The easiest trial run would be at the state level. Making each states Senators part of the same class instead of being from two of the three separate classes would require a Constitutional Amendment, since it would change the terms of some sitting Senators to something longer or shorter than the Constitutional 6-year terms.
It also (presuming Senate terms aren't changed besides what is necessary for transition) would mean each state would have six years between elections for the Senate rather than alternating 2- and 4-year intervals.
> Unfortunately this is subject to game-theoretic problems because it tends to make the Senatorial output of a state less polarized (California might have one Republican and one Democrat instead of two Democrats), which could result in a political swing towards those states without proportional Senators (Wyoming might still elect two Republicans with FPTP).
This can be mitigated by increasing the number of Senators per state; even 3/state would probably make it a non-issue, it would make 2/1 splits common initially, so you wouldn't have most states even with the others 2/0.
I think the easiest "trial run" for proportional voting in the US is using two-winner elections (preferably Schulze but maybe traditional STV) to elect Senators, because this doesn't require any changes in representation. Unfortunately this is subject to game-theoretic problems because it tends to make the Senatorial output of a state less polarized (California might have one Republican and one Democrat instead of two Democrats), which could result in a political swing towards those states without proportional Senators (Wyoming might still elect two Republicans with FPTP).