I've compared the pedestals extracted form nomSciOps runs on July 7 and july 13 with the actual baseline pedestals from June 28. The histograms of pedestal differences in ADC units are show on the following plots:
Top plots show the pedestal drift by july 7 for x8 ranges (left) and x1 range(right).
Bottom plots show the pedestal drift by July 13 for x8 ranges (left) and x1 range(right).
According to these plots there is a slow pedestal drift of upto 6 adc units in 2 weeks in the most sensitive channels.
At the same time the CAL temperature during this period in survey mode was stable within 0.2 degrees - see the points of June 28, July 7 and July 13 on the next temperature plots: 
temperature of the central tower AFEE (tower=5, afee=0) vs time
temperature of corner tower AFEE board (tower=0, afee=0) vs time
The pedestal difference between July 13 and June 28 is factor of 2 bigger than between July 9 and July 7, where the real AFEE temperature difference of ~1.5 degree was measured:
LEX8 pedestals difference between July 9, 11:00UTC(pointing mode - maximum temperature deviation) and July 7, 17:00 UTC (end of long period of survey mode)
HEX8 pedestals difference between July 9, 11:00UTC(pointing mode - maximum temperature deviation) and July 7, 17:00 UTC (end of long period of survey mode)
It is not clear what could be the reason for the pedestal drift while measured AFEE temperature is stable - may be the temperature of GCFE chips is different from AFEE board temperature and is changing slowly ? - A hypothesis to verify.