Skip to content

KCSC Training

Chapter 1 › 1.8 › Slides

Slides · 1.8

When a Sensor Lies

Read · 12 slides

1 / 12

When a Sensor Lies

Recognize bias faults, saturation, zero shift, hum, cable noise, sensor resonance, temperature effects and a loosening mount
In the time history and in the spectrum

Bias Out of Its Window

Near the supply: an open circuit; near zero: a short
Inside the supply but outside the window: a damaged sensor or the wrong supply
The check: the reading, the sensor on a known-good cable, a known-good sensor on the same cable

Saturation and the Baseline That Decays

Clipped peaks, then a baseline displaced and decaying back
The ASD of the same record can look nearly normal
The check is the time history at full bandwidth, not the spectrum

Four seconds of a random response with three overloads clipped at plus and minus 50 g, each followed by a baseline displaced downward and decaying back — Figure 1.9. An IEPE channel driven into saturation: the time history at full bandwidth, with clipped peaks and the baseline stepping away and decaying back exponentially, beside the ASD of the same record, which sits close to the ASD of a clean record of the same motion drawn under it.

Twelve milliseconds around one overload: a ring clipped flat at both rails, then the baseline stepped below zero — Figure 1.9. An IEPE channel driven into saturation: the time history at full bandwidth, with clipped peaks and the baseline stepping away and decaying back exponentially, beside the ASD of the same record, which sits close to the ASD of a clean record of the same motion drawn under it.

Zero Shift on a Shock

An offset after the pulse that is not motion, bleeding back at the chain's time constant
Integrate to velocity: a table starts and ends at rest, so the velocity comes back to about zero
A velocity that ramps away is the sensor, not the pulse

A 20 g half-sine shock with compensating pulses, recorded twice, one returning to zero and one whose baseline sits 1 g off zero after the pulse — Figure 1.10. Two records of the same shaker shock and their integrals to velocity: one physical, whose velocity rises through the pulse and returns to zero, and one with zero shift, whose baseline sits off zero after the pulse and whose velocity ramps away without end.

The two velocity integrals: one rising through the pulse and returning to zero, one ramping away without end — Figure 1.10. Two records of the same shaker shock and their integrals to velocity: one physical, whose velocity rises through the pulse and returns to zero, and one with zero shift, whose baseline sits off zero after the pulse and whose velocity ramps away without end.

Hum

A comb of narrow, steady peaks at the line frequency and its harmonics, there with the drive at zero
The check: amplifier on, drive at zero; then break the loop and repeat
If the comb goes, the article was never involved

The ASD of a response channel with the drive at zero: before, narrow peaks at 60 Hz and its harmonics standing far above a flat floor; after an isolated stud, the floor alone — Figure 1.11. The ASD of a response channel with the drive at zero, before and after an isolated stud: before, a comb of narrow peaks at the line frequency and its harmonics standing far above the floor; after, the floor alone.

Spikes and Dropouts That Follow the Cable

Spikes that line up with the cable moving; dropouts to the baseline and back
Both grow with level, because the cable moves more
The wiggle test: drive at zero, the cable flexed along its run, each connector pressed and turned

Two channels on the same fixture over three seconds: the clean one unchanged at four marks where the cable moved, the faulty one with spikes and short dropouts to zero at each mark — Figure 1.12. A response channel's time history with a cable fault: spikes and short dropouts to the baseline that line up with marks showing when the cable moved, against a clean channel beside it on the same fixture.

A Resonance That Belongs to the Sensor

The same peak on every test, whatever is on the fixture: the mounted resonance
On one channel and not its neighbors, at a frequency the survey never showed: the mount
The check: the same mode on another channel; one channel alone is the sensor until cleared

A Mount Coming Loose

The high-frequency floor rises; spikes at the peaks of the motion
The mount's resonance walks down from run to run, and the effect grows with level
The check: a wrench on the stud, a fingernail at the glue line, the rigid-body check

Drift With Temperature

A slow drift while the chamber temperature changes, settling when it stops
The check: a record with the drive at zero across a temperature change
The cure: a shear design next time, a settled temperature before every run now

Right Shape, Wrong Size

A wrong sensitivity, a gain twice, the wrong units or detector, the wrong patch, axis or way up
Each gives a clean signal of the right form at the wrong level or sign
Found only by comparison: the calibrator, polarity, the second chain, the rigid-body check

The Order of Suspicion

The chain is checked before the article is blamed
The chain's checks take minutes; the article's take days
A finding that survives the chain's checks is a finding the analyst can defend

Decide

Every signature has a five-minute check, run before the article is blamed
The bias, the time history at full bandwidth, the velocity integral of every shock, the drive-at-zero record, the wiggle test
The neighboring channel on the same mode, the wrench on the mount, the rigid-body check
What the check found, and what was done about it, in the log beside the run

This page is part of the KCSC Training course and is licensed to one named reader for viewing. It does not print. Lab kit pages print from the Lab Kit menu.