With the crankshaft fully roughed-in the 'easy' part is over, and now it's all about discovering and dealing with all the tiny details that affect its final accuracy. As I get closer to the final finishing operations there's less and less stock available to protect me from my mistakes.
The rear of the crankshaft on the full-scale Merlin is internally splined for a rear driveshaft. Rather than risk the crankshaft with a difficult splining operation, the quarter scale model uses a separate internally splined coupler that's Loctite'd into a half inch bore in the rear of the crankshaft. The drawings show this coupler having a circular array of mounting holes, but the crankshaft drawing doesn't show a corresponding array of tapped holes. The recommended Loctite can probably handle the torque requirements of the magnetos and maybe even the supercharger, but I'm not sure it will be adequate for the starting torque requirements of an electric starter. So, I decided to drill/tap an array of holes in the rear end of the crankshaft for six 3-48 mounting screws. These holes should have been drilled much earlier in the workflow, but I've had my head down in the crankshaft drawing which I had been assuming was complete. I wasn't happy with the setup that I had to use at this late date to support the crankshaft for drilling, but it was adequate for the forces needed to drill the tiny holes.
I left excess stock on all the crankshaft webs, and now it must be removed before machining the counterbores at the ends of the bearings. These counterbores will eventually be plugged in order to seal up the pressurized oil passages between the main and crankpin journals. Since I can't trial fit the crankshaft into the crankcase, I built partial CAD models of my crankcase and the crankshaft workpiece in its current state using measurements of the actual bearing locations. This allowed me to overlay the two and determine the best way to distribute the removal of the excess web material. The center bearing establishes the forward/aft position of the crankshaft in the crankcase, and so I provided a .0015" clearance for it. I followed the drawing for the remaining six bearings which calls for .040" clearances. I used a radially mounted right-hand boring bar to finish the left sides of the webs. Rather than invert it and run my lathe in reverse as I did previously, I made a left-hand tool to finish the right sides. It started out as a boring bar for a triangular insert, but due to a measurement error I ended up using a trigon insert, instead. I also ground a form tool to machine a radius on the neck of the crankshaft just behind the front flange.
Now the real fun begins...
I had .070" excess material on the main bearings, and so I decided to experiment some more with my bifurcated cutter. Since I wasn't 100% happy even with my new insert, I honed its front edge on a diamond hone and added more rake to the insides of the Britnell notch. I think its edge is now as keen as anything I could have done using HSS. My results using the improved insert was a slightly smoother cutting action and an improved d.o.c. resolution to less than .001" while cutting near the supported ends of the crankshaft. This depth control will be important later when I try to finish all the bearings to a common diameter.
I still had some serious workpiece deflection while turning the three center bearings, however. Initially, I wasn't aware of it because the edge of the tool was hidden deep inside the webs and under a thick layer of the moly-based cutting oil I've been using. The symptom that alarmed me was a ridiculously high TIR of two to three thousandths on a freshly turned bearing. The deflection was created by the compromised rigidity of the workpiece which is a result of the now fully machined webs particularly around the offset crankpins. The real problem for a crankshaft is that this deflection can't be compensated by taking a heavier cut. The offset crankpins cause the workpiece's reaction to the cutting tool to be a function of the angle of rotation of the workpiece. This can be measured using a dial indicator on an adjacent bearing while cutting. The actual amount of deflection continuously changes as the workpiece rotated. At high rpm the symptom would have been a squeal as the cutter chattered against the workpiece. At my 50 rpm turning speed, though, it showed up as a non-circular turned bearing. Because of the limited space between the deep bearing webs it's very difficult to get accurate measurements, but I'm reasonably certain that I measured a couple thousandths difference between two almost orthogonal measurements taken on the diameter of the center bearing. This closely correspond to the unreasonable TIRs I measured just after turning the bearing. The bearing breathes instead of wobbling as it spins, and it's hard to notice a problem with the naked eye.
Some quick and dirty things I did to understand the problem included wood shims forced between the crankpin webs as well as a length of close fitting drill rod inserted through the hole in the center of the crankshaft. I was able to significantly affect the results but not cure the problem since I'm sure even the oak was reacting to the high cutting forces.
In any event, I completed the web machining, and this produced identical final finished spaces between the webs. By this time I had removed .020" of my .070" safety margin from the main bearings.
After re-turning the main bearings with the wood shims in place, I removed the workpiece from the lathe and then re-installed it to check the repeatability of the TIR measurements. I found the runout to be similar for all the bearings, but it had nearly doubled to an unbelievably value of .010". This result led me to believe I may have yet another problem with my headstock fixture and, in particular, the newly machined buttons. I suspected the center-drilled holes weren't deep enough and that my setscrews that bear against the spigot flats might be pushing the workpiece around slightly. I made new buttons with deeper center-drilled holes, and I re-turned the bearings with the wood shims in place and then repeated the test. Again, I measured almost the same ridiculously high run-out, and now I have only .030" safety stock remaining.
After spending a couple days doing various experiments but no additional cutting I concluded the issue related to the fixture was probably due to my method of tightening the setscrews. When I previously checked the consistency of the fixture I wasn't paying close attention to any particular tightening sequence. I had been just tightening the screws against the flats, but my workpiece at that time was a heavy chunk of rigid steel. Now the workpiece is much more flexible, and the setscrews need to be iteratively tightened so the workpiece is allowed to find its center with little or no net stress.
To check out these theories I machined a set of aluminum cervical collars which fit between the crankpin webs. These are not jammed into place but were lapped for a close sliding fit. I think it's important to not fixture the workpiece under any stress that can affect dimensions because after machining and when the part is removed from the fixture these dimensions will change. I Ty-Wrapped these collars in place, and for good measure I also inserted a length of close-fitting drill rod through the previously drilled hole through the center of the workpiece. (I could really use a steady rest behind the center of the crankshaft, but my lathe carriage is in the way, and there is no room for a follow rest.) I carefully tightened the setscrews to allow the workpiece to find its own position on the headstock center, and then I turned just enough material off the center bearing to clean up its TIR. I then removed the workpiece from the lathe so I could re-fixture it and re-check the TIR. Thankfully it now repeated.
And so, to answer Peter's earlier question about whether stiffening spacers are needed while turning a crankshaft, my answer is a resounding 'yes'. And thanks to Charles who warned me about being too quick to finish the most important surfaces on the part.
It's now obvious to me that there is a major advantage, besides surface finish, to grinding crankshaft journals rather than turning them. The grinding forces against the workpiece will be much less than a cutter's turning forces, and the resulting deflection even on a 10-1/2" long crankshaft would be nearly insignificant.
Now, I plan to finish up the last few semi-finishing operations before returning to the bearings. Hopefully, I've paid my dues and the problems are understood well enough to successfully finish up. - Terry