Sometime in 1970, our dealership received a shipment of Kawasaki's cute little 90cc G3s, and I began to use one as my commuter bike. Every now and then, it would half-shift; the shift lever would only go halfway and stop, and the bike would not be in any gear. Pull the clutch and poke again, and it would obediently engage the next ratio. What could be causing this?
I thought about this as I hummed back and forth between the shop in Arlington and home in Watertown. What would stop the occasional shift at the halfway point? Gear engagement is not made by sliding gear teeth into and out of mesh—a few vigorous first-to-second upshifts and the teeth would begin to be mangled. Instead, the end faces of gears carry hefty squarish pegs called “dogs” usually in sets of three. To engage a gear, the shift lever operates a claw that rotates a shift drum. Into the cylindrical surface of that drum are milled wiggly slots, each of which drives a shift fork by means of a peg that engages a slot in the drum. Each C-shaped shift fork moves a gear in the gearbox, which is free to slide axially. It does this by riding in a groove machined into the gear.
People who abuse transmissions often either bend the shift forks from their stomping efforts to shift or cause them to become blue with heat from trying to hold them in engagement when the shift dogs are too damaged to transmit torque without kicking back.
When you move the shift lever, the shift drum rotates a specific amount, moving one shift fork to disengage the gear you are now using, then engages the next gear by sliding one set of three dogs into engagement with another set on the appropriate gears; this all becomes clear when you have the parts in front of you.
Now to the heart of the matter. When one dog set is moved along a shaft toward another, one of two things happens: 1) the dogs of one gear slide neatly into the spaces between the dogs of the other; or 2) the two dog sets hit face to face. (By “face,” I mean the dog surfaces on which the gear would rest if you set it down—gear axis vertical—on a flat surface).
In case one, no problem, the shift completes normally. In case two, because the gearbox has oil in it, the well-lubricated dog faces slide off of each other and drop into engagement. What could interfere with this process? Machining burrs on the dog faces, that’s what.
I wanted to test my theory. Arriving at work I pulled my G3’s engine, removed the cylinder, head, and outer cases, and split the crankcases. This engine has only one cylinder so this was not a very time-consuming process, and I was curious about the shifting problem.
Soon, using my trusty collection of transmission snap-ring pliers, I had all the gears off their shafts, lined up on a work surface with all their shims, clips, and bearings in order for reassembly. One by one, I picked up the gears and examined the faces of their engaging dogs. They had been rapidly machined flat on a lathe, leaving a surface of tool marks.
I decided to give this little gearbox the same treatment I had learned to give racing gearboxes: to lap the dog faces on a sheet of 220-grit abrasive paper (lying on a flat surface) until all tool marks were gone. Then, using a high-speed flexible shaft tool with an abrasive-charged rubber wheel, I slightly radiused the edges of the dogs (sharp edges on gear dogs are vulnerable to hammering when shifted).
I washed up the parts and reassembled the engine. Riding it home later that evening and thereafter, I never again had a half-shift. Evidently, with the dog faces now smooth and without burrs or sharp edges, when dog sets hit face to face rather than face to slot, they quickly slid off each other and into engagement. As they should.
In later years, I would see more than one motorcycle manufacturer make this same kind of discovery and respond by improving critical surface finishes in their gearboxes.
With so many of life’s problems being too complex for solution, it’s a satisfaction to get one little thing right.