Hi Carl,
I wouldn't take that value of 0.012% settling time to be the "minimum", but just what has been tested for the device. At 5.5 microseconds per sample, the settling time percent would probably look more like 0.015% or 0.020%. This is a limitation of data acquisition devices where higher sampling rates are going to lead to less accurate measurements. Also, that's not to say that the values are always going to be off by that percentage. I would interpret it as the maximum error that could happen at that measurement time window.
Mark M.
Applications Engineering
National Instruments