Looked at whether the frequency of missing F1 reference times depends on event type. It doesn't. F1 reference times are missed ~0.35% for event types 1-8.
Larry and I looked to see if the 1190 reference time is missing for events that are missing the F1 reference time. A scientific sample of two events (that miss both F1 reference times), show that missing F1 reference times are not correlated with missing 1190 reference times. (The 1190 may never miss it's reference time.)
So either the F1 is simply 0.35% inefficient, the electronics that delivers the reference time is 0.35% inefficient (but only for the F1), or the F1 reference times are at the edge of the F1 TDC window. I thought I checked this earlier, but I will check again.