Checking the PySb code for possible causes of those crashes I spotted the fact that on the SET_OUTPUTS command implementation a burst of 24  single modbus coil setting commands was generated. I have inserted a small delay between the commands (0.2 sec) and I was not able to replicate the crash while commanding the relays from the VPM interface or from ITF_LOCK. The use of the relays in the DOWN and LOW_NOISE_1 states of ITF_LOCK has been provisionally restored.