I was looking forward to working on the coolant pump, but after studying its drawing I decided to make a change to the shaft seal. Since I didn't want to start machining the castings until I actually have the seal in my hands, the pump will have to wait until after the seal arrives.
For now, I turned my attention to the starter countershaft. This simple shaft is driven by a cluster gear which, in turn, is driven by either the manual or electrical starter. A spur gear on its lower end will eventually drive the crankshaft through a geared one-way bearing located on yet another countershaft. The starter countershaft will be supported by a pair of ball bearings. The top bearing will be radially constrained in the bulkhead that was earlier machined, installed, and line bored with a lower bearing recess inside the wheel case. This countershaft will have to be assembled in place in a difficult area inside the wheel case. Assembly will include its bearings, a pinion depth spacer, spur gear and key, and a backup setscrew. In the most recent drawing that I have, a 1/16" diameter radial pin is used to secure the bevel gear to its 3/16" diameter shaft.
This tiny pin caught my attention, and so I decided to do some rough calculations to see if it would really hold up while supporting the engine's starting torque. During my 18-cylinder radial build, I found that a torque of some 14 ft-lbs was required to turn over the completed engine. The radial has six more cylinders than the Merlin that contribute additional resistance, but the torque requirement will also increase when the crank is spun up fast enough to start the engine since the cylinder pressures have less time to leak down. I conservatively estimated that the starter countershaft's pinned bevel gear will have to handle some two ft-lbs of torque after accounting for the gear reduction between it and the crankshaft. Upon converting this torque to a pair of shear forces acting across each end of the pin, I came up with a 40 kpsi tensile strength requirement for the pin. Very little margin would be available from common alloys, and so I decided to shrink fit the parts together and use a short length of spring or 'piano' wire for the pin as a back-up. Eventually, I also increased the diameters of the shaft to 1/4" and the pin to .078". The steels used in the tough control rod stock commonly found in RC hobby shops can often provide well over 100kpsi.
I wasn't willing to accept the dimensions in the drawing for the starter countershaft until its mesh with the gears on the manual and electrical starter shafts could be verified in my actual wheel case. And, I couldn't do this without having the cluster gear that is common to these three shafts. No design details were provided for it other than suggested part numbers for two commercial bevel gears that could be assembled to create it. I derived its dimensions using trial-and-error machined spacers and shafts just as I did for the pump shaft. This gear, which will be located on the shaft for the electrical starter, simultaneously meshes with gears on both the manual starter shaft and the starter countershaft. The gears used to make the cluster were individually trial-mated with their drive/driven gears so the required relative distance between their front faces could be determined. Once this distance was known, the larger gear was machined so the smaller gear could be shrink-fitted into it. The larger gear was bored for a .00075" interference shrink-fit with the smaller gear. In order to simplify the pressing operation, the larger gear was also shortened so that after the gears were joined their rear faces would be flush. A radial pin will eventually secure the cluster gear to the electrical starter shaft, and it will back-up the shrink-fit as well. The temporary Delrin shafts for the manual, electrical, and starter counter shafts were then reused to recheck the gears' mesh.
My method for fitting the gears down inside the wheel case was a bit subjective. I typically blocked one gear of each pair and then tried to rock its mated gear back and forth to check for backlash. No movement indicated the gears were likely too close, while a degree or so of backlash seemed to provide the clearance needed for quiet and silky smooth operation. Setting the bevel gears up for full tooth engagement was made difficult by the limited visibility of the tiny blued gears down inside the wheel case. I used a magnifying glass, flashlight, and, in one case, a borescope to set the engagement.
After the gear fits were verified, the actual components for the electrical starter shaft assembly were designed and machined. No details were provided for either the electrical or manual starter assemblies as these were expected to be customized by the builder. I designed the electrical starter shaft housing to support a 1/4" diameter shaft with a pair of ball bearings. A spacer within the housing sets the position and controls the thrust of the cluster gear in both axial directions. The bottom end of the shaft was machined for an Oldham coupler for flexibility later when the motorized section is developed. For ease of assembly/disassembly the cluster gear was bored for a close slip fit on its shaft and secured with a .075" diameter pin. The pin is a close slip fit in its bore and is held in place with low-strength (purple) Loctite.
With the electrical starter shaft assembly installed and the cluster gear in final position, the components for the manual starter could be designed and machined. Its twin bearing housing is internally similar to that of the electrical starter, but the driven end of its shaft is a 5/16" hex that can be spun with a suitably adapted socket in an electric drill. There's only a minimal flange on the wheel case to support the manual starter, and so the closely fitted nose of its housing was profiled so it could be extended as far as possible into the wheel case. I machined its 1/4" diameter shaft from an appropriated portion of a handle from a hex wrench. The wrench wasn't hardened, but it machined like a tough alloy that should hold up well in this particular application. The bevel gear on this shaft was also secured with a .075" diameter pin, and its depth and axial thrust were limited in both directions with a pair of machined spacers.
Finally, with the starter shaft assemblies completed and installed, the components for the starter countershaft could be safely machined. The required length of the shaft was determined once again using temporary spacers to trial fit its bevel gear to the cluster gear. The spur gear on the front end of the shaft was machined from the blank used earlier to determine the location of the starter countershaft within the wheel case based on its mesh with its driven gear on a second countershaft. The blank required only minor machining to finish it including a 1/16" broached key slot. At the last minute I decided to integrate the mesh-setting spacer into the spur gear since I had determined the bevel gear's exact spacing requirement. In general, this isn't good design practice, but in this particular case I felt it would help simplify a difficult assembly. Assembling the starter countershaft in place wasn't as difficult as I had feared, and when completed, all the gears in the starter cluster were silky smooth when test driven with a drill.
I performed two .00075" interference shrink fits on two quarter inch bores, and I learned a few things along the way. Of course this is an excessive amount of interference on such a small bore, and a low TIR fit would have been difficult to achieve with a cold press. In the past, I've tried heating the female part in an oven for a shrink fit. But, by the time I got it over on to the press and set up, the part had typically cooled so much that the heat was of little benefit. This time, I set the female part up on a stainless steel plate on the press, and then I heated it in place with a propane torch to 400F. With little additional set-up I was able to press the parts together within seconds. I also turned a simple close-fitting flat bottom sleeve to go around the narrow shaft during pressing. This helped to keep the shaft vertically aligned with its bore during those first few critical seconds.
I didn't expect such a simple shaft to require so much additional work, but the starter shaft assemblies were fun to design and satisfying to complete. The shaft seal hasn't yet arrived, and so I'll likely next tackle the second countershaft. This one will be quite a bit more interesting as it will include a couple clutches and an integrally cut spur gear. - Terry
