2) Cumulative error: 1k laps, first 15km very well in place, but after that the cumulative error appeared and the result at 42km already 130m cumulative error.
look at your single recorded lap-distances. They are sometimes more and sometimes less than 1000m.
If the algorythm would work fine, summation of the single distances to 1000 m would be 0 m. But in reality the 1000 + values are bigger than the 1000 - values. The result is a positive abberation from x.000m
I have done two runs with the firmware x 17,7 km, the "error" was + 40m (pace of 5:30min/km), 26 km, the "error" was +130 m (pace of 6:15min/km). On the first look the "error" seems to be always positive but to be sure much more recordings are needed.
Maybe this is a way to solve the problem:
distance over all = d
lap-distance = ld
number of laps = n
A full lap should be triggered always when d > ld * (n+1)
With every full lap n is getting one more.
The disadvantage of this solution is, if you take a manual lap the new lap should count from this point the next lapdistance but if a lap always is triggered from the beginning of the recording, the next lap would come to early.
So it would be better to optimize the new algorithm. I don´t know how the auto-lap is triggered at the moment. In the old firmware the lap was always full, when the lap distance was reached. The reason for the positive error in the past firmware-versions was, the distance can´t be measured meter by meter. If after 998 m came 1006 m for example, the full lap was triggered at 1006 m. As result the more than 1000 meters were accumulated to more and more absolute "errors".
With the new firmware x lap-distances with less than 1000 m are possible. It looks like the new algorythm regards the "step-size" betwen two values. Maybe the step-size depends on the speed. This means, the (positive) cummulative error would be bigger the faster you run (more distance in the same time), if the step-size is a fix value. To trigger a full lap means, to trigger at a distance calculated by lapdistance - "step-size". In this case the positive and negative errors will accululate to 0. But you (globalsat) have to determine the right step-size which depends on the speed. At the moment it seems the step-size is to small. Please check your algorythm of calculation of "step-size", if this description is the way in the new firmware.
You have to