I'm not sure how the first alternative you suggest would work (I mean I don't know what transmission blocks are). The second one seems more promising to me, so I'll discuss this one.
For example: Are you using a motor? or. linear actuator? or Are you just specifying a force or torque or a motion profile on the input? What is the geometry? etc.
What I want to model is a spring attached to a scissor mechanism. Say the scissor mechanism has three attachement points A, B and C, with the distances AC and AB free to change arbitrarily but always staying proportional to one another. I want to attach a mass to point B and place a spring between either A and B or A and C (I don't think where I attach the spring makes much difference). All masses except the one attached to B should be neglected.
Then I want the whole thing to be a component with A and C serving as connections to other components. And then I want to attach many instances of this component in a linear sequence : C1 connected to A2, C2 to A3 and so on.
For what it's worth, my goal is to design some kind of linear catapult. I'm fairly sure I could do it without SystemModeler, by writing a Lagrangian for the whole thing, and then deriving the Euler-Lagrange equations with Mathematica, but the Modelica approach is conceptually more appealing and might be easier if I want to make the model more realistic later.
PS.
For example, look at idealGear and idealGearR2T. You can model the scissors mechanism as a transmission.
Yeah you're right that seems like the exact analog to what I'm trying to do. I guess I'll have to study how this works.