Hundredth of millimeters functional dimensioning (industrial context)

[Nicolas3 6]

Hello everyone.

I am using Ultimaker 3 for professional purposes and I would like to raise you some questions. Actually, after using UM3 (with PLA) for some random pieces (like tool holders, etc.), I would like to print specific pieces that requires dimensional precisions.

I would like to respect few hundredth of a millimeter for dimensional precision. But after some tries, I can't fin the right parameters.

I've tried several layer height (0.2mm and 0.05mm) but still I am off the limit (several thenth of millimeters).

I would like to know if other parameters have an influence on dimensional specification ? (like printing speed -> Wall speed), and if some of you have faced the same issue ?

Then, do you think that we have to adapt the design of the part that I have to print in order to take into account the radius of the print core ?

Or do you think that the plastic shrinkage of PLA can have an influence ?

Thanks in advance,

Nicolas

[Labern 775]

PLA shrinks heaps. It pulls in when the nozzle moves like a rubber band.

There are so many factors that can effect the accuracy of the finished product.

Print Speed, Temp, Layer Height, Acceleration, Filament Type, Filament Brand plus many more.

It depends on the model how to best combat this issue. Internal dimensions of holes are the most effected.

There is a Horizontal expansion setting that you can use to help resolve that.

Sometimes its best to print out a section of your model where you need the accuracy. Measure it and adjust your design. This way it can be a lot faster and save a lot of plastic.

I normally design holes 0.3mm bigger then actual for example.

[geert_2 558]

In addition to what Labern said: I found that blobs, ringing effects around corners, calibration of the first layer ("elephant feet") and similar deformations also affect accuracy on an UM2. So they may exist also on an UM3 to some degree.

Speed and temp have a visible big influence: print a 10mm x 10mm x 10mm test block and manually change speed or temp half way. You will clearly see the difference with the naked eye.

If the layer height is 0.1mm, then there are already layer lines, thus ripples of about 0.05mm, in the Z-direction. Blobs and these layer lines generally cause a part to be 0.1mm wider than designed in my models, or holes to be 0.1mm narrower (for big holes) to 0.5mm narrower (for very small holes) than designed.

So I guess post processing (sanding, grinding, smoothing by dissolving, painting, polishing, drilling out holes) will be required anyway, for good accuracy.

[neotko 1,417]

IMO bowden can't deliver precision, but with the zge (directdrive) so far I'm getting down to 0.03mm for x/y and 0.10-.0.15 max to holes.

To avoid expansion, a good cooling system helps to 'frooze' the extrusion (for pla/platec). Using also 0.35 extrusion for a 0.40 nozzle.

There are many factors, mainly getting use to X material and repetition under the same temp/speed. After 2-5 tests you could get down to 0.05 precision comparing x/y outside and circles inside and doing adjustments on the model.

Anyhow, all that adjustments will go to caca if you change even 1 element (even color from the same material can affect).

[yellowshark 153]

Hi Nicolas,

Lots of good points above. In my experience if the hardware and software are set up correctly/appropriately then on the x/y axes you should achieve at least 50 micron accuracy, consistently. You can achieve better results maybe with variation but of course once you get to say 20 or 30 micron accuracy then how good you are at measuring is a big factor. If you have not been trained then you probably are not.

Managing shrinkage can be assisted by

• Running your extruder as cool as possible

• Having you environment at say at least 20c or warmer.

• Leaving your part in the printer until the bed and printer have cooled to ambient temp.

• Leaving your part in the environment for 24 hours (relevant for N Europe in the winter, maybe not Florida all year round!).

I appreciate the 3rd point may be awkward.

PLA ranges from semi-crystalline to fully amorphous; crystalline polymers shrink more than amorphous polymers.

Thicker walls cause more shrinkage in injection moulding, I assume the same applies with 3D printing.

How much you squash down your 1st layer (nozzle to bed distance) also can impact your results. I had a part recently where I had to squash it down a lot to get adhesion, which causes elephant feet and I had to take 100 microns off the x/y dimension on the bottom 300 microns of the part to get the dim. right.

This squashing also impacts the Z dimension, which is harder to resolve as you are constrained by the layer height unless you are slicing with S3D (I am not aware of a Cura add-on which lets you modify layer heights but I could be wrong).

We never print a dimensionally critical part above 30mm/s but If you do need to print faster then I guess the answer is to review the results and try to modify the part dimensions until you get a match.

I do not know what precise tolerances you have to work to but with a physically new machine you should be able to achieve 100 micron accuracy quite easily and 50 microns with some care.

HTH

Edited March 17, 2017 by Guest

[Nicolas3 6]

Thanks a lot for your answers.

Indeed, I'll carry some tests with all the parameters you've mentioned above.

3D printing sounds like a "delicate" science with you guys.

I'll keep you up.

Nicolas

[Nicolas3 6]

Guys,

I did it. As you've said, some parameters affected the accuracy of the printing.

I've used :

Ultimaker PLA

layer height = 0,05 mm

Vprinting = 30mm/s

I've haven't tried to adjust the temperature, but in winter I will "play" with this parameters.

And yes, indeed, changing the filament has an influence.

Thanks a lot,

Nicolas

[kmanstudios 1,120]

Would your nozzle size also be a limitation as a restriction on the x/y axis? Kinda would like to know that since it could be a factor for me as well.