Final version: http://www.thingiverse.com/thing:591203
what it is:
This chain holds together purely due to its shape.
That is the axles aren't prevented from sliding out due to friction, spring, magnet or some other force but solely due to the geometry of the chain segments.
Only the very last segment that ties the chain to a loop
needs to introduce one single snap together location that breaks this rule.
The chain can only be opened beginning at this location and can be only opened sequentially in one direction.
Also this is "hierarchical locking demo II"
an improved and somewhat more useful version of
"hierarchical locking demo I "
the problem with this model:
1) there's not enough clearance to fit the central spacers in and without them the chain can fall apart.
2) the caps can bend outward too much providing the chain another way to fall apart.
how it works & history
cla ... chain link assembly <<
the "cla" subsumes
1) the main-chain-link-part
2) the retainer-part (not the one on the side but the one in line)
3) the optional roller part
The trick lies in the retainer part.
When you add cla(n+1) to cla(n)
the retainer part of cla(n+1) locks the axle between between cla(n) and cla(n-1) into place while still allowing you to insert te axle between cla(n+1) and cla(n).
It would also be possible to fuse the retainer-part to the main-chain-link-body preventing it from matching its rotation to the preceding main-chain-link.
This would require the retainer body to fan out though
a) making it unprintable without bad overhangs
b) making the chain really fat and limiting the chain bending angle below 90°
c) bloacking the possibility to attach utiliarian stuff to the links
(I did that fanout in three precieding failed attempts.)
In atomically precise small scale factories chains can play a crucial role for transport of molecule fragments and molecular building blocks.
If a diamondoid crystalline-molecule chain at the lower physical size limit is only hold together by Van der Waals forces then thermal fluctuations pose a threat for the chain to break open at any time and at any point. Especially in an envirounment with elevated temperature.
The risk of accidental thermal induced chain opening could be prevented with axle caps that are fused on with covalent bonds. But then the links can't be taken apart ever again. Then the chain can't be changed in length and can't be rid of links that are not broken but damaged in other form (e.g. broken attached tooltips). Or if different shaped links are used it cant be remixed.
Using positive locking / form closure combines resilience against opening and the freedome of recomposition in a very satisfying way.
This chain is designed with the
"samesize detail rule for bulk limit nanodesign"
in mind which is also good for 3D printers.
possible extensions and improvenments
A related exercise would be to modify this design (or create an alternate one)
such that multiple chains can be put together side by side in a similar shape lock style.
This would make it possible that the single weak points are e.g. located on opposite ends of the chain. That means if the first links breaks open the chain wont come apart but instead has a good chance of closing up again.
The total number of closing points that are held togehter only be weak small-area-VdW-force then amounts to only two no matter how long or wide the chain is made.
Lay your fingers inbetween the fingers of your other hand to see roughly what I mean.
Chains a few segments wide allow for stronger bigger-area-VdW-force
Then the chain becomes effectively thermally unopenable below completely destructive melting temperatures.
Print at least eight full sets to make a somewhat interesting loop.
Use two colors - this makes it easier to understand how the chain works.
Note that this design is far from perfect. Before it's closed up It's pretty wobbly. The good side of this is that you can close the loop by force although it shouldn't be possible to connect the last link with the first.
Closed up it becomes sturdy enough that you can drop it from a table.