Originally Posted by Andy_K

Here's a slightly more visual explanation (motivated by my own efforts to convince myself that what I said in my previous post was correct).

The model I was using to calculate cable pull based on the pivot radius is assuming that the amount of cable pulled is approximately the same as the distance traveled by the anchor point. I mentioned in post 18 where I introduced this method that the cable pulled was actually the chord cut across the virtual circle traveled by the pivot point but for most levers that's a pretty good approximation. This assumes that the pivot point is near the top of the virtual circle. (See Figure 1 on the attached image.)

If the cable is anchored significantly left of the top (relative to the cable exit) of the virtual circle, the approximation would break down, as seen in Figure 2. Here the amount of cable pulled is actually the difference in the distance from the original anchor position to the exit point and the distance from the final anchor position to the exit point, assuming nothing restricts the cable's freedom of movement within the virtual circle.

[...]

In any event, this is a fine tuning effect and the simple approximation based on the inner circle tells you enough to understand approximately how the brake will respond.