is 149,62mm instead of 150mm.
So overall I have an X error of -0,30% and an Y error of -0,23%.
This nicely fits my earlier stated ± 0,3% shrinking for PLA.
So I think my printer is accurate to better than 0,1%. In the end it doesn’t make a huge difference. Apparently there is a slight difference of 0,07% between X and Y.
From my point of view, the most accurate printers assuming a shrinkage of 0,25…03% are Prusa MK4, X1C_2, Prusa XL, Ratrig, X1C_1 and A1. While QIDI, A1 mini, Hornett and MK3 are far off.
To settle this discussion some, I think it would be a good idea to separate filament shrinking and printer motion system. For that, don’t measure printed parts but movement of the printhead itself.
E.g. move the printhead to an arbitrary position, measure the distance from the side, then move the printhead in one direction by 100mm e.g. with the touchscreen controls and measure distance to the wall again.
Repeat for the second axis.
If the difference is less than 100mm, then that is the amount that the printer itself contributes undersized parts.
Did you only measure outer dimensions? Then also other inaccuracies of 3d printing add to the measurement error. Vector3D put a lot of good thougths into the CaliFlower. He averages inside and outside dimensions to cancel out over or under extrusion for example.
Horizontal holes have been known forever to be too small in 3d printing! Are you sure, that your other printers do not do that? All printers I have worked with so far typically undersize holes by ~0,2mm.
I suppose rather than reinvent the wheel or print objects, it might make more sense to use a calibration object. In my case, I used the CaliLantern, and IIRC, I used Sunlu transparent PETG to do the print. I’m quite sure I didn’t use shrinkage compensation in the slicer, because at that point in time I wouldn’t not have been aware that such a feature existed. Unfortunately, it doesn’t seem as though Sunlu has datasheets for its filament, so if I were to do it over again, I would try to print with a filament that has a good datasheet to go along with it. Which filament maker has the best, most accurate datasheets for their filaments? Anyway, drawing upon generic shrinkage data from google:
and comparing to my results:
It looks as though 1. my x and y error were very close to the same, and 2. much if not the entirety of the error may have resulted from the PETG shrinkage. I most likely used the generic profile, because this was before orca slicer and at the time I didn’t know enough to really do much of anything in terms of tweaking my printer settings. The machine would have been about 7 or 8 months old at the time and hadn’t been used much, so it was in good condition.
I don’t think that the Calilantern specifically addresses the issue of wrong sized holes though. I guess that’s a limitation. Is it really a separate thing? If so, maybe I could suggest to the author that he add something for measuring errors in hole diameters.
The Calilantern is a pay-to-use print, but I read somewhere that someone published an equivalent, free version that gets you the same kind of measurements. It’s a very tedious set of measurements, and not something I would want to do very often. IIRC, the Califlower is a lot fewer measurements, and maybe it’s good enough for the concerns raised here.
Edit: Here’s a link to the free Califlower alternative:
https://www.printables.com/model/164261-calibration-bro-califlower-calilantern-calibration
Here’s the original. $6 doesn’t seem much to pay if it generates good data that can be directly compared to what others are getting.
https://www.printables.com/model/682023-califlower-calibration-stl-calculator
I think the author may have updated the Califlower since I last saw it, since it now looks different than what was in his original youtube video.
Good luck!
Going to shorten quotes for brevity.
Fair enough, but this does imply that those who receive the desired measurements from their X1C’s have printers that are overshooting eg. moving ~100.3mm for a 100mm model. (Assuming that they are printing with no compensations.) Thanks for hunting down some actual shrinkage data.
Califlower with accompanying XLS from Vector3D some months ago. I still have one printout in ColorFabb PLA regrind at 150% scale. These are the measured dimensions:
Outer X …
Inner Y measurement is 149,69mm.
The average of all Y values is 149,62mm instead of 150mm.So overall I have an X error of -0,30% and an Y error of -0,23%.
Again thanks for providing actual useful data for this thread. It looks like you’ve got slightly better accuracy than I do. I’m seeing about ~0.5% shrinkage from BL Basic PLA but this could just be the filament.
From my point of view, the most accurate printers assuming a shrinkage of 0,25…03% are Prusa MK4, X1C_2, Prusa XL, Ratrig, X1C_1 and A1. While QIDI, A1 mini, Hornett and MK3 are far off
Fair enough again. There is a tendency towards a mean error of about 0.25% ish shrinkage. However those 2 X1C’s still bother me. Why are they printing so differently? I think this graph lends credibility to the theory that BL simply has a wide window for QC when it comes to dimensional accuracy. That is why some people are printing PLA parts with no compensations and seeing exactly the dimensions they asked for (their printer is printing large and “compensates” for shrinkage) and other people are printing PLA parts the same way and finding small parts that need compensation.
I think it would be a good idea to separate filament shrinking and printer motion system. For that, don’t measure printed parts but movement of the printhead itself.
E.g. move the printhead to an arbitrary position, measure the distance from the side, then move the printhead in one direction by 100mm e.g. with the touchscreen controls and measure distance to the wall again.
Repeat for the second axis.
If the difference is less than 100mm, then that is the amount that the printer itself contributes undersized parts.
Wholeheartedly agree but I see a small issue: The printer can likely move 100.0mm when commanded to by the user, however input shaping and all the processing done in the slicer is not active for manual jogging, so it’s possible that a 100mm commanded move from the model would still not travel the whole distance in actual printing. I think the way to test this would be give the printer a sliced 1 layer, 1 wall, 100-250mm square and strap a pen to the printhead and let it mark on a piece of paper. I’m not going to pursue this yet but it would be a good test.
Did you only measure outer dimensions? Then also other inaccuracies of 3d printing add to the measurement error. Vector3D put a lot of good thougths into the CaliFlower. He averages inside and outside dimensions to cancel out over or under extrusion for example.
Yes, the model was mostly intended for the round vs linear issue. The 7mm measuring slots came out as 6.95 and 6.97. Am currently printing the free Califlower Bro version so I’ll report back once that’s done.
Horizontal holes have been known forever to be too small in 3d printing! Are you sure, that your other printers do not do that? All printers I have worked with so far typically undersize holes by ~0,2mm.
Indeed they are! That model was for Tdoening. I’ve noticed any hole geometry (rectangles included) tends to come out small. Nature of additive manufacturing that I compensate for (with a drill)
Between the data provided by Alex_vG, NeverDie, and myself I’d like to make a few statements for anyone who finds this thread and wants a TL;DR. Hopefully everyone can agree on these.
1. Filament shrinkage is real and should be accounted for.
2. Not all Bambu Labs CoreXY printers print parts at the same size.
3. Not all Bambu Labs CoreXY printers are delivered with a square gantry.
4. Bambu should enable a way for users to compensate for shrinkage in Bambu Studio
1 Like
So, to mitigate that, maybe we should all be putting +0.2mm in the XY hole compensation field of the slicer? Unless you/somebody know of a more accurate number, I’m going to go with that. Seems like it should be the default. Better than zero.
https://wiki.bambulab.com/en/software/bambu-studio/xy-hole-contour-compensation
I would caution against that. XY Hole and Contour Compensation often hurt more than they help, here’s an example.
Lets say I want to print this part in this orientation. I’ve decided my holes need to be 0.1 bigger and my contour needs to be 0.2 bigger. If I use XY Contour and Hole here’s what happens:
First half of the print goes great. The printer sees a rectangle with a rectangular hole in the middle and applies contour compensation to the outside and hole compensation to the inside.
When the printer gets to the cutout layers it no longer sees a rectangle, it sees a C shaped contour with no hole.
At this point the inner surface of the rectangular hole will have a 0.3mm ridge in it. The hole is no longer being made 0.1 bigger and since the C shape is now all contour it gets a 0.2 compensation inward.
If you’re printing something simple, XY Hole/Contour compensation can work. It leads to serious problems when used with more complex models.
4 Likes
Nothing is wrong with that. I just don’t understand why if you haven’t found a solution to “this well known” issue in 3 months, why you wouldn’t instead buy the cheaper competition that doesn’t have the same issue. The only reason I can see is that this isn’t a deal break as you implied it was.
1 Like
You say “well known issue” in quotes but this thread is the 10th most viewed thread in the entire X1C forum. I think that speaks for itself. I said I bought the Bambu for the AMS, there is currently little game in town for multi material systems. Prusa is an option, but is nearly as expensive as the X1C while being significantly slower and does not include an enclosure.
Discussion between the two of us is pointless. You doubt my motives and I get nothing useful from you. Others here are contributing to the discussion. Your method is to oversize parts and then machine them later. If that process is acceptable to you, great. Enjoy your approach. See post 94.
3 Likes