Inverted raw times before using cuts
Calibration data appears as a second peak around t1-t2=400 channels in the time above threshold spectrum.

* t1 -t2 in channels:

After using the calculated slopes one sees that this second peak is exactly at the 200 ns corresponding to the internal calibration.

• t1 - t2 in ns::
  

As an example I chose the wires 3, 58 of the first MB to study the difference between them.
As it appears below the difference between the two wires is about 74 channels.

• t1 in channels:
  
  But amazingly 76 ns

• t1 in ns:
  

As a result of the huge slopes variation.

• slopes in ns/ch:
  
  Then I compared another file where the global offset is different by 200 ns.
  And found the following t1 spectrum.

• t1 in channels:
  
  But after calibration:

• t1 in ns:
  
  We see almost the same offset difference of 79 ns between the two wires.
  as the slopes remained in a range of 2% unchanged.

• slope difference ns/ch: