Even though I decided to scrap the first crankshaft, I finished it up so I'd have a test bed for the remaining setups and operations. I also wanted to be sure there weren't any surprises waiting for me inside the crankcase or with the starter assembly before getting getting too far along with the second one.
Without grinding capability, the Offy's eight inch long crankshaft with its eight half-inch journals is a difficult part to accurately machine. In comparison, the Merlin's crankshaft with its three-quarter inch journals was considerably more rigid and much less troublesome. Even with rod journal packing, .006" deflections could be demonstrated with only modest finger pressure on the center of the Offy's crank. A drawing shows the deflections measured in four different directions with finger pressure on one of the center rod journals with the other three packed. A journal that must be unpacked for machining appeared to be responsible about half of the worst-case deflection. I was able to reduce the .006" deflection shown in the drawing by about half by packing the main journals as well. Even with carefully shimmed packings, tailstock pressure adds its own deflection by distorting the long flimsy workpiece.
The deflections make it difficult to turn a truly round journal to a precise diameter using what is essentially has to be parting tool. They also affect the centerlines of the rod journals' axes and, potentially, their alignment with the main axis. The exact locations of the rod journals' axes isn't a major issue, but they need to be parallel to the main axis to prevent the pistons from binding in their bores.
After semi-roughing a pair of rod journals, I was satisfied with their measured .001" TIR. I continued on and performed the same operations in different setups on the other journals but then discovered the TIR of the first pair had increased to some .010". The subsequent machining operations had apparently changed the workpiece and its reaction to the tailstock. In order to turn the rod journals to their finished diameters, the finishing operation had also to move their axes some .005" in the presence of a .003" workpiece deflection.
Maintaining a consistent tailstock pressure between setups (and throughout an operation in a given setup) can be tricky since the way the workpiece reacts to tailstock pressure changes as material is removed from it. For example, after finishing all the journals, I needed to remove excess material from the nose of the crankshaft so I could trial fit the crankshaft inside the crankcase. This involved turning down a one inch long length to just under a half inch in diameter which I knew would weaken the workpiece and allow it to distort. I watched runout developing in the main journals as more and material was removed from the nose. When finished, the nose ran true, but the TIR of the main journals were now at .005".
Along the way I discovered that centrifugal force on the packed but unbalanced workpiece created an additional deflection. To get around this and its accompanying surface finish problems, the maximum spindle speed was limited to 50 rpm. Some of the final .001" passes on the second crankshaft were actually performed while rotating the spindle by hand.
One of the photos shows the new grooving tool purchased for all journal operations on the second crank. Its style is identical to the one used to machine the first crank's rod journals (as well as the Merlin's), but its narrower .155" width allows it to fit in the narrow space between the cheeks of the Offy's main journals. It was also bifurcated and lapped on a diamond plate until its cutting edge was keen enough to take a consistent .001" depth of cut on a steel test rod. The corners of its chip breaker also had to be lapped flat to reduce the cutter's tendency to dig in during side-to-side cutting. In use, the depth of cut was limited to a maximum .005" (diameter), and the cutting edge was touched up and re-indicated for each journal.
The new workpiece was prepared identically to the first one including the end spigots, reference flats and center-drills. The journals on the main axis were again roughed out on the mill while the workpiece was supported in the vise. Ten-sided polygons instead of six were used this time in order to reduce the trauma to the workpiece during the interrupted cuts. This time I also roughed-in the surfaces for the front and rear ball bearings at the same time.
The workpiece was set up on centers in the lathe and its o.d. immediately skimmed. Left overnight, its TIR creeped up from essentially zero to .0015". All five bearing surfaces on the main axis were then semi-finished to within .030" of their final values using the new tool, and the cheeks were faced to their finished values. The measured TIRs were essentially zero.
The workpiece was returned to the mill where the rod journals were roughed in. After completion, the workpiece was temporarily re-installed in the lathe where I noticed the main journal TIR's had increased by a couple more thousandths. The rod journals were then semi-finished, and their cheeks faced to their final thickness.
On the following day the TIRs were rechecked. The main journals now measured between .005" and .007". The #2 and #3 rod journals were at some .012", and the #1 and #4 rod journals measured .009". All journals were still round to within less than a thousandth and so the entire TIR increase was attributed to workpiece distortion in response to its previous machining.
I allowed the workpiece rest for a couple days in hope that it might self-heal but no joy. The workpiece was moved back to the mill where the final cheek profiles were machined. When the journal TIRs were later remeasured, there didn't appear to be any further changes.
I should add that rod journal TIR's aren't obvious unless they're measured with a dial indicator. Runout created by a journal that's out of round by only a few thousandths will be irritatingly visible to the naked eye, but 10X that amount due to a displaced rod journal axis won't even be noticed. If one could be sure this displacement doesn't also include a skewed axis, its only effect would be just a degree or so change in valve timing and not worth worrying about. - Terry
Without grinding capability, the Offy's eight inch long crankshaft with its eight half-inch journals is a difficult part to accurately machine. In comparison, the Merlin's crankshaft with its three-quarter inch journals was considerably more rigid and much less troublesome. Even with rod journal packing, .006" deflections could be demonstrated with only modest finger pressure on the center of the Offy's crank. A drawing shows the deflections measured in four different directions with finger pressure on one of the center rod journals with the other three packed. A journal that must be unpacked for machining appeared to be responsible about half of the worst-case deflection. I was able to reduce the .006" deflection shown in the drawing by about half by packing the main journals as well. Even with carefully shimmed packings, tailstock pressure adds its own deflection by distorting the long flimsy workpiece.
The deflections make it difficult to turn a truly round journal to a precise diameter using what is essentially has to be parting tool. They also affect the centerlines of the rod journals' axes and, potentially, their alignment with the main axis. The exact locations of the rod journals' axes isn't a major issue, but they need to be parallel to the main axis to prevent the pistons from binding in their bores.
After semi-roughing a pair of rod journals, I was satisfied with their measured .001" TIR. I continued on and performed the same operations in different setups on the other journals but then discovered the TIR of the first pair had increased to some .010". The subsequent machining operations had apparently changed the workpiece and its reaction to the tailstock. In order to turn the rod journals to their finished diameters, the finishing operation had also to move their axes some .005" in the presence of a .003" workpiece deflection.
Maintaining a consistent tailstock pressure between setups (and throughout an operation in a given setup) can be tricky since the way the workpiece reacts to tailstock pressure changes as material is removed from it. For example, after finishing all the journals, I needed to remove excess material from the nose of the crankshaft so I could trial fit the crankshaft inside the crankcase. This involved turning down a one inch long length to just under a half inch in diameter which I knew would weaken the workpiece and allow it to distort. I watched runout developing in the main journals as more and material was removed from the nose. When finished, the nose ran true, but the TIR of the main journals were now at .005".
Along the way I discovered that centrifugal force on the packed but unbalanced workpiece created an additional deflection. To get around this and its accompanying surface finish problems, the maximum spindle speed was limited to 50 rpm. Some of the final .001" passes on the second crankshaft were actually performed while rotating the spindle by hand.
One of the photos shows the new grooving tool purchased for all journal operations on the second crank. Its style is identical to the one used to machine the first crank's rod journals (as well as the Merlin's), but its narrower .155" width allows it to fit in the narrow space between the cheeks of the Offy's main journals. It was also bifurcated and lapped on a diamond plate until its cutting edge was keen enough to take a consistent .001" depth of cut on a steel test rod. The corners of its chip breaker also had to be lapped flat to reduce the cutter's tendency to dig in during side-to-side cutting. In use, the depth of cut was limited to a maximum .005" (diameter), and the cutting edge was touched up and re-indicated for each journal.
The new workpiece was prepared identically to the first one including the end spigots, reference flats and center-drills. The journals on the main axis were again roughed out on the mill while the workpiece was supported in the vise. Ten-sided polygons instead of six were used this time in order to reduce the trauma to the workpiece during the interrupted cuts. This time I also roughed-in the surfaces for the front and rear ball bearings at the same time.
The workpiece was set up on centers in the lathe and its o.d. immediately skimmed. Left overnight, its TIR creeped up from essentially zero to .0015". All five bearing surfaces on the main axis were then semi-finished to within .030" of their final values using the new tool, and the cheeks were faced to their finished values. The measured TIRs were essentially zero.
The workpiece was returned to the mill where the rod journals were roughed in. After completion, the workpiece was temporarily re-installed in the lathe where I noticed the main journal TIR's had increased by a couple more thousandths. The rod journals were then semi-finished, and their cheeks faced to their final thickness.
On the following day the TIRs were rechecked. The main journals now measured between .005" and .007". The #2 and #3 rod journals were at some .012", and the #1 and #4 rod journals measured .009". All journals were still round to within less than a thousandth and so the entire TIR increase was attributed to workpiece distortion in response to its previous machining.
I allowed the workpiece rest for a couple days in hope that it might self-heal but no joy. The workpiece was moved back to the mill where the final cheek profiles were machined. When the journal TIRs were later remeasured, there didn't appear to be any further changes.
I should add that rod journal TIR's aren't obvious unless they're measured with a dial indicator. Runout created by a journal that's out of round by only a few thousandths will be irritatingly visible to the naked eye, but 10X that amount due to a displaced rod journal axis won't even be noticed. If one could be sure this displacement doesn't also include a skewed axis, its only effect would be just a degree or so change in valve timing and not worth worrying about. - Terry
