It looks like the wheel case and its components will end up being the most difficult assembly in this build. There are so many shafts, countershafts, and shafts inside other shafts that the interactions among all the various gears make it difficult to come up with a construction sequence. John has patiently answered some of my dumb bunny questions through email, but the documentation could sure use some sub-assembly drawings. I took a couple drawings with me to Cabin Fever so I could study them safely away from the shop and its temptation to start cutting too early.
I eventually decided to continue construction with the supercharger bearing plate. This scratch-machined plate will support the rear bearings for the main rear shaft as well as the countershaft that drives the supercharger. When completed there will be 10 shafts, 15 bearings, and 14 gears packed inside the wheel case in front of this plate, and they will all interact with the two shafts supported by it. So, it's important that its bearings be accurately located with respect to the crankshaft centerline and, eventually, the axes of all the other shafts as well. It didn't seem practical to attempt to match-machine much of this plate to the wheel case since the important features on both parts face one another and are aren't simultaneously accessible.
There are mixtures of spur, bevel, and helical gears inside the wheel case. I had naively planned to make all the gears for this engine. But, after gaining some appreciation for why the wheel case is called what it is, I felt the learning curves for the bevel and crossed-axis helical gears would be best left for another project. Dealing with all the shop-made cutters and the tooth approximations I'd likely have to make while trying to maintain all the interacting center-to-center running distances was just too much to pile on top of what is already ahead. I still plan to machine the spur gears, but I placed an order for the bevel and helical gears from a gear manufacturer in England that was recommended by the documentation. In terms of the distances we travel here in Texas, that gear factory is likely located next door to the original Merlin engine plant.
When I received the involute cutters purchased for this project back in June of last year I spot tested the tooth cutting profiles of four of the cutters by cutting two pairs of test gears and checking their meshes at their theoretical center-to-center running distances. Most gear cutters that I've purchased are imported, and I've found their quality to be very inconsistent. The 32-pitch gears I made were for three of the shafts in the wheel case; but at the time I was only interested in verifying the tooth profiles, and so I didn't complete their machining. The 50/24 tooth gear-pair ran fine with no binding and minimal backlash at its theoretical 1.156" center-to-center distance, and so I put the cutters away for later use. I decided to install the gear-set back into its test fixture to see what kind of errors I'd be willing to accept inside the wheel case. This particular pair ran with noticeable drag but no binding with a spacing of 1.148", and at 1.164" the backlash was too great for my taste. My plan, of course, will be to aim for the theoretical center spacings for the spur gear pairs, but I'll use measured spacings for the bevel and helical gears once I receive and test them. I felt that a +/-.004" error, if necessary, would probably be acceptable on all but the high rpm supercharger gear set.
I needed a consistent way of supporting the wheel case for the numerous machining operations to come; and so I mounted it, front flange down, to a faced 1/2" flat plate. I then machined two opposite ends of this plate parallel with the top surface of the chain cover and the other two ends were machined perpendicular to it. When completed, I had the wheel case mounted to a support plate that was square and aligned to its reference top edge within a few tenths.
The next operation on the wheel case was to drill and tap the seventeen 2-56 mounting holes for the bearing plate. The mounting holes were individually located in the centers of their cast blind bosses on the wheel case mounting flange, and their coordinates were recorded for use in SolidWorks where the bearing plate was laid out. The locations of the rough-cast front bearing pockets for the same two shafts in the wheel case as well as the parallel starter countershaft were also measured in the same set-up. Circular features in these castings have been remarkably accurate. A dial indicator showed these 'rough cast' bearing pockets, were within a thousandth or so of being perfectly circular. They are, of course, intended to be machined to finished dimensions.
Their measured locations, however, disagreed with those in the drawing by about .020". This would be enough to bind the gears on all three shafts and is roughly the same size error I discovered in the wheel case's centerline when I machined the chain cover. This was a big concern because adjusting their locations will affect the gears on the other shafts, and the whole avalanche of possible consequences was too much for me to follow especially since I don't yet have most of the gears to play with.
At first I thought all of this might have been caused by me when I initially squared up the warped wheel case casting. But, further measurements showed a possible interference problem between a large countershaft gear and the wall of the wheel case. The complex shape of the interior walls put my measurements into question, and so I decided to double-check the fit by cutting a simple gear set and installing it on a pair of temporary snug-fitting Delrin shafts that I pressed into the cast bearing pockets. Sure enough, the gear was not centered between the walls of the wheel case and, in fact, was jammed against one side.
In the entire lot of castings this was the first casting error that I had run into that hadn't been flagged in the documentation, and so I spent many hours convincing myself that it was real. I spent even more time experimenting with new locations for the bearing pockets that would eliminate the interference, provide the correct center-to-center running distances for the gears on the three parallel shafts and minimally impact the other shafts. Unfortunately my test gear on the starter countershaft had only .007" clearance to the wheel case, and every fix I came up with moved it even further into the wall. In the end, I could only guess about the ripple-down effects on the various cross-shafts and their gears.
Finally, with a best-effort drawing to work from, I began construction on the bearing plate by chucking a slice of 6061 in my lathe's 3-jaw. Since I was starting with a piece of scrap that was only marginally thicker than the finished dimension of the plate, I cut a plexiglass spacer to help accurately support the material in the chuck. I've found that plexiglass works well for fixtures since it cuts easily on a bandsaw, and its thickness is extremely uniform.
The o.d. and the mating flange of the bearing plate blank were turned for a snug tested-fit to the wheel case bore, and the bearing pocket for the central main shaft was bored. The blank was then flipped around and mounted in a collet chuck where its face was indicated-in before being turned to its finished thickness.
The blank was then secured, rear side up, to a piece of MDF with a close fitting plug pressed into the center bearing pocket so it could be indicated-in on the mill. The clearance holes for the flange mounting bolts were spotted and drilled using the measured wheel case coordinates. Button head screws were added through the flange mounting holes to further secure the perimeter of the blank to the MDF. The blank was shimmed and checked for z-height consistency across its face to insure the bearing pocket for the countershaft was bored perpendicular to the bearing plate. Excess material was then removed from the rear of the plate where a stiffening brace and some inspection holes were machined. The bearing pocket for the countershaft was bored with a boring bar instead of being interpolated. Finally, the 1-72 mounting holes for the two bearing retainers were drilled and tapped.
It was this last operation that turned around and bit me. A mysterious errant move of my Tormach's spindle gouged the newly machined brace on the plate as it moved into position to start drilling the retainer holes. I was able to modify the design of the brace to clean up the damage and re-reference the workpiece, but the source of the hiccup was never found. Unfortunately, I had also managed to place two of the eight bearing retainer holes in the wrong locations thanks to the overly cluttered model I had been using to investigate the bearing relocations.
After assembling a pair of newly designed bearing retainers I discovered that one of the gears in front of the bearing plate will end up so close to one of the .060" thick retainers retainers and its 1-72 button head screws that it will likely rub. So, I redesigned the bearing retainers to use .032" thick brass plate and flat head 0-80's. This meant that I had to re-fixture the bearing plate and re-drill and re-tap it for the new retainers.
In the end, the bearing plate fits snugly into the wheel case, and all seventeen mounting holes line up perfectly. The new bearing retainers hide my drilling errors, or I would have started over on a new plate. However, I think I may have gotten only a taste of things to come with this wheel case. There be dragons hidin' in there. - Terry