Some trouble with trajectory

I use pronterface, Sfact and teacup, and i have some differences in trajectory between the layer 0, the others and the last. I join a picture, that's more effective than a long explanation. It's a thin wall cube 20 x 20 mm. I hope someone could help me to understand and solve that problem.

Possibly false triggering of the endstops causing it to stop early.

---

[www.hydraraptor.blogspot.com]

Well printing is in the middle of the bed (?), so how can verify, or reset trigger of the endstops ? I don't understand why the first layer is good and the second one starts 2 mm earlier. And why the last was done at 25%?

Teacup doesn't monitor the endstops except when homing. It may be lost steps on Y or X axis, depending on how the part was oriented on the printer. The first layer is often printed more slowly than subsequent layers, which would explain why the base turned out OK but the second layer failed. Turn up the power on that motor slightly and see if it improves. Or it could be too much power causing a stepper driver to overheat and cut out for a second, in which case turn the power down or improve the cooling on it.

I'm going to experiment with a rectangle instead of a square to determine which axis is wrong and how the piece is oriented.
Currently what is the most successful firmware?

I printed a rectangle, so the y-axis is wrong. I tried to increase or decrease engine power, but nothing has changed.
I attach a part of Gcode, and I don't know why :
From here
G1 X107.2 Y87.8 Z3.2
G1 E3.1
G1 X92.8 Y87.8 Z3.2 F1200.0 E0.32
G1 X92.8 Y112.2 E0.5422
G1 X107.2 Y112.2 E0.32
G1 X107.2 Y87.8 E0.5422
G1 F3000.0 E-3.0
; S1.0
; S0.8
G1 X107.2 Y87.8 Z3.6
G1 E3.1
G1 X92.8 Y87.8 Z3.6 F1200.0 E0.32
G1 X92.8 Y112.2 E0.5422
G1 X107.2 Y112.2 E0.32
G1 X107.2 Y87.8 E0.5422
G1 F3000.0 E-3.0
; S1.0
; S0.8
G1 X107.2 Y87.8 Z4.0
G1 E3.1
G1 X92.8 Y87.8 Z4.0 F1200.0 E0.32

To here, it's OK! And the print stops
If I order a move manually, the print restart to finish Gcode but with an offset

G1 X92.8 Y112.2 E0.5422
G1 X107.2 Y112.2 E0.32
G1 X107.2 Y87.8 E0.5422
G1 F3000.0 E-3.0

May be it's due to firmware? an idea?