The 5475HB is a Karbonite servo, no metal gear to the best of my knowledge.

Please explain in detail how you’re setting up the servo and TX.

The problem off center or deviations from center is atypical and varies with every TX.

I'll try and explain the conundrum:

Ideally your TX will transmit a pulse width of 900-1500-2100 micros seconds at end-point/neutral/end-point. But it’s not an ideal world and manufacturing and component tolerances skew the results.

Since your servo may not center exactly at 1500us and the TX outputs a different number too, things are amiss. This is where Hitec programmable digitals leave all others behind. With a Hitec programmer you can reset the servo to recognize the TX’s output as center and maintain linear travel either side of center.

As an example lets say your TX is outputting 1530us at neutral and maybe the servo is slightly of too, which is very common. Once you utilize sub-trim or trim to center the servo you’ve introduced a deviation and or offset in to the travel arc that now follows at end-points.

As you note the travel percentages are unequal, which is a result of using trim to center the servo. The neutral offset has to be absorbed; if the offset is 30 degrees in one direction the end-points should reflect same.

Another variable maybe your mechanical linkage set-up, if something is unequal here you’ll note this problem too.