Is the collocation a problem? Currently, we are using utf8_general_ci instead of utf8_bin.
This is the problem. I have to switch the collation of the token column to utf8_bin or utf8mb4_bin (as suggested by the B8 developer). Seems to be a local problem in my installation because it is correct at github.
applied-geodesy.org - OpenSource Least-Squares Adjustment Software for Geodetic Sciences