When Don Bently worked on making solid-state versions of the eddy-current measurement system (it was actually originally designed in the 1930s by GE engineers using vacuum tubes), he had a choice
between using N-P-N transistors or P-N-P transistors. At the time, transistors were quite expensive, so he chose the least expensive of the two: P-N-P (apparently, PNP transistors they were less expensive to manufacture 50 years ago than their NPN counterparts).
Because the circuits used PNP transistors, a negative bias voltage was required rather than a positive bias voltage. Don chose -18V. This was later changed to -24V to allow more linear range from the transducer.
At that time, the industrial instrumentation community had not yet standardized on +24 vdc, and by the time they did, there were so many Bently Nevada eddy current vibration sensors installed that changing to +24V rather than -24V was not greeting with enthusiasm by users. Hence, it has remained -24V to this day.
