Over the past 2 hours, starting 20:45, w.3m.near has gone down and up. Other w's don't show this behavior. ldiag is okay, spd and tc seem reasonable to me at first glance. The w time series looks okay, just with shifts. We'll need to see if this behavior persists.