Ahgghghh damnit Joe I had a real long comment drafted, figured it merited getting on the forums, didn't remember my password, had to reset it, check my handful of email accounts to find whichi email tied to my user, and reset it. By which point you had addressed some of my comments. It's still on my clipboard so I'm going to paste it despite you addressing things like cycling.
A plug spring, is wayy more elegant than what I was thinking. Remember buddies, k is only constant within the linear region, you'll go into the plastic (that is, region on the stress/strain curve, not material) failure zone well before you see a fracture failure (just like metal - the permanant deformation region of yield comes before the fracture). and game over. But yeah, that's really a insanely elegant idea (I think) since even if one of the two springs approaches permanent deformation, the other one will compensate? I think? Not even close to someone in materials science. My experience with springs are limited to legos and struct coils and..I think that's it so, grain of salt. Or, maybe it will just introduce 2^n variables, since you now have to material combinations[1] with two diameters, with different loading profiles. Someone in materials or metals should jump in here.
Adding two metals in a plug form and playing around with them will almost certainly keep you in the elastic zone (say 1.5 to 10 as k performs 100% linear, for at least a few hundred cycles. I bet you can get a bag of chinese springs for a couple bucks on ebay, and benefit entirely from their lack of quality control. Then you can play around with the material combinations, and mayb get you something which will perform for a few thousand cycles before the fatigue profile changes.
Just a tangential point: Fictive.com has a service that uses something similar to a sintered metal process which will 3d print your Solidworks models (I think, I don't do all that fancy art stuff). It';ll likely be way out of the 'cost/result' region, but it'd be really really fun to play around with, which I presume is what most of us are here (edit: this was a comment from Joe's original youtube thread)
People are 3d printing custom intake headers for performance cars with the entry/mid level (500k-ish?- no idea at all- the platform looked no bigger than 24x36) metal printing. Beforehand, the company has to manually work with you/do a lot of manual geometry analysis (since I think their production volume at this point is so low that it hasn't merited automation yet, some brief FEA just because COMSOL is super cheap, compared to a field failure -> customer&reputation loss. I digress - 3d printing green-> sintered metals is going to be suppppppper cool. I expect to see hackaday post of any day now from some dude who "borrowed" metal grains from work
--------
[1] You benefit from the material variance to for *your* setup. But the inconsistency, would you go into production, would introduce so many variables that you'd be banging your head at the table. Especially because you don't know which component is responsble of the variance in the assembly. At which point, just bring in COMSOL and order from some place that has batch reports for tracibility-- so there's that downside of Chinese materials-- you now-- the whole tracability thing that any competent manufacturer will deliver to you in the packing slip