In all this I have noticed the two seating tabs on the yoke (external clip style) are different in the machined areas.
One tab is relatively large, with plenty of material and a defined machined seating area.
The other tab looks to be machined down past its half way mark leaving behind a much smaller tab when compared to the other side.
If I'm understanding your description correctly; the physical retention "tab" within the yoke's u-joint cap saddle relief radius as machined is not necessarily going to be consistent in width as the outside dimension is controlled by the yokes' as-forged exterior surface profile. Here, a good quick test is to take a known good or new proper sized u-joint assembly (caps including the cross-trunnion) and just lay/press it into it's intended seated position with the observation of whether the length of the u-joint is tight between the two "tabs", and also fits tightly in the radius.
Beyond just poor previous (monkey
) efforts, and just wrong parts assembly causing damage, with some often not so ingenious fixes, often if the u-joint becomes excessively worn, and particularly if one or both of the retaining u-bolts were not torque adequately at some point in the past, this can lead to the caps in the assembly slipping and causing a hammering effect, which will given time displace material causing excessive resultant distance to be presented between the two tabs, and allowing the entire driveline assembly to be displaced off center.
The more I look into this yoke the more it appears it was machined about .022" off to one side.
If the yoke is off by that much, and the driveshaft is off by a bit too, but still in tolerance, the combined errors may be where my vibration resides.
These values as you have presented if accurate are way out of acceptable sums; something closer to .003" deviation would be expected, in excess of say .006" begin to expect vibratory effects, .012" or more, your just wasting your time!
Acquire a dial-indicator, fixture it to the rear-end housing (not something that can change position in the relationship
), mount a complete u-joint assembly (one that has proven to 'register' properly as described previously
) and indicate off the machined stop surface of the cap exposed, rotating the yoke repeatedly until one is confident with the process and the execution with repeatability.
Scott.