Back to an old subject, I thought it might add some background to the
problems of crank attachment by showing some examples of failure and
wear from which the square taper suffers.
http://pardo.net/bike/pic/fail-002/000.html
Among these pictures that Pardo collected are some that show crank
interfaces with cracks at the corner of the square bore and spindles
with fretting marks on the flanks. There is also a spalled bearing
that attests to the bearing load of such a small spindle even with 1/4
inch bearing balls.
Note that the load is carried by the small area on the spindle where
there is no fretting erosion, the clean part above the area with
scrape wear, while the scraped area serves more for withstanding
bending loads form the outboard pedal than transmitting pedaling
torque. This also shows that these interfaces are dynamic rather than
fixed as they appear to the observer. That is why these designs are
being changed as we have seen from Campagnolo and Shimano.
Although not everyone sees these failures, I have had plenty and that
is why I am no as skeptical about the reasons these design changes
come about. Unfortunately some newer contributors, like ISIS, either
haven't seen enough failures or don't have the experience of trying to
solve such problems. We've seen it even by Shimano with two versions
of Octalink, both suffering from the same failure mode.
Jobst Brandt