The spikes in phase sigma values on 2 out of 3 receivers looks to be clock related given the spikes occur on all signal measurements simultaneously. One possible source for this is the clock steering of the receiver which is enabled by default. The steering algorithm may be making larger adjustments on some receivers than others.
I suggest the customer disable clock steering by issuing the command "CLOCKADJUST DISABLE" to see if that improves the consistency between receivers. Given the GPStation-6 has a built in OCXO, it is not really necessary to have the clock steering enabled.