The secondary cranks with their stub axles are shown below. Note that the stub axles screw into the cranks but that the outer bearing sits partly on the crank.
The left-side stub axle with components is shown below. Note how the sprocket (13 tooth) which is located between the two bearings, screws onto the stub with a right-hand thread (because the secondary crank counter-rotates). The stub also has a right-hand thread into the secondary crank. One end of the stub as shown still needs to be cut shorter (compare with picture above). The bearings are off-the-shelf sizes, and the sprocket is from a standard derailleur cluster. Given the directions of rotation, the right-side sub-assembly uses left-hand threads. For the same reason the left-hand pedal is fitted on the right-side, and vice-versa in this application. However, a standard left-hand threaded 13-tooth sprocket was not available. Instead, bearing-grade Locktite adhesive was used to ensure that the right-side sprocket does not unscrew under load.
The right-side primary crank is shown below, length 120mm between centers. The component on this side was fabricated from 10mm mild steel plate bored out for the outer (larger) bearing, and the inner bearing housing was brazed onto that. Note the retainer plate for the tapered joint between shaft and crank.
Initially I made parts from steel steel, but following initial feasibility tests, a more expensive but much lighter 5083 aluminium alloy is being used where possible. The left-side primary crank was cut from 10mm aluminium alloy plate, shown here partially assembled:
Irrespective of the type of alloy, the thickness of the plate was determined by the need to minimize the Q-factor (distance between pedals). However, use of relatively thin 10mm plate posed the additional problem of joining this to the shaft. Conventional bicycle cranks use a very sturdy square internal tapered mounting which is much thicker than this. Instead, I have used a split tapered collet which slides onto the square (non-tapered) shaft end and a retainer plate that forces the collet into the crank body. The taper into the crank tightens the hold of the collet onto the shaft in the manner that a chuck grips a drill. At the same time the tapered fit of the collet into the crank transmits the torque between shaft and crank. A similar arrangement is seen in the so-called "TaperLock" adapters. The parts of my homemade version are shown below. The 17mm diameter shaft is made from silver steel, which is a medium-carbon alloy.
A rough estimate of the strength of this joint prior to actual road testing was possible because the 1/2 inch square shaft ends fit precisely into a conventional torque wrench. I used a Britool EVT 2000 wrench that is certified and calibrated up to 230 Nm. In the picture below you see a test in which a torque of 160 Nm (approx. 100 lb ft) was applied to the shaft, with the crank held in a vise, and no signs of slippage.
As seen above, the length of the shaft is such that not only torque but also considerable bending strain would occur when the wrench was pulled. Moreover, this was repeated in a five-minute workout on the torque wrench in a fairly realistic simulation of operating conditions. At 200 Nm I felt a slight yield. Inspection showed that the shaft had started to give, but the taper sleeve held.
It is by no means a precision machine (of mainland Chinese origin) and one has to take very great care and be very patient, to obtain accurate results on such equipment!