I use nopheads math, and added a few other things to account for n-sided holes (a 4 sided hole has a larger diameter than a 6 sided). You also need to account for extra material in the corners, like for nut capturing I cutout a bit at each vertex, the plastic extrusion can’t make perfect bends.
External accuracy is a function of the slicing software and layer height. I believe nophead did the math for that too.
I’m dealing with some substantial Slip-stick which also may be why I’m experiencing as large a factor as I am to get the real world metal parts to fit into the plastic…as has been stated here, Z seems to be okay, and there’s some fudge factor in the model to compensate. It’s probably some combination of over extrusion and printer kinematics that I’m compensating for. The part in question takes two 20x20 extrusions…the holes are 20.2x20.2, and if printed at 100% results in some delamination by the time I get the parts inserted.
Wow. So glad I saw this. Tried a few times to get a semi mechanical cell phone to Microscope adapter to print in PLA, but the sizing was off for a few components. Still fairly new to FDM, so I’m very tempted to try it one more time now that I’ve seen this.
Been there!
I typically adjust my models (e.g. slightly larger holes) instead of scaling. But I agree with @Eclsnowman 's solution: calibrate as close as possible and then scale according to the plastic being used. Also, @Stephanie_A makes a good point about Nophead’s math. If you haven’t read his Polyholes post, I highly recommend it, as well as pretty much all of his 3D printing posts, which are always densely packed with well thought out information/tips. His blog (like RichRap’s) is required reading for the serious 3D printing junkie, IMHO:
LOL- im doing that same thing right now! Its been a while since I updated the firmware on my printer - and I cant find the file - so rather than redo everything Im just sticking with the 105% until my new printer is done! I wish you could extract the firmware out of the arduinos!
Adding on what @SirGeekALot said, there’s also an ebook version of that on the reprap developer bookshelf compiled by @Gary_Hodgson that makes for great bed time reading on something like a kindle (afk me how I know ;))
I hang with a buncha First Order Geeks. That makes me insanely happy.
I think correcting material-related inaccuracies in the slicer is appropriate but not constant inaccuracies in the machine itself. Those should be calibrated out.
I do not agree with adjusting dimensions in the model to compensate for either of the above. Models should reflect reality and employ sensible tolerances to permit them to be valid for future technologies.
@Neil_Darlow “I do not agree with adjusting dimensions in the model to compensate for either of the above.” I 100% totally agree with you. I did not say this was a “best practice”–it is quite the opposite actually. But it is a “quick & dirty” solution for me when I am making one-offs for single use (pretty much most of the time) and the tolerances aren’t really important (e.g. I want the screw to fit through the hole without needing further processing, and I don’t care if the hole turns out a tiny bit too large).