In order to turn the liners' o.d. features, they were supported by their bores on an expanding collet. Although many builders routinely make such mandrels, I opted to use a commercial import that threads into a 5C collet chuck. Once installed, it was turned to fit the liner bores and then left in place on the lathe until all the machining on all the liners was completed. Before turning the soft collet for a close slip fit inside the bores, it was important to place the collet segments under a bit of outward pressure using their expanding screw. The total available adjustment range was only some ten thousandths, and since this corresponded to about a quarter turn, an initial loading of 1/16 turn was sufficient.
Unfortunately, the runout measured at the mandrel's tailstock end varied as much as a couple thousandths when the screw was adjusted for a different diameter. The workpiece couldn't be arbitrarily taken off and on the mandrel nor flipped end-for-end between operations intended for perfectly concentric features (maybe because the bores hadn't yet been honed as 'Bigrigbri' suggested.) Since this was how I planned to use the mandrel, I had to re-indicate the workpiece each time it was remounted. This involved rotating the workpiece on the mandrel, tightening the expanding screw, indicating the o.d. used as a reference and then repeating the steps until the minimum TIR was found. Fortunately, it was always within a half thousandth or so. There were only a few features on the liners that needed to be truly concentric, but I chose to re-indicate each workpiece before every operation. Witness marks added to the i.d. of each liner with a marking pen eventually eliminated the tedious process.
One of the sketches shows the liners' features that required machining. For consistency, all the liners were machined in batches one feature at a time. A real risk with this approach, though, was that an error could have propagated throughout the whole batch of parts before being discovered. Features one and three are the most critical, and these must be concentric with each other. Feature one is the lower half of the combustion chamber seal, and number three is the lower half of the coolant jacket's upper seal.
Feature one was the first turning operation, and it also became the reference o.d. for indicating all subsequent operations. Counterbores in the heads' combustion chambers will seal against the top ends of the liners which sit above the cylinder block. These seals rely on an important machining detail described by a pair of notes - "no sharp inside corners" and ".010 R Fillet Typ" - on the drawing for the liner. I've included a close-up sketch of the area of interest for a typical seal.
A 1.405" diameter counterbore, concentric with each combustion chamber, was machined into the heads much earlier. The top end of each liner was specified to be turned with an outside diameter of 1.400" as well as a .010" radius fillet in its outside corner. It's very important that the ends of all six liners sit at the exact same height above the top surface of the cylinder block. When the head is bolted down to the block, the liners are sandwiched between them, and the sharp edge of the counterbores are forced into the liners' outside fillets to create the seals. Modeling showed there should be a .003" crush height available for deforming the soft edge of the aluminum counterbore into the hard corner fillet of the liner. Since I turned the ends of my liners using a standard lathe insert with a .008" nose radius instead of grinding a custom .010" radius cutter, I changed the outer diameter of the liners' top ends to 1.402" to obtain the same .003" crush height.
The integrity of the seals also relies on machining that was done over a year ago when the head and cylinder block castings were bored. During those operations it was important to accurately machine all the counterbores to the same diameter and spacing to match the bores in the cylinder blocks. I was anal about machining all the counterbores on both heads to within a few tenths of one another so a common end diameter could be later used for all the liners. Back in post #61, just after completing the head and cylinder block machining, I turned a set of close-fitting Delrin fixtures which simulate a portion of the liners including their top ends and shoulders. I used them to trial assemble the head and cylinder block pairs just after machining them so I could verify the counterbores in the heads were actually aligned with the cylinder bores in the blocks. This had been one of my 'must have' machining milestones in this build.
After the top end of the first liner was machined, it was lightly set into each head counterbore so the crush distance could be checked with a feeler gage. Depending upon the particular counterbore used for the measurement, the crush heights ranged from zero to something under .001", but most were close to zero. The explanation for this that I took away from the modeling was that even though the 'sharp' edges of the counterbores had never been intentionally broken, they likely ended up with slight chamfers up to .004" in width. I turned this into an opportunity to improve the seal by turning two different radii fillets at two different depths on the liners. I've included a drawing from my modeling to illustrate just what I actually did. Basically, a typical counterbore's chamfered edge fits nicely into the .008" radius fillet, and the .016" fillet provides a additional crush height of .002". These fillet radii match the nose radii of standard lathe inserts. The crush heights of the finally machined parts measured between .0015" and .002".
The outside diameter of each blank was next turned down to 1.500" which became the final o.d. of the liner's top shoulder as well as its largest diameter feature. A fixture was then machined for the headstock end of the mandrel to use as a positive z-axis stop. The blanks were then reversed on the mandrel with their finished top ends tightly against this stop so each could be faced to the exact same final length. This stop was used to reference the z-axis for the rest of the operations.
The coolant jacket was then machined using a DCMT insert. This 55 degree rhomboid cutter simplified the turning operation and left good looking tapered edges at the ends of the jackets. The recessed jacket provided a nice tool entry point for finish machining the coolant jacket's upper seal using a grooving tool modified with a bit of relief so it could be used for very light side-to-side turning.
The number three feature is the lower half of the coolant jacket's top seal. The liner diameter just below the shoulder was specified to have a close slip fit in the cylinder block, but since it's not a press fit, it's not a functional seal either. Instead, the seal is formed between the top deck surface of the cylinder block and the bottom surface of the liner's shoulder. The 'sharp corner-into-a-fillet' trick can't be used for this seal because it might have created inconsistencies in the heights of the shoulders and issues for the combustion chamber seals. Instead, the upper coolant seals rely upon a pair of smooth flat surfaces pressed together by the sandwiching forces of the head when torqued down to the cylinder block. The integrity of this seal can be affected by the parallel alignment of the liner inside the bore of the block since it affects the parallelism of the sealing surfaces. This alignment is influenced by the lower coolant jacket's compressed o-ring as well as the liner collar support hardware between the cylinder block and crankcase.
As far as I can tell from my research, the liners' coolant jackets in the full-size engine were similarly sealed. There were certainly plenty of documented problems with leaks, but evidently they were eventually solved. With the change to the liner's wall thickness that I've made, there is now sufficient space to add a -027 Viton o-ring as a back-up seal. With this modification, the combustion chamber becomes the only metal-to-metal seal. Adding this o-ring may increase the assembly difficulty involved with getting the liner into the cylinder block, though. Some very preliminary tests have shown the assembly may already be fairly difficult because of the need to compress and stuff the lower o-ring into the bottom end of the assembly. Before making a decision about using the upper o-ring, I'm going to machine the liners' lower support hardware and do some actual assembly testing using a single modified liner. I don't want the o-ring groove to become one of those mistakes that propagates through the entire batch of finished parts.
The last feature to be machined was a 60 degree chamfer that I added to the bottom ends of the liners. This chamfer will provide rod clearance near the bottoms of my thicker-walled liners, but more importantly it will hopefully ease the insertion of the ringed pistons into the liners. In the photos I've seen of the full-size engine's assembly, each entire cylinder block will have to be inserted down onto the crankcase over all six ringed pistons simultaneously. - Terry.
