Originally Posted by STI_FFY
I'm glad you said "potentially", otherwise your conclusion would not be logically sound.
To eliminate uncertainty you would first have to determine how much the measured deflection actually translates into additional clearance, since the two are almost certainly not directly proportional. Assuming it does, you would then have to establish that Subaru did not take that additional clearance into account when specifying the gear lash.
From a purely academic standpoint your results are interesting. I wonder what this looks like for the 6MT, which is considered by many to be bulletproof. You wouldn't happen to have one of those lying around?
Exactly... It also depends on where I measured. The spot at which I measured the largest amount of dimensional change is probably a pretty good indicator of lash between the ring and pinion gears though. Most manufacturers spec anywhere between +/- .001" and +/- .003" for their tolerance on gear lash and R&P clearance regardless of case design... Perhaps this is why Subaru's spec is a narrow range (only 1 thousandth in either direction), they might have been expecting a particular amount of flex and set the tolerance in a more narrow range to make up for tolerance stacking between the case and gears.
Hey if anybody has a 6mt they would like to donate, I'm all for it... Though I doubt I'll get too many offers
Originally Posted by tom@heartboy
i would try and beef up that fixture. i think a lot of the measured flex might be in the fixture itself. get some good 4x4 1/4 wall tubing with some larger mounts. i dont think galvanized yard tubing mig welded together would be strong enough.
< spent about 10 years testing composite/metal samples to failure.
I'm glad you brought that up... Actually the fixture flexed less than I expected, though it did have some give. I was waiting for a sharp guy to come along and point this out
I'll explain how I got around this problem.
Here's the foundational concept behind my measurement technique: ALL measurements were taken in relation to bolt locations on the bell housing of the case. The dial gauge and it's mount point were not in any way contacting the fixture and should not have picked up any flex in the fixture. If I wanted to take measurements relative to a fixed point off of the case, then the fixture would have had to be exceptionally stiff.
The disadvantage of course is that any flex in the bellhousing could also translate into additional movement of the dial gauge. Though my main point was to find peak areas of flex in the case and basically to prove that a measurable and significant amount of flex actually does in fact occur in the Subaru 5mt case rather than focusing on measurements taken against a fixed reference point away from the case.
The whole test fixture project is more a proof of concept and way to find the weakest points so I know where to apply reinforcement. I could build a very nice fixture for $400-500 and LOTS of labor time that would give me acceptably accurate measurement in relation to a fixed point, and that would give me essentially the same findings, but with a level of accuracy beyond what I need.
<-- Also professional experience testing products to destruction and 4 years in a metrological bachelor's degree program (Metrology, the science of measurement, not to be mistaken with metallurgy or meteorology).