
By Daniel Foster – Condition Monitoring Services Consultant
In condition monitoring projects, measurement accuracy often depends less on the sensor itself and more on how it is installed. A recent customer case involving the Bently Nevada 163947-00-05-02-84-90 non-contact current sensor illustrates this point clearly.
The sensor was fully functional. The data, however, was not.
Background
-
Site type: combined-cycle power plant
-
Objective: trending motor current for early fault detection
-
Monitoring scope: multiple identical motors with comparable loads
-
Issue reported: inconsistent current values between units
From a monitoring perspective, the variation exceeded acceptable tolerance.
Data Anomalies Observed
-
One motor consistently showed lower RMS current
-
Load and process parameters were confirmed identical
-
No alarms, no dropouts, no saturation
-
Trends were stable—but wrong
This is often more dangerous than obvious failure.
On-Site Verification
Physical inspection revealed subtle but critical differences:
-
Sensor-to-conductor distance varied by several millimeters
-
One sensor was slightly angled relative to the current path
-
Cable bundles caused uneven magnetic field distribution
These deviations were within what many installers would consider “acceptable.”
They were not.
Root Cause Analysis
The 163947-00-05-02-84-90 measures current indirectly via magnetic field intensity. Small geometric changes result in:
-
Reduced magnetic coupling
-
Systematic under-reading of actual current
-
False conclusions when comparing machines
The sensor output remained stable, which masked the underlying measurement error.
Corrective Measures
-
Reinstalled sensors using a fixed alignment template
-
Standardized distance from conductor across all units
-
Separated power cables to reduce field distortion
-
Re-baselined current trends after mechanical correction
Once geometry was corrected, all motors showed consistent current behavior.
Why This Matters
-
Condition monitoring relies on relative accuracy
-
Stable but inaccurate data leads to incorrect maintenance decisions
-
Installation tolerances must be treated as measurement parameters
-
Non-contact does not mean non-sensitive
Professional Assessment
The Bently Nevada 163947-00-05-02-84-90 performed exactly within specification. The fault lay entirely in installation practice.
In monitoring systems, precision begins with mechanics, not software.
— Daniel
Excellent PLC
