Your problem is most likely nothing to do with your code or wiring. R/C servos are designed to work with internal feedback from a built-in pot that tracks with the gear train over the approx 180% travel and are very linear when operated as designed like that, that is a 1 to 2 millisec signal pulse tracks well from 0 to 180 degrees of mechanical travel.
When one modifies a R/C servo to become a continuous rotation 'servo' the pot is disconnected either mechanically or electrically. The result is that the center 50% (1.5 millisec) signal is the approx stop position and any deviation plus or minus from 1.5 millsec pulse, causes a variable speed in one direction or the other. However the deviation from stop is in no way linear and to expect two modified servos to track identically when wired to the same control signal is probably an assumption not possible. The internal servo 'tuning' were never designed to work in a continuous rotation mode as the whole take a standard R/C servo and convert it to bi-directional variable speed geared motor drive is just a popular hack and results in it not being a precision device anymore.
Lefty