After two years, three sensors failed verification. Destructive examination of two failed units found no visible moisture, but their resistance was approximately 20 ohms below the expected value.
Observed Failure Evidence
| Item | Evidence | Engineering significance |
|---|---|---|
| Stated range | -196 to +250 degrees C | The -90 degrees C process temperature falls within the stated range, but range alone does not establish resistance to repeated thermal cycling or handling. |
| Population | Approximately 20 installed; three failed verification after two years | The observed failure fraction is approximately 15%, subject to the approximate population count. |
| Electrical symptom | Resistance approximately 20 ohms low | A low reading is consistent with an unintended lower-resistance path, but does not identify its physical cause. |
| Internal inspection | Four channels containing platinum-wire turns; white powder escaped from two channels | The material was not identified, and the original internal condition may have been disturbed during disassembly. |
| Moisture inspection | No visible moisture found | This rules out only visible moisture in the two opened samples, not all insulation or sealing faults. |
Root-Cause Assessment
Turn-to-turn contact is a plausible explanation for resistance approximately 20 ohms below expectation. However, the examination does not prove whether the winding lacked adequate fixation during manufacture, whether fixation degraded during repeated transitions between testing down to -90 degrees C and an ambient environment of 25 plus or minus 10 degrees C, or whether disassembly released material and windings that had previously remained fixed.
The evidence also does not identify the white powder or establish that it should have been fully sintered. Treat the powder identity, the required construction method, and the dominant failure mechanism as unknown until the manufacturer supplies construction criteria or a controlled failure analysis confirms them.
Diagnosis and Replacement Selection
- Retain the third failed sensor intact. Record its resistance at a controlled temperature and compare it with the applicable sensor characteristic and verification record; do not open it before nondestructive testing is complete.
- Compare all approximately 20 serial-number records, verification results, exposure histories, handling practices, and cable condition. Determine whether failure correlates with a production group, number of cold cycles, minimum exposure temperature, or mechanical handling.
- Request a manufacturer failure analysis for an intact failed unit. Ask specifically whether a resistance shift of approximately -20 ohms is consistent with winding contact and whether repeated cycling between -90 degrees C service and 25 plus or minus 10 degrees C ambient is covered by the sensor qualification.
- Define replacement requirements from the duty cycle, not measurement range alone: operation at -90 degrees C, repeated thermal transitions, several-hour tests, manual installation and removal, required accuracy, mechanical construction, and verification interval.
- Before fleet replacement, qualify candidate sensors through repeated representative cold-to-ambient cycles. Measure resistance at a controlled reference condition before cycling, at planned checkpoints, and after cycling; accept a candidate only if it continues to meet the required verification tolerance.
No evidence supports naming a longer-life substitute. Select a replacement only after its supplier confirms suitability for the actual thermal-cycling and handling duty and the qualification test demonstrates stable resistance.
FAQ
An unintended contact between platinum-wire turns is a plausible cause because it creates a lower-resistance path. The opened samples suggest this mechanism, but they do not prove whether it originated in manufacture or developed during service.
No.
Cycle the candidate through the representative -90 degrees C test and ambient recovery duty, then check resistance at a controlled reference condition before, during, and after cycling. Confirm that it remains within the required verification tolerance before deployment.
