A local magnetic measurement near a suspected lamination short contains the ordinary core field as well as any response from the abnormal current path. The ordinary field can change sharply near joints and edges. Moving a probe slightly can produce a difference larger than the effect being sought.
Plan the probe locations around the physical hypothesis. Identify the candidate current path, predicted field component and a reference region whose normal geometry is sufficiently comparable. An arbitrary nearby quiet point may be a poor background reference.
Draw the placement map
Mark coordinates, orientation, standoff and the probe’s sensing volume. Include the joint surfaces or cooling passages that influence the main-field gradient. Retain a method for returning to the same positions after a repeat trial or controlled state change.
Choose points that discriminate the proposed field pattern. A collection of measurements all on one side of the path can miss a directional change that would strengthen or challenge the hypothesis.
Reference the excitation
Record actual excitation and its phase where synchronized components are used. A different waveform or amplitude can change background field. Normalize only through a relationship demonstrated to be appropriate over the selected range.
Reference the geometry
Compare the suspect configuration with an intact or otherwise justified reference under the same probe geometry. Evaluate position uncertainty using a small controlled displacement on the reference. This shows how much difference can arise from registration error alone.
Reference the probe’s disturbance
Check whether the probe, holder or lead changes joint clearance, local contact or the field itself. A sensor inserted into a narrow region can alter the condition it is intended to observe. Where this effect is significant, use another arrangement or include the disturbance in the model and limitation.
Retain vector and time information
Several components or phase-resolved traces can help distinguish a local pattern from a main-field amplitude change. Their interpretation still depends on calibrated sensor response and known orientation. Do not assign fault current from one component with an unsupported conversion factor.
Compare the measured difference pattern with a justified forward model or representative known specimen. Include the probe’s spatial averaging if it matters near the predicted gradient. A point-model maximum cannot be compared directly with a finite sensing volume.
The output should be a placement drawing and a registered difference map with its position sensitivity. It can support or challenge the proposed conductive path. If the result is unresolved because ordinary field gradients dominate, the map identifies which probe position or complementary observation is needed next, rather than turning a background peak into a fault conclusion.
Compare positional pickup with the candidate field
Catalogue nominal sheet thickness is 0.23 mm [1]. Assume placement shifts by that distance, local background gradient is 4.0 mT/mm and the candidate fault-related difference is 0.50 mT. Placement alone creates 0.92 mT, exceeding the candidate signal. To keep this contribution below an assumed 0.10 mT budget requires displacement below 0.10/4.0 = 0.025 mm. This is a fixture-design target for the example, not a probe accuracy claim. Demonstrate position recovery or select a lower-gradient location, then include sensor, orientation and excitation uncertainties [2] before converting a field difference into current.
Manufacturing controls. Joint overlap, layer displacement, duct opening and clamp geometry shape the background gradient. The probe holder must maintain standoff without changing the same gaps or contacts whose field it is intended to observe.
References
- Nippon Steel Corporation, Grain-Oriented Electrical Steel Sheets, catalogue D004je_03_202004, 2020, Specifications; Size and Form Tolerances.
- JCGM, JCGM 100:2008, Evaluation of measurement data — Guide to the expression of uncertainty in measurement, 2008.