TRG vs GPS Statistics of Mar 1999 run

Relative time between spills (via TRG 48-bit clock) (*0)

same as above, but zoomed in [2.17 ... 2.21] and with histogram

TRG oscillator drift from 50 MHz (*1)

Absolute spill time difference between GPS1 and GPS2 (*3)

Relative spill time difference between GPS1 and TRG (*2)

Relative spill time difference between GPS2 and TRG (*2)

Note: y axis is log scale

(0)*	About the data gap between March 6 - 11, 1999: The data logging option for the GPS control software (gpstrg) was originally only designed for debugging during the initial K2K beam days. It was turned off on March 6, 1999, when I was satisfied with the operation stability of the GPS system. But then the option was decided to re-enable on at March 11, 1999, to maintain full data backup since the Online DAQ system appearantly lost 5%-10% of the GPS events due to network timing problems - which were fixed a few weeks later.
(1)*	Both, the TRG module and LTC module each use an internal, free-running 50 MHz crystal oscillator (independent) for their clock counters. The two oscillators are not temperature compensated! Therefore, any change of the room temperature in the North Hall control room causes them to drift. Cooler temperatures usually result in higher frquencies than at warmer temperatures.
(2)*	Absolute TRG clock vs. GPS time measurements cannot be made due to the mentioned temperature-dependent TRG oscillator drifts, as shown in the wave forms of the plots above. The waves are inverse proportional to temperature changes in the North Hall.
(3)*	During this run period, GPS2 had one 12.5-minute satellite synchronization problem, caused by a firmware bug on the Motorola UT-plus Oncore module. During this 12.5-minute section GPS2 data is totally wrong and should not be used (GPS1 is ok, though): 03/25/1999 from 01:38:51 until 01:51:29.