At Jun 27 20:35 UTC (15:35 CDT) the 10 meter TRH data stopped. It resumed on its own about 1.5 hours later. From the archive file:

data_dump -i 4,60 -A ncar4_20110627_120000.dat.bz2  
...
2011 06 27 20:35:48.5645  0.9701      30 TRH15 24.64 55.05 6442 1670\r\n
2011 06 27 20:35:49.5345  0.9701      30 TRH15 24.64 54.87 6442 1665\r\n
2011 06 27 22:07:08.7843    5479      13 \x006412 1642\r\n
2011 06 27 22:07:09.7387  0.9544      30 TRH15 24.35 54.09 6413 1645\r\n
2011 06 27 22:07:10.7089  0.9702      30 TRH15 24.34 53.84 6412 1638\r\n

I noticed in the daily plots that it stopped again on Jun 28 about 13:00 UTC. I logged in, ran rserial (rs 7), and it resumed after I hit a few carriage returns.

Since a carriage return seemed to help, I changed the xml on that system for the TRH sensor, adding a timeout of 10 seconds and an init_string of \r:

<serialSensor ID="NCAR_SHT" class="DSMSerialSensor"         
            baud="9600" parity="none" databits="8" stopbits="1" timeout="10"
                init_string="\r">