I published this on the RepRap buliders blob http://builders.reprap.org/ some days before and was asked to publish it here also.. so here is my post from the Builders blog...

nspired by a paper from Kruth et al. (2005, Benchmarking of different SLS/SLM processes as rapid manufacturing techniques) I also designed a benchmark model for the RepRap. It looks quite random but by printing it you can check whether your RepRap prints geometrically correct. I included the following features:

-vertical holes for M3, M4, M8 screws
-vertical hexagons suitable for M3, M4 M8 nuts
-horizontal teardrops M3, M4 M8 and hexagons for suitable nuts
-sharp edges with angles of 15, 30 and 45 deg
-thin walls 0.5mm, 1,0mm, 1.5mm and 2mm horizontally and vertically
-stair effect at sloped walls with angles of 15, 30, 45, 60 and 80 deg
-round corners, radius 4mm to 16mm
-gear teeth with 2mm wide teeth
-in addition to this, you can also check perpendicularity and parallelism of your printed objects, which were the general intentions of designing a benchmark.

You can download both the STL and the AoI file from SourceForge

/mendel/mechanics/test-parts

Printing it took about two hours, so why not watching a fancy movie while your RepRap prints its benchmark.

The very first version included a complete horizontal hexagon for M8 nuts, but apparently this made the object to big in height, so I just cut it off, leaving just a small offset ;)

The gear teeth were not built as expected but as a round corner with a bubbled rim. I am also missing the 0.5mm thin walls. The 1.5mm walls consist of only two parallel strings and are not solid, there is a small gap between them in both directions.