From the RBT FAQ...
From the RBT FAQ...
From: Jobst Brandt <jbrandt@hpl.hp.com> Since the advent of the crank cottage industry, crank failures have become more common than previously because most of these "home made" cranks, usually produced in prototype quantities on numerically controlled (NC) milling machines, have not been designed with sound engineering practices, nor have they been tested destructively in repeatable tests on testing machines. My comments are mostly aimed at major brand production cranks. Cranks break primarily because they are aluminum and because they have high stress principally at two places. Aluminum has no distinct fatigue threshold in contrast to steel, so that with increasing use and load cranks are destined to break at the two most failure prone places, the pedal eye, and the junction of the spider fingers and the right crank. The pedal eye is weak because the joint is incorrectly designed, but being standard, it may not be changed since it appears to work. This joint always moves under load and through its fretting causes an undercut in the face of the crank. Removal of a pedal, that has been ridden any considerable amount, will reveal a recess in the face of the crank with cracks around its circumference caused by fretting. The cracks often propagate into the crank and cause failure. A solution to this problem would be a 45 degree taper in place of the flat shoulder at the end of the pedal thread. The thin web between the crank and the adjacent legs of a five legged spider is also a place where most cranks crack. The Campagnolo C-Record as well as Ritchey cranks address this problem by using the crank itself as the fifth leg of the spider, and transmitting pedal torque directly from the crank to the chainwheels. By this arrangement the spider merely supports the chainwheels radially and laterally and the driving torque is delivered by a solid anchor.