I had a weird issue today and MCE couldn't really give me any direction. I'm wondering if anyone else has seen this or knows why it was happening.

We had to replace the double Elsco guide rollers with singles to lower the landing system because it would contact the top of the hoistway on a counterweight buffer test. Apparently it was never tested after the mod. 8 cars and all but 1 have the old style i-Land with the big encoder wheel (don't know why this 1 had the compact I-Land).
The first we did was one of these, moved the magnets, did a scan, fine tuned the floor offsets and good to go. The second car was the one with the compact i-Land with the small encoder roller but I don't know if that has anything to do with it. With the magnets moved dead center at floor heights we do a scan several times, on manual mode and auto but every time the floors are way out. Top floor was really close but the bottom was an inch low and all intermediate floors were 3/8"-5/8" high. When I would do the floor calibration and enter the measured values, now the top is good, the bottom is now an inch high and the intermediate floors are 3/8"-5/8" low. So I tried entering half the measures offsets, now the bottom is 1/2" high and the intermediates are around 1/4" low. So I do the floor calibration again, still 1" low at the bottom, intermediates around 1/2" high and the estimated offsets are less than 10 thousandths of an inch. This time I just entered 0 for all the measured offsets, send it and now it nails all the floor heights. WTF?

Does anyone have any idea what's going on?