How a 105 LSD (aka ZF LSD) works (potted
Below is a parts picture of the slippery bit of the diff.
Essentially, there's a multi plate clutch for each rear wheel.
Assuming the wheels are on the ground & have traction, drive arrives via the tailshaft, through the diff & into the limited slip clutch part.
The drive ring (part #4) moves out to clamp the clutch plates (part #s 2 & 3) to transmit drive to the axle/wheel. The outward movement of the drive ring (#4) is accomplished via the ramps (you can see these as cutouts on the inner sides of each of the drive rings (#4).
If a wheel loses traction the drive ring loosens its clamping of the clutch pack & drive is reduced/removed from that wheel.
There's a whole bunch of wearing bits in there, but the most important (in this instance) is the clutch plates. As the clutch plates wear & get thinner the drive ring has to move further out to clamp the clutch plates. Because that outward movement is accomplished via the drive ring ramps it uses more rotational movement to do this. But, within reason, it will eventually clamp the clutch plates & transmit drive.
This shows up as scary amount of "freeplay" in the tailshaft rotation before the diff takes up the slack caused by the thinner clutch plates.
Hope that helps (& makes sense, I've been to a party & might have had a drink or 2)
edit to add: It's not that weird a method of LSDing, Hewland used a similar system in its racing tranaxles for many years. It was similar enough that there were court battles between ZF & Hewland over who owned the design.
Yes I understand all that but what I don't get is what supports the inner ends of the axles. There must surely be diff side bearings(?) and they should support the axles not the clutch plates. The clutch plates should only be for torque transfer.edit to add: It's not that weird a method of LSDing, Hewland used a similar system in its racing tranaxles for many years. It was similar enough that there were court battles between ZF & Hewland over who owned the design.