Those who like tinkering with their rubidium standards may find the following paper of interest http://tf.nist.gov/timefreq/general/pdf/1013.pdf You can potentially drop the Allan deviation of a rubidium package by a factor of ten using the modification detailed in the paper. Bruce