Q. is it in your perl code, or firmware code ?

A. In the Perl code

Almost right. Transit time is found by looking for instances of zero injector pulsewidth. It seems to be mostly proportional to RPM to the point of being able to neglect anything else with relative safety. It's only used to produce a time-corrected Lambda reading, which is then used - along with the other log data - for working out statistics for the RPM and tpsDOT acceleration bins. The added amount and times can be derived from this, but it's not a straight forward algorithm because there are more unknowns than equations...