I wrote:
> It's pretty unusual for a spring to lose tension. Contrary to popular
> opinion, proper springs don't "go soft" from use -- ever.
Someone asked me off-list (so I'll preserve his anonymity):
> So why do almost all of my Shimano Cranes have a limp top spring?
It is possible that spring was very poorly designed and/or made, or more likely they were damaged from being stressed beyond what their designer envisioned.
Let's limit this to steel springs 'cuz that's what I know, and that's what derailleurs almost always have. Other metals are mostly the same though.
The "springiness" (spring constant) of a steel spring, operating below the yield point, is determined by the shape (length, cross-section etc) and by the Modulus of Elasticity ("Young's Modulus) which for steels is nearly a constant at 30 million psi. (Some steels, notably stainless, are a bit lower than that, but you're not too far off to say all steels have the same modulus regardless of alloying, heat treat state etc.)
The total force applied by the spring depends on the spring constant together with the pre-load, or distance the spring has been deflected from its rest position.
Since the sectional shape of the spring (thickness of the wire) didn't change in use, and the modulus didn't change (and thus the spring constant didn't change), then what did change? The spring may have been stressed beyond its yield point, and "taken a set". This changes the rest position of the two ends of the spring, and thus changed the pre-load at any given point in the rotation.
Proper steel springs generally have enough carbon in them to be heat-treatable, and are heat-treated to be strong enough to never reach their yield stress in normal use. If a spring bends permanently (takes a set) in normal use, then it was a rather poor spring. That's an understatement; more like an embarrassment to its maker really. That should never happen. Any chance the Cranes you're talking about got rotated well beyond the range they're supposed to go? They have a tab near the upper pivot that's supposed to keep them from rotating too far, but I have seen Cranes where that tab was bent or even broken off, from being hit so forcefully that the tab moved out of the way and let the derailleur rotate too far. I think maybe getting the chain caught in the spokes can do this. If the rotation went beyond what the designer of the spring thought was possible in use, then it could be permanently bent.
The good news is: bent springs can often (I'd say usually) be bent back. Take the spring out, grab the tail ends with some vise-grips or some such, and rotate it far enough in the opposite direction to make it take a new set, back where it needs to be preload-wise. That's often as good as new.
Unfortunately, "good as new" wasn't good enough in the first place, was it? So you might just bend it again, the next time you over-rotate it. Talk to Shimano about that (and I wish you luck with that)...
Mark Bulgier
Seattle, WA USA