Has anyone tried sending their failed logs to R-Track? I would think
they might be able to pinpoint the glitch in the checksum calculation.
There's probably an errant byte somewhere, maybe due to the memory
wrap-around, maybe not.