Below are two screenshots of the overlay of laps from two different sessions at Pitt Race using RaceChrono. The first is with the internal GPS on my S8, the second with an external Garmin bluetooth GPS receiver. I know which I'd trust the results of more.....and no there were no off track excursions with the first pic



S8 internal GPS


S8 w/Garmin External

 

Assuming your Garmin receiver updates at around 10 Hz like their VIRB camera, I'm not surprised to see how much smoother the trace is vs. a 1 Hz phone. Thanks for posting that.

On the other hand, most start/finish lines are on a straight and usually at some distance after the last corner. And certainly in my case with a NA S2000 at altitude, the rate of acceleration is not great and reasonably constant over 1 second. So even with only 1 Hz sampling, it seems that Harry's could interpolate the two waypoints before and after the finish and be reasonably accurate about the timing crossing the line. All that said, I think RobRob is right that any variation in exiting the last corner (at least on this track where there are three corners linked together) coupled with the non repeatability in setting the start/finish with the Garmin software can result in a variation of at least few tenths. In fairness to Harry's, as I said above, I think its approach (assuming it does what I described) should give the more repeatable and comparable time vs. what others measure.

@cracknut, I agree that the corner speeds reported by Harry's are useless as the graphs all have sharp transitions and I know my mid corners speeds don't change that fast.