Short answer: a TUS data system must preserve the time-aligned temperatures across documented survey locations, plus furnace/work-zone identity, sensor and instrument traceability, survey stages, events, corrections and the applicable acceptance context. A graph without that context is not a reconstructable survey record.
A Temperature Uniformity Survey evaluates temperature variation within a defined working zone under specified conditions. Sensor count, placement, stabilisation, recording interval, test duration and acceptance rules depend on the governing revision, furnace classification and customer procedure. They must be verified, not copied from a generic web checklist.
Build the survey record in layers
| Layer | Examples to preserve |
|---|---|
| Authority | Standard/customer procedure and revision, furnace class/instrument context where applicable |
| Equipment | Furnace, working-zone dimensions/identity, acquisition instrument and channel map |
| Sensors | Survey sensor ID, location, type, calibration status and authorised correction |
| Time series | Raw and corrected values as defined, common timestamps, setpoint/stage context |
| Events | Start, stabilisation, required recording period, interruption, adjustment and end |
| Result/report | Defined statistics, acceptance source, observations, disposition and report version |
Turn the approved survey plan into a data model
Before acquisition starts, give each planned location a stable identifier and map it to the physical sensor, instrument input and displayed channel. Keep those identities separate: “L3” may describe a position in the work zone, while “TC-047” identifies a particular survey sensor and “CH-06” identifies an acquisition input. If a failed sensor is moved or replaced, the record should show when the mapping changed instead of silently reusing the old label.
The same discipline applies to the furnace itself. Preserve the equipment identifier, working-zone definition, load or operating context required by the procedure, control/setpoint stage and survey plan revision. A graph titled only “Furnace 2 TUS” cannot explain which working zone, layout or configuration was actually assessed.
Time alignment is part of the measurement story
When channels are scanned sequentially, “same timestamp” may represent a scan cycle rather than simultaneous conversion. The difference may or may not matter for a slow stable survey, but the acquisition behaviour should be known. Reference and furnace-system clocks also need a defined relationship if their events will be compared.
Separate three ideas that are often collapsed into one setting: how frequently each input is measured, how frequently the live display refreshes, and how frequently durable values are written to the survey record. The configured behaviour should retain the changes and exceptions relevant to the approved method without overwhelming review with redundant data. See sampling rate versus logging interval for the system-design distinction.
Make locations unambiguous
Channel 1 is not a permanent location. Store the survey layout or location identifier used by the approved plan, and retain any change made during setup. A later reviewer should be able to connect the plotted line to a physical survey point without relying on a handwritten note that stayed in the test case.
Handle corrections transparently
If the procedure requires correction factors, preserve the entered value, sign, source calibration record and corrected result. Do not overwrite the raw reading. Version the calculation/report method so a regenerated report can be explained.
A useful report can show corrected data while still retaining the original acquisition values. That makes it possible to answer four basic questions later: what was measured, which correction source was used, how the correction was interpreted, and which calculation/report version produced the displayed result.
Plan for exceptions
- What is recorded if a survey sensor opens or becomes intermittent?
- Can local logging continue if a laptop, LAN or cloud link fails?
- How are a paused/restarted survey and operator observations represented?
- Does the report expose missing or invalid data rather than connecting a smooth line across it?
Worked record example: a sensor is replaced during setup
Consider an illustrative survey plan in which location L4 is initially mapped to sensor TC-104 on channel CH-04. During the setup check, TC-104 is found open-circuit and is replaced by TC-119 before the defined recording window begins. A weak record simply shows “CH-04” throughout. A stronger record retains the failed-sensor observation, the time of replacement, both sensor identities, the calibration/correction source for TC-119 and confirmation that the final map was the one used for the survey.
If the approved procedure permits the survey to proceed, the final graph can still use location L4 as the stable visual label. The underlying audit data explains that L4's sensor assignment changed before the test window. If a replacement or interruption occurs during the actual window, the event remains explicit for the responsible reviewer to disposition under the governing procedure.
Common mistakes that make a survey hard to defend
- Plot-only reporting: an attractive trend without equipment, plan, sensor, correction and timing context is not a complete survey pack.
- Location names that drift: reusing “TC1” to mean position, sensor and channel makes replacements impossible to reconstruct.
- Hidden gaps: drawing a continuous line across lost, open or invalid readings can imply evidence that was never recorded.
- Statistics over the wrong window: minimum, maximum or spread must use the defined analysis period, not an automatic full-file range that includes ramp-up.
- Manual correction without provenance: a typed corrected value needs the raw value, sign convention and calibration source behind it.
- Unversioned layout: changing the survey map after capture without retaining the earlier version breaks the link between physical point and channel.
- Generic pass/fail logic: the software must not substitute a memorised limit for the verified requirement and approved procedure.
What the final survey pack should let a reviewer do
A reviewer should be able to move from the report summary to the time series, then to a channel/location/sensor map, correction source, event log and raw values without guessing. The pack should distinguish automatic calculations from operator observations and identify the procedure, report template and calculation versions. If an older pack is regenerated, the system should preserve the original authoritative record and make the regenerated output traceable.
Storage also matters. A survey file on one laptop is not a long-term retrieval strategy. Define the record owner, naming/index fields, authorised access, backup, retention and export format. The principles in long-term heat-treatment log archiving apply equally to survey evidence.
Sources
- SAE International — AMS2750H Pyrometry (official revision listing; verified 2026-08-26).
- AIAG — CQI-9 Heat Treat System Assessment (official manual page; verify the governing edition before project use).
Use the TUS planning helper to organise readiness items. It deliberately avoids publishing normative sensor counts or placement rules.