Talking to NI support, they have given me an idea to try.
They have said to start measure in DC coupling, and check the voltage read. It should be (Current excitation value times accelerometer impedance)
If no accelerometer is connected, this measure should be zero or little offset, or something very different.
I can't try it until next monday, as the system is installed on field, and I'll be out of office.
Try it, or wait my reply next week
Thanks to all,
Aitortxo.