The Feb/March 2005 issue of OTL had an article on clutch splines where I suggested some reasons for the clutch spline failures on some 6-speed boxers. Misalignment between the engine and transmission was a possible cause, and I illustrated a few ways in which misalignment might be present. In the last two years I've inspected some bikes which experienced such failure, trying to find a common cause. What I'm presenting here is my opinion based on my observations.
Let's take a look at two of the diagrams I made:
A lateral misalignment would probably result from mismachining of the engine case or the transmission housing. Here, the input shaft is too high (or the crankshaft is too low). One side of the hub would be pressing against the shaft and the other side wouldn't. The load would be distributed across the entire side of the hub, and as the parts rotated all of the surfaces would bear the load at some point. A certain degree of this misalignment is normal, due to manufacturing tolerances.
This was suspected in a series of failures (on the same bike) in a case on bmwsporttouring.com. As a result, I started measuring every bike I could for this misalignment. I still do, but so far I have yet to find one where this alignment is suspect. One problem with this measurement (where the axis of the crankshaft is compared to the center of the bore for the input shaft in the transmission) is that the crankshaft floats in the crankcase bearings with up to 0.1mm clearance. When the engine is not turning quickly, the crankshaft isn't centered in the bore because there are other forces (gravity, alternator belt tension, etc.) on it. I always find the crankshaft to be lower than it should be, but it can be lifted so that it is higher than it should be. It's a very difficult measurement.
A bent housing (also referred to as the clutch carrier) will cause the hub of the friction disk to try to rock on the input shaft. The load will change to a different location every time the clutch is engaged. When the clutch housing replaced the flywheel in 1981, there were problems like this.
BMW doesn't publish specifications for the required alignment of the housing. Before 1981, though, flywheel runout specifications were published and the limit was 0.10mm. Most modern housings I have measured have about that much runout at the three clutch mounting tabs.
In one particular case, a bike was trailered to me after its second spline failure. I measured everything I could, and found no lateral misalignment. Clutch housing runout, though, was noticeable. Of the three mounting ears for the clutch; if the first was considered to be the standard then the second was 0.12mm off and the third was 0.20mm off. Inspection of the clutch surfaces showed that the clutch disk was overheating one area of the pressure plate and cover, which is consistent with misaligment within the clutch...
This is about 25% of the full article, which is published in the June 2007 OTL in its entirety.