XY Zero not aligned

Hi all.
After some testing, i got my "XYZ Zero tool" working. But I've some problem with XY zero (Z0 is correct)...

Summarizing: in my Axis I created 3 spinboxes linked to 3 variables: Material thickness (#1000), Tool diameter (#1001), Tool lenght (#1002).
In my Zero Gcode, I read the values in the spinboxes, then start with the zero procedure.
Pinpoint: I don't really use the tool lenght (meaned without the shaft lenght), I just created it maybe for some use (if need it)

The procedure works: I use an aluminium square as touch plate (15mm tick) Z, Y and X (in this order) are zeroed, but if I then go to X0 Y0, the tool is not where I expect to be....
In the attached image, I expect to see the tool in the green area (where the dotted lines cross), but in reality the tool is in the red cross.... the distance is really low (about 0.7 mm center-to-center - sorry for the inaccuracy)

For the test, I'm using an end mill 0.8 mm bit.

the procedure should work this way: after input material tickness (5 mm) and tool diameter (0.8 mm), Z is zeroed, then rised to Z=20mm-material tickness (this way, whatever is the material thikness, Z rise always at the same height)
Then Z moves along Y 8 mm and goes down to a fixed position. I read the current Z height: if the difference [value-5.5 ] is greater than 1.5, the Z goes down to a fixed position (1 mm below the square border), otherwise the procedure stops and I get a warning (it happens if the material is more thick than 13.99 mm and the bit could hit the material surface and damage itself).
If no warning happens, the Y and X procedure goes fine.
At this point, XYZ are zeroed. But the X0 Y0 doesn't match the position I think (the green circle).

When setting the touch offset, I use the (square - material tickness) for Z, and the (tool diameter / 2) for X and Y. Is that theoretically correct?
If so, why the X0 Y0 is in the wrong place?

Attached also may zero Gcode (a little bit messy, sorry for that, is still the test version, not yet cleared)

Thanks all.

Andrea

After some test and better measurements, I realized it was such a parallax error. The bit is in the right zero.
Think I need to fix a little bit the square so to get more room for the bit in "parking position" (zero).