check
check
check
check
check
check
A Hall current sensor with noisy or offset output is usually responding to real inputs it was not meant to measure: the primary conductor sitting off-center or at an angle in the aperture, magnetic fields from adjacent high-current cables or busbars coupling into the sensing element, a shielding or grounding arrangement that lets common-mode noise in, bandwidth/filter settings mismatched to the application, or an un-zeroed output where the model provides zero adjustment. The first safe check is geometric and documentary — confirm the conductor's position in the aperture and the presence of nearby current paths against the installation drawing — before any electrical measurement. All measurements on live current-measurement circuits are qualified-personnel work; do not probe energized primary conductors.
The sensor measures the field, not the wire. Any current — the conductor you want, the cable next door, the busbar behind the panel — contributes magnetic field at the sensing element. Noise and offset often mean the sensor is faithfully reporting your neighbors.
Conductor position changes the reading. Off-center or angled conductors alter the field distribution inside the aperture; most of the offset/noose complaints start with "we re-landed the cable".
Offset at zero current is a condition, not always a defect. Residual magnetism, nearby DC paths and uncompensated offset add up; where the model provides zero adjustment, it is applied to a documented zero state — never to mask a real nearby-current offset.
Bandwidth cuts both ways. Measuring fast transients needs bandwidth; a noisy reading on a slow control signal needs filtering or a narrower-band model — the selection-side logic is in the Hall selection guide ({{URL_C04}}).
Collect the waveform before adjusting. A scope capture at the sensor output (by qualified personnel) separates pickup (line-frequency ripple, switching spikes), offset (constant shift) and oscillation (stability problems) — each has a different fix.
