Elephant foot compensation test (that REALLY works)

Why calibrate Elephant foot compensation at all?

Well, I have to admit that I used “standard” values for EF compensation for years. But I stumbled over this matter when designing my printed pinhole lens where I use the EF compensation feature to arbitrarily widen the pinhole on the first layer, so this was my first application where a precisely set compensation was vital for the whole project. 
Searched and found some test files on printables.com and thingiverse.com that claimed to “test” EF compensation, but none of them served the purpose in a way that could be called “correct” or at least “helpful”. So I had to make my own test file.

Using the file is easy:

Just print it using zero elephant foot compensation, then hold the print up against a light source. Ideally, you should see some (up to five) small holes in the plate. The smallest of these is the one whose number (printed underneath the hole) should be entered as EF compensation value.

If you don't see a hole at all, reprint the file with an arbitrary EF compensation value (I suggest 0.2 mm). Again, hold the print up against a light source. Look for the smallest visible hole and add its value to what EF compensation you have previously set (so, for example, if you printed using 0.2 mm and the smallest hole in your print says “.15”, set a value of 0.2 + 0.15 = 0.35 mm).

If you want to positively confirm your settings, you can reprint the file using your determined settings. In this print, all five holes have to be visible.

Printing the file is, well, not THAT easy (but easy enough):

Elephant foot compensation is a top level setting, meaning it is influenced by each and every other calibration setting. There is no sense in tuning in EF compensation and then change filament feed and flow, retraction values, nozzle/bed temperatures, or z distance/height map (the latter done automatically by most of today's 3D printers) - all of these settings should have been made prior to EF compensation.
This model uses structures that are finer than needed for ordinary printing, so your current acceleration/jerk settings will definitely be wrong for this as well, but we will compensate for these simply by using a very low printing speed for the first layer.
Additionally, please make sure that your nozzle is really clean (to Bambu users: The automatic “nozzle cleaning” is not sufficient). And use a non-structured, clean buildplate.

Now enter these slicer settings:
Layer height (all): 0.15 mm
Line width (all): [your nozzle's size]
Ironing: OFF
Arachne feature: OFF (Wall generator: Classic)
Printing order: Inner walls/outer walls/infill
Wall loops/perimeters: 4
Bottom shell layers: 4 (or more)
Speed first layer: 4 mm/s (Infill may be up to 15 mm/s)

All other settings are either arbitrary or depend on your current material calibration.

Tip 1: You can print multiple copies of the file at the same time (sequentially) and assign different EF compensation values to them in order to extend the range of values tested.

Tip 2: Of course you can refine your results by entering an interim value (like 0.175 or 0.215) as the base EF compensation value for the second (or third, fourth…) print, but in daily practice there are too many unpredictable factors to really make use of a setting more precise than +/- 0.05 mm.