The PSD3_time_sync_data.txt file contains the information needed to correct the DAS data files.

The GPS became disconnected from the DAS and the PC clock started to drift. The pdf file shows a plot
and explains when the drift started and ended. It appears the drift is linear. 
The data files that need correcting are located ftp://ftp1.esrl.noaa.gov/psd3/bao/CU/Emily/CarriageData/

Please apply a linear correction to the DAS time stamp (Col 1) in the raw files. If my calculation is correct it should be 
about a 20sec correction over ~ 15 days.
met01005315_raw.txt through Feb 22 1500 UTC through Mar 8 2134 UTC
I have the raw files backed-up.

Details on the PSD3 file from Nick Wagner. It explains why some of the GPS server data are missing.
The first column is the local computer time and the second column is the GPS time measured by the tower server computer. 

1) On March 10th, I changed the network configuration in an attempt to use remote desktop from Boulder. When I switched the
configuration the time sync stopped working.

2) The time sync program times out after 100 ms. So, if the time server didn't respond fast enough or network slowed, the time
sync software will not report a GPS time. The two larger gaps are times when the time sync was not running. I noticed the time
sync software was not running and restarted it. I think it was not running because PSD3 was restated.

3) I don't know the reason for the spikes but I'm going to ignore them.