# Accurate calibration/dimension test

Apr 29, 2017

### Thing Apps Enabled

Does it matter what size nozzle you use?

I have used 0.4. But the method in the excel sheet is independent from the nozzle. Even over or under extrusion are compensated.

AWESOME !!!

Got the deviation to minimal(0.02% & 0.09%) after 2 iterations!
Standard deviation X = 0.09
Standard deviation Y = 0.12

Perhaps if/when you update the files you could make a variation that combines both X and Y together?

Thanks

Nice to hear that it is useful. A combined version is a good idea.
Your printer is quite accurate. Just for the stats which printer and slicer have you used?

My printer is kind of a cross between Prusa i3 and Mendel 90.
I used Cura running off Repetier host

Anet A2 using Simplify3D
X std. dev current = 0.28
X std. dev expected after correction = 0.19
Y std. dev current = 0.32256413
Y std. dev expected after correction = 0.141300166

Just noticed that in the excel calculator for X, Y and Z, Row 23 D - H doesn't have a formula

Thats by purpose because the mean of the C16 and C23 is used as a reference. Note: Always the difference to this reference is evaluated instead of absolute length. This eliminates patially some effects which should not be concidered when calibrating (wobble, under/over extrusion).

Thank you very much! Great information and excel calculation sheet!

Quick question, I understand adjusting the e-steps.

What am I supposed to do with the "blue" highlighted "tolerance needed to xx.x% of the prints fit"?

Thanks again!

That's a guideline when designing parts in CAD.
For example: You design a shaft for a bearing with a 7mm inner hole. If the shaft diameter is exactly 7mm in the stl file sometimes the printed shaft will be smaller and sometimes larger than 7mm. So sometimes it fits sometimes not. (in the best case this is a 50/50 chance). So either you will have to sand, drill, etc until the parts fit or you design the schaft to be slightly smaller. To have a guideline how much smaller it should be engineers often use 3 times the standard deviation of the manufacturing process. (3sigma).
Thats a short explanation of a quite complex topic. I hope it answers your question.

I derived a guideline for my designs from this: 0.2mm for tight fits 0.4mm for normal fits and t least 1mm of space if parts shouldn't touch.

Of course you can try to tune your printer until it is perfect, but to be fair this is very costly and actually impossible. Therefore it is better, easier and cheaper to accept the limits of your printer and forsee tolerances right in the design.

Thank you for responding so quickly!

I've worked in manufacturing environments for a few years now and I've heard the terms 3sigma and 6sigma thrown around quite often and have really meant to look into them so I appreciate the basic explanation and i'll look into it further!

I have a Duplicator i3 V2.1, i've done the Z-brace mod, squared the frame and calibrated my filament e-step. Using white hatchbox ABS filament, running Slic3r, 240 C extruder.

Standard deviation for X-axis before adjusting the Steps/mm was 0.64
After making the calculated adjustment to the steps i'm at 0.08

Still need to print and check the Y and Z axies, I know the Z has been pretty spot on in the past when I have measured various parts so I don't expect to have to adjust it.

So your saying due to having a standard deviation of 0.23 and 0.24 you generalized and have a higher % chance of parts fitting tightly by adjusting by 0.2mm?

This sounds reasonable although it's kinda a bummer to have to adjust designs in that sort of way, i'm sure different filament chemistry's effect the deviation as well? So when printing the same model with different filaments you may come out with different results and tolerances of the actual parts produced.

Thanks again! This is a great tool and resource!

It is a standard procedure in mechanical design and a design not concidering the limitations of the manufacturing process is not accepted in real engineering. Normally this is done by defining assymetric tolerances which are condiddering when manufacturing, but in 3D printing the CAD and CAM (Computer aided manufacturing) is merging. So I don't see any alternative.
The difficulty is to get a good set of standard values which fit to every printer, slicer, filament. The statistical methods used are a first try but they defenetly do not cover all effects (filament, temperature, vibrations, etc). So I concider them more like a rough guideline rather than precise values.

Here are my results:
Standard deviation X: 0.24mm
Standard deviation Y: 0.23mm
Standard deviation Z: 0.1mm

Printer: Anet A8 (Many modifications: Belt tighteners, Ramps 1.4 DRV2285, Calibrated)
Slicer: Slic3r Prusa Edition

Hello, my printbed is to short in Y-direction, only 150 mm. Any idea to modify the part and formula to fit?

Just uploaded a smaller version (145mm). Not tested yet but it should work.

ok, I have tested it, everything is working.

Great, thank you :-)