I’m struggling with coming to terms with dimensional accuracy when printing with PLA. Are there any methods that can be used to determine the correct amount of inflation needed for each axis of a part that don’t involve printing the part out first, measuring and then printing the final part?
I think you’ll find the answer here is no.
Even high end SLS nylon printers have to go through a calibration print. Depending on the precision you require, it could be necessary to perform a calibration print or more than one if you change from one filament color to another or from one filament brand to another.
One cannot be certain of the thermal stability of any given batch of plastic in the FDM world, at least not at the hobby level.
I suppose the multi-ten-thousand dollar and up printers with extreme production control of filament would get away with having a foundation for calibration, but I’d still venture a guess that a calibration print is still performed with the high end machines.
@James_Ling I specialize in high accuracy PLA prints with very repeatable results. You can’t do it without printing something first. Here is what I do.
First I find the proper temperature for a spool of PLA. With my nozzle about 3" off the bed I extrude out a contant stream of filament at a temperature I think should work for it and watch what it does. If it curls like a pig tail then it is too cold. If it stretches then it is too hot. When it extrudes out a consistent strand I know I’m within +/- 5C of a good temperature.
At the same time I check the diameter of the extruded strand to confirm I do not have an obstruction in the nozzle or filament guide. A 0.4mm nozzle will typically extrude out a 0.46 - 0.50mm strand for me.
I then use this temperature as a starting point to calibrate the roll of filament using these calibration clips found here: http://www.thingiverse.com/thing:1037301
If the inner and outer shells have a gap that is not closing after compensating for underextrusion then bump the temperature up by 5 and that generally fixes it.
As Fred mentioned each batch and color will behave differently when it comes to thermal stability. I do this calibration for every spool. Takes about 1 - 1.5 hours but well worth it to have consistent prints for the rest of the spool.
I then print these accuracy blocks to check for thermal stability.
The clips have a 0.2mm clearance between them and my printers can print within 0.1mm of tolerance accuracy. Most well tuned hobby printers can meet these tolerances. I design these clearances into all my assemblies for 3D printing. Snug fits have 0.2mm clearance and linear moving parts have 0.3mm. Rotating moving parts have up to 0.4mm. The end result is the assemblies have 4 - 12 thousands of an inch of clearance and will all fit and work together.
The result of this technique for me is that I can print high tolerance assembly parts on multiple machines with different filaments and have the parts be interchangeable.
Thank you, this helps. My naive process has been to print the target part once, measure, then print the final version, for each different part (for parts that must have specific tolerances).
What I think I’m getting from your posts is that calibration can be performed once per spool, and once dialed in, that calibration and any necessary adjustments to part dimensions will be valid for the bulk of that particular spool.
That is correct. I calibrate each spool once and mark the calibration on the spool. Then when I am slicing parts I check what spool I will be using and enter those parameters to get an accurate print.
One more additional piece of information is that I calibrate it for a particular printer and nozzle. Switching nozzle sizes requires a new calibration. For example a 0.4mm nozzle calibration may be 1.74 and 195C while a 0.3 mm nozzle on the same machine will be 1.82 and 205C using the same filament. Sometimes if I switch the filament to a different machine I will run a calibration check to see if I need to compensate for the unique PID tuning of that particular hot end.
Let me know if this helps you get the results you are looking for.
Another caveat to be aware of when printing precision parts. If you have to use mesh leveling or distance sensing to compensate for a bed that is not parallel to the X and Y then your prints will be skewed and you can forget about any precision printing that relies on Z being perpendicular to X and Y. The smaller the footprint of the part the less of an issue this might be but it is something to keep in mind. I prefer my machines to be mechanically precise without relying on software to correct for alignment issues. I see the software as a way to confirm the precision and not compensate for errors.
When I can lay down a 0.1 mm thick layer across most of a 10 x 10 bed and have it stick well then I know I’m within my tolerances.