Accuracy

[CS3D 0]

Hi, 

I would like to know what are the printed conditions and best parameters to obtain the maximum accuracy on a Ultimaker 2+ ? As decribe on the specification sheet of the Ultimaker 2+ 

For exemple I am printing a simple circle/disk that is suppose to be 55mm diameter (PLA / layers 0.1mm / speed 30mm/s) but when I mesure it after it has been being printed I obtain an average variation of the diameter of 0.3mm.

The tension of all the belts has been done as well as the X and Y axes alignment.

All the pulleys are clean and tighten

 

Thanks in advance for your help  and advice

[yellowshark 153]

Hi, this has nothing do you with your printer, probably, but is due to the way 3d printers print circles. I.E. they do not draw circles they draw lines and as a result the diameter of a circle will always be smaller. So you have to give your circle a bigger diameter at design. It may take you a couple of goes to get it right but once you have done it a few times you will get a feel for your printer and will know what increase to make.

 

For straight lines you should be able to hit +/- 0.05mm accuracy. I will not guarantee anything better but can normally hit 0.03 or 0.02.

As long as you are well set up then printing slow is the key to accuracy.

[CS3D 0]

Hi, 

Thank you for your answer. It is really helpful.

But when I talk about "variation of the diameter of 0.3mm" I mean that the diamater for my 55mm circle varies from 54.89mm to 55.18mm

Thanks

[gr5 2,189]

I'm guessing most of the variation is within 1cm of the build plate (aka bed)?  The bottom layer is very hard to get the same diameter as the layer above because of how hard you are squishing the bottom layer.  And if you increase the distance between the bed and the nozzle in your leveling then the part doesn't stick as well.  So this is the hardest layer to be the same as the layers above but it can be done by using a negative value for "horizontal expansion first layer".

 

The next few layers above tend to move inward for a few layers maybe 3mm, and then they move outward again, and then by 10mm from the build plate you should see extreme consistency.  This bowing in is related to the temperature of the build plate.  So now you have more compromises:

You can lower the build plate temperature to say 40C, but now large parts are more likely to warp up at the corners and come off the bed.  For small parts if you need perfect cylinders (say under 3cm in diameter) then this might be the way to go (40C bed).  But for larger parts - more than 80mm across you need to keep the bed above the softening temp (about 52C for pla).

 

In addition to all this, note that as PLA comes out of the nozzle it cools rapidly (in milliseconds) and PLA sticks to itself (which makes the prints better than for other materials) while still liquid so all corners and circles on a given layer (all circles drawn as circles by the X/Y movements in a given layer) are going to pull inward as the stretchy-liquid-rubber-band like material is pulled inward.  I usually compensate this by, for example, making all my vertical holes about .3mm larger than the goal diameter.

 

In summary - the variations you are seeing are probably caused by the material and the temperature and the movement more than the machine itself which is consistent to much better than .3mm - probably about 10X better than that (.03mm).  

 

A photo might help though.  We may be talking about different things here.

 

[CS3D 0]

Hi gr5,

 

Thanks for the answer, it make sense !

I print PLA so my bed is at 40°C (nozzle 205°C) but my printed circle is only 10mm high (Z) so I will have a try tomorow with a higher one and I will let you know (with pictures).

To be a bit more precise, all the diameter variations are mesured at the same height, so technically speaking my circle is not perfectly round at a determinate height.

 

Thanks

[yellowshark 153]

Hi, I certainly and I suspect @gr5 too have misunderstood you, although both our answers are still relevant. Are you saying that on any layer, although probably the top layer, you are taking diameter measurements from different points on the circumference and that these are differing measurements which on average show a 0.3 error?

 

I have never seen that though to be fair I guess I rarely take more than two measurements, at right angles to each other. Also how big is the difference between the measurements. Personally I would not be concerned by 10 or 20 microns and depending on my mood might overlook 30 microns - sorry you said that already, yes quite large!

 

Another thing not to overlook, and you might be, but if you are not an engineer (typically) and trained on how to accurately take measurements with a gauge then your measurements will probably not be good enough to get the level of accuracy you seem to chasing. I know; I am not trained and my measurements are always different and wrong compared with my son  who is a qualified mechanical engineer.

 

And to reinforce @gr5 's point that if you really squish in your filament on the 1st layer to get good adhesion then yes you will see the errors he notes. If I were to print something with required accurate holes then I would deliberately increase the nozzle to bed distance that I normally use; at the end of the day it depends on how much tolerance you are allowed.

[kmanstudios 1,120]

Just to follow up, there is a new feature that is called "Initial Layer Horizontal Expansion" to use as an offset of that issue of the bottom getting squished out a bit. Also, someone on the forum made note that it is good to put a bit of a rounded bevel on the bottom to help combat that as well when designing a part.

         |

         |

_____|    <--------Straight corner

 

        |

        |

____/  <--------Beveled to combat squish or even elephant's foot.

[gr5 2,189]

@yellowshark makes a good point.  We need a better definition of what seems unexpected.  If your cylinder is not round that is definitely a problem and kind of common - it implies your axis are not square.  This should not happen but may be easy to fix.

 

Anyway I guess you need to show us some photos of what you are talkinga bout.

[tinkergnome 927]

8 hours ago, gr5 said:

it implies your axis are not square.

 

...or the chosen polygon resolution (count of faces) is too low?

 

At 50mm diameter one would measure quite a few deviations, if one prints a polygon with (let's say) 20 edges instead of a true cylinder, right?

[CS3D 0]

Hi everybody,

 

And again...thanks for your answers !

Here are some pictures of my "problem"

My 3D model is 50x50x10mm (x/y/z)

I guess I have enough edges

@yellowshark >> yes I am taking mesurement at different height of the circle (top, bottom and middle)

I am using a sliding digital caliper. I know it is not that precise, but I think its enough to notice that my circle is not round (pictures enclosed)

The tension of all the belts has been done as well as the X and Y axes alignment.

All the pulleys are clean and tighten

I print PLA (205°C) with a bed at 40°C 

The nozzle is a 0.4mm

Layer height is 0.1mm

Speed is 40mm/s 

I did print another one at 30mm/s...it is a bit better but not as good as it should be, I still avec an average of 0.25mm error and the diameter

 

Thanks