New servo code

A fruitful discussion on the ppsi mailing list culminated with this message. I upgraded my tree to the new code and it works (for me anyway) much better. This is the chart which shows it settling down:


This shows none of the oscillations that used to appear. But maybe they are hiding in the flat part?



No -- this looks excellent.

Looking at the offset between the time and a reference clock, I see:


This chart starts ten minutes later than the other chart as it was still trying to get the offset down to a manageable level. I'm also not going to blame the servo code for the wander -- these are two different GPS receivers -- only one of which is a timing receiver. I suspect that a wander of up to 1µs is to be expected.

All in all, this is a major impovement!