Pyrometry is not just a quality record. It is the evidence that establishes whether a furnace is qualified to run a particular process, whether the resulting loads can be released, and whether customer requirements are being met. In practice, temperature uniformity surveys (TUS), system accuracy tests (SAT), instrument calibration and sensor management determine how much of a production schedule remains available. For heat treatment furnaces serving aerospace, automotive, defence or medical customers, that evidence is the licence to operate, and it is renewed on a calendar whether or not the plant is ready for it.
The expectation has shifted from “tell me you do it” to “show me the evidence.” AMS2750H is the working revision in 2026 and has been in use since 2023 for many customer flow-downs; AIAG continues to catalogue CQI-9 at its 4th edition. A plant can have extensive temperature data and still struggle to demonstrate that it used the right method, correct sensors and complete records.
The scale of that exposure deserves to be stated plainly. A pyrometry specialist who sits on the AMS2750 writing team puts pyrometry at the root of roughly 80% of Nadcap nonconformances [16]. No other single discipline in a heat treating plant carries that concentration of audit risk, and almost none is managed as casually as a box of thermocouples in a store cupboard.
All numerical figures in this article are indicative benchmarks drawn from published surveys, plant reports, regulatory and consensus standards texts, and supplier references. They vary by furnace type, class, instrumentation type, load, alloy, survey method, throughput and local regulation. Verify them against the revision of the standard and the customer specification governing your equipment.
The compliance triangle: sensors, instruments and the survey that proves both
Three tests answer different questions:
- Instrument calibration checks an instrument’s response to known simulated inputs.
- SAT checks the installed system: the control or recording system is compared against a test sensor.
- TUS establishes furnace capability within a defined qualified work zone and tolerance.
Product temperature profiling answers a separate question: what temperatures did the load experience, and did the cycle produce the required heat-treatment outcome? Confusing a load profile with TUS evidence is a recurring source of avoidable nonconformance. PhoenixTM notes that many compliance problems sit in this gap: a plant may have plenty of temperature data, but not the type of evidence the applicable standard requires. [6]
The documentation burden is real. One commercial compliance platform describes 26 separate items to check and document for each TUS. [7] A successful survey therefore depends on method, sensor traceability, correct instrument status, data collection and a complete report, not simply a “pass” result.
One distinction causes expensive confusion at the purchasing stage. Asking the furnace builder to run a uniformity survey before shipment is sensible, and it is the right moment to discover that a work zone cannot hold its tolerance. But a survey performed in the builder’s shop does not satisfy the initial TUS requirement. The buyer remains responsible for the initial TUS, the initial SAT and instrument calibration before any production hardware goes into the furnace [15]. A machine that arrives with a certificate is not a machine that is cleared to run.
Furnace class describes the permitted temperature-uniformity tolerance. Instrumentation types A through E describe the measurement and control arrangement. The class and instrumentation type together determine ongoing testing obligations; the furnace class alone does not. [1]
Table 1. Furnace class, uniformity tolerance and survey obligation, indicative 2026
| Furnace class | Uniformity tolerance | Typical duty | Documented interval example |
|---|---|---|---|
| Class 1 | ±5.0 °F (±3.0 °C) | Tightest-uniformity applications | – |
| Class 2 | ±10.0 °F (±6.0 °C) | Precision production heat treatment | Type B: monthly TUS |
| Class 3 | ±15.0 °F (±8.0 °C) | General industrial heat treatment | – |
| Class 4 | ±20.0 °F (±10.0 °C) | Broader-tolerance thermal processes | – |
| Class 5 | ±25.0 °F (±14.0 °C) | Applications allowing wider variation | Type B: quarterly TUS |
| Class 6 | ±50.0 °F (±28.0 °C) | Processes with the widest stated tolerance | – |
| Note | Furnace class and instrumentation type jointly govern test frequency. After four consecutive successful surveys, an interval may be extended where permitted. | – | Illustrative only; the revision in force governs. |
Tolerance figures are shown in the cited furnace-class reference. [3] The two interval examples and extension principle are drawn from published pyrometry guidance. [4] This table is illustrative, not a substitute for the applicable standard, its current interval tables or the customer specification.
Sensor management: the cheapest line item with the fastest drift
Sensors can dominate survey uncertainty. A peer-reviewed uncertainty assessment covering an oven, autoclave, thermal bath and refrigeration unit found thermocouple contributions of 30.1% to 51.9% of expanded uncertainty. Equipment stability and uniformity contributed 31.8% to 61.6%, while the data logger was the smallest of the three contributions. [5]
The commercial lesson is straightforward: buying a more precise logger is often the least valuable first improvement. Sensor condition, handling and furnace stability deserve attention before a capital upgrade to the acquisition instrument.
AMS2750H tightened expendable and nonexpendable sensor definitions. Braided fibreglass- or plastic-insulated sensors remain expendable even when placed in a closed protection tube; the confined space can drive drift as insulation additives burn off. [2] The H revision also caps base-metal or refractory TUS sensor reuse at 270 uses for nonexpendable sensors or six months from first use, whichever comes first. [2]
The standard clarifies what counts as a use. A single ramp from room temperature to the first survey temperature, onward to the maximum survey temperature, and then back to room temperature is one use. Two separate excursions from room temperature count as two uses. [2] A calibration agency must provide calibration points no more than 250 °F or 140 °C apart. [2] Where required, expendable base-metal sensors must meet IEC 60584-1 tolerance class 1. [5]
Good field practice makes a measurable difference:
- Weld thermocouple junctions; do not twist the wires together.
- Keep sensor tips at least three inches from any surface.
- Avoid bending and re-bending wire.
- Keep the door or wall entry point consistent. Moving the hot-to-cold transition along the wire can shift the reading.
- Use a purpose-built port rather than routing survey sensors through a door, where a port is available. [4][12][13]
Sensor type and sheath selection matter as much as handling. Base-metal types cover the mainstream heat-treating range, noble-metal types are reserved for the high end and for certain atmospheres, and the wrong choice tends to surface as drift rather than as an obvious failure [13]. Two habits protect a fleet: mark the hot zone on each sensor and never re-cut or re-bend that zone into a different position, and keep the calibrated spool’s certificate physically attached to the spool so traceability survives contact with the shop floor [12].
Sensor wear is not visible by eye. Reusing survey sensors without a controlled use count means the plant is guessing at its own measurement uncertainty, and cannot reliably demonstrate otherwise during an audit.
Correction factors, offsets and the arithmetic that fails audits
Instrument calibration must be performed as-found, with at least three simulated inputs: the minimum, maximum and at least one point in the middle third of the qualified operating temperature range. For a Class 2 furnace qualified from 300 to 1,200 °F, that means 300 °F, 1,200 °F and at least one point between 600 and 900 °F. [2]
Cumulative correction offsets may not exceed the uniformity tolerance. For a Class 2 furnace instrument, the limit is ±10.0 °F or ±6.0 °C. [2] Extrapolating correction factors beyond the calibrated range is prohibited except where performed by a recognised standards body. Between known calibration points, interpolation must be linear or use the nearest-point factor; whichever method is selected must be defined and applied consistently. [2]
A change in offset carries a further requirement: its effect across the entire qualified operating temperature range must be evaluated and documented. [2]
A classic audit finding occurs when calibration certificates show correction factors applied algebraically, but production does not apply them. In that case, the certified system and the running system are effectively different systems. This remains a common pyrometry nonconformance. [8][9]
AMS2750H also moved from requiring a calibration label on or near the instrument to requiring that the most recent successful calibration status be readily accessible at or close to the equipment. That allows a suitable digital status board to satisfy the access requirement. [2] But access does not make the arithmetic optional. Offset stacking is a particular risk: successive small corrections may each be permissible while cumulatively walking a control loop away from the process it is meant to hold.
Survey scheduling: turning the calendar into capacity
A TUS consumes furnace time as well as laboratory time. Under AMS2750H, data collection cannot begin until all TUS and furnace sensors are at least 100 °F or 55 °C below each survey temperature. Once collection begins, readings from all TUS sensors must be recorded simultaneously at least every two minutes. After stabilization, the soak continues for a minimum of 30 minutes. [2][5]
Scope depends on the equipment and application:
- A furnace qualified across multiple temperatures may require surveys at the required temperature points.
- Multi-zone furnaces require survey planning that accounts for their zones and applicable instrumentation type.
- Continuous furnaces may require consideration of both the highest and lowest traverse speeds.
- Initial and periodic surveys do not necessarily have the same scope.
- A furnace used only at a single set point may be surveyed at that one temperature. [2]
Sensor count is not a judgement call. The standard defines a minimum number of test sensor locations, assessed either volumetrically or by planar method depending on the geometry of the work zone, and a survey that under-populates its grid is nonconforming even when every reading falls inside tolerance [7]. This is where a pre-built rack pays for itself twice: it fixes the geometry once, in a quiet week, rather than improvising it against a production deadline.
Continental Furnaces rule of thumb, not a standard: budget one full shift of lost production per batch furnace for a three-temperature survey, before accounting for rig construction, sensor installation and report review. Pre-building and pre-wiring survey racks, and maintaining a dedicated survey kit, can save hours per survey. Across a furnace fleet, those saved hours compound. [4]
Two parameters now receive explicit attention in survey reporting: recovery time and lag time. Recovery time is how long the furnace takes, after a representative load is introduced, to bring all sensors back inside tolerance; lag time is the spread between the first and last sensor reaching the lower tolerance. Both appear as analysed outputs in commercially available TUS software aligned to AMS2750H, and both are referenced by material specifications such as AMS2770 for wrought aluminium parts [14]. A furnace can pass uniformity and still show poor recovery, and that is precisely the furnace that will under-soak the heaviest load on the coldest morning of the year.
Interval extension requires active control. Production must not begin after permitted extension days expire. If equipment has been out of use beyond its due date plus extension, it must be recalibrated and resurveyed before return to service. A plant may exceed the interval with approval from each cognizant engineering or quality organisation; however, the next survey must cover all temperatures required for the initial survey, and the interval continues at the frequency in use at the time of the last periodic survey. [2]
Where pyrometry quietly costs money: withdrawal, hold and scrap
A failed calibration, SAT or TUS removes the affected equipment from service. [2] Loads already inside may become suspect, requiring assessment, customer notification and action under the quality system. The commercial asymmetry is stark: a survey costs furnace hours; a withdrawal can disrupt the entire downstream plan.
Put the two costs side by side and the economics stop being arguable. A three-temperature survey on a mid-size batch furnace consumes furnace hours that are visible, planned and budgeted. A withdrawal consumes things that are not: parts already released to the floor, customer promise dates set weeks earlier, expedited freight, and in aerospace work the possibility that a customer’s own build schedule slips and the conversation turns commercial rather than technical. The survey is the cheaper half of the risk, and it is the half that can be scheduled.
Short cycles create a separate process-recording trap. Records must contain at least six data points per recorded sensor per time-at-temperature cycle, at intervals no greater than 10 minutes. If time at temperature is under six minutes, collection must occur at least once per minute. [2] Fast cycles in the wire and cable industry are particularly exposed to this requirement.
AMS2750H also assesses hot and cold recording-sensor locations individually. A TUS can pass overall and still require relocation of one sensor if that location does not meet its own assessment criteria. [2]
Nadcap heat-treat findings repeatedly involve incomplete pre-audit documentation or self-audits, missed requirement flow-down, repeat findings, missing furnace charts inside TUS packages and misunderstanding SAT extension eligibility. Repeat findings can affect merit status. [8][9][10] PhoenixTM reports that auditors increasingly ask whether the TUS method matches the cited standard, whether calibration traceability is complete, whether records are consistent across shifts and sites, and whether the same evidence can be produced again without special effort. [6]
The most instructive finding in current practice is not a failed test at all. It is maintenance completed without quality signing off that no further pyrometry testing was required: a door seal replaced, a log left unsigned, a control thermocouple changed without a corresponding alternate SAT decision [16]. That is a documentation failure with the same consequence as a failed survey, and it is by far the cheapest one to design out of a system.
Table 2. Pyrometry failure modes, what they look like and the first corrective action
| Symptom or finding | Underlying cause | First corrective action |
|---|---|---|
| TUS passes, then fails at the next interval with no process change | Sensor reuse limit or drift accumulated in control thermocouples; no use log | Reconcile sensor batch and use records; inspect control sensors and repeat testing as required |
| SAT discrepancy on one zone only | Fractured or misplaced control sensor, wrong extension wire, or correction factor applied on paper only | Check the sensor-to-instrument chain and verify production offset application |
| Recurring audit finding on calibration records | Generic metrology sticker template does not identify instrument or furnace number and omits pyrometry-specific requirements | Replace the template and audit records against the governing pyrometry requirements [9] |
| Survey fails at one temperature in a multi-temperature survey | Hot or cold location shifted, load or rack change, zone-tuning drift, or insulation loss | Review location and configuration history; investigate the failing temperature and zone |
| Short-cycle process records rejected | Fewer than six data points or collection interval over 10 minutes on a fast cycle | Confirm the cycle time and set collection frequency to meet the applicable rule [2] |
| Equipment returned to service after an idle period without tests | Out-of-use rules not applied to calibration, SAT and TUS | Verify due dates and extension status; complete required tests before production [2] |
Symptoms overlap. A metallurgical or pyrometry review should precede corrective action.
The pattern behind most of these rows is the same: the furnace did not change, the record did. Equipment that passed last quarter and fails this quarter without a process change is telling you that a sensor aged, a location drifted or a correction stopped being applied, not that the refractory suddenly went bad. Read the record before you rebuild the furnace.
2026 technology and market updates
AMS2750H is the working revision, and it bites in maintenance, not in paperwork
The changes most likely to affect shop-floor routines include the 270-use and six-month sensor caps, individual hot- and cold-location assessment, the prohibition on correction-factor extrapolation, and full-range evaluation of new offsets. TUS data must be recorded simultaneously at two-minute intervals. Alternate SAT is triggered by events such as sensor replacement, roll or spool replacement, or instrument calibration, not merely by a paper frequency. [2]
CQI-9 stays at the fourth edition while the interpretation sharpens
AIAG continues to catalogue CQI-9 at the 4th edition. [11] For automotive supply chains, the pressure is increasingly on repeatable evidence: consistent survey methods, traceable calibration records and procedures that do not depend on one employee knowing the unwritten process. [6]
Rev J is already in draft, and the checklists move with it
The AMS2750 committee has restarted work on a Rev J, expected to be a clarification cycle rather than a rewrite: the substantive engineering in G and H is broadly sound, and the remaining work is about closing interpretation gaps and keeping the intent of each requirement visible [16]. Two practical instructions follow. First, any internal pyrometry procedure that quotes a clause number needs a revision field and a review date, because a supplier checklist revision lands shortly after every specification revision. Second, the people who will have to live with the next revision should be in the room where it is written; supplier representation on the AMS2750 and Nadcap task groups is open, and it is where interpretation questions actually get settled [16].
Digital pyrometry records consolidate
Compliance platforms are integrating with logger brands including Fluke, Honeywell, SSi, Datapaq, Eurotherm, Phoenix, Yokogawa and Future Design. Some also bring in supplier calibration data and certificates, accept results from ISO/IEC 17025-accredited pyrometry laboratories, and provide furnace-by-furnace status views. [7] Survey software report templates can pre-populate required TUS criteria to reduce omissions. [6][14]
The direction of travel for 2026 is gating rather than reporting. Compliance systems are being integrated with production scheduling so that a job cannot be released to a furnace whose calibration, SAT or TUS is past due, replacing the hope that an operator notices a wall chart with a system that simply refuses the transaction [16]. For a plant running more than a handful of furnaces, that is the highest-value control available, because it converts a quality argument into a scheduling constraint that nobody can quietly bypass at two in the morning.
This is a market trend, not an endorsement of any single platform. Software cannot repair poor sensor management or an offset error. A dashboard is only as reliable as the evidence and records entered into it, and a lockout is only as good as the asset register behind it.
Uncertainty budgets move into the conversation
AMS2750 requires third-party pyrometry service providers to hold ISO/IEC 17025 accreditation from an ILAC-recognised body, with a scope that covers relevant laboratory standards and field service. [2] ISO/IEC 17025 brings uncertainty evaluation into the laboratory’s work; uncertainty analysis is therefore appearing in TUS reporting even though the aerospace specification itself does not mandate a particular calculation. [5]
Plant engineers should ask for the uncertainty breakdown. It identifies where improvement is likely to pay back. If the equipment contribution dominates, the next investment should focus on maintenance and furnace performance, not automatically on metrology spend.
What this means across your furnace fleet
- Heat treatment furnaces: In aerospace and automotive flow-downs, specify furnace class and instrumentation type in the purchase specification, not after installation. [15] Continental Furnaces’ heat treatment furnace range can be considered alongside qualification and surveyability requirements.
- Steel rolling mill operations: Roll-heat-treatment, normalising and annealing furnaces compete for the same available hours as the rolling programme itself, so in a steel rolling mill the survey window belongs in the production plan rather than in a quality folder. Survey access, zone symmetry and sensor ports are worth arguing about at specification stage, when they are free.
- Aluminium processing: For an aluminum melting furnace and holding furnaces, radiation-survey obligations can apply when aluminium solution treatment above 800 °F is required by a material or process specification. [2] Control-sensor discipline also supports yield. Consider the applicable melting and recycling furnace range.
- Steel melting and recycling: A melting furnace for steel or a metal recycling furnace depends on sensor survival and suitable protection for reliable process control. See melting and recycling furnace solutions.
- Galvanizing: In a hot dip galvanizing plant, sensor discipline supports consistent preheat control and process repeatability. Explore the hot dip galvanizing plant.
- Wire and cable industry: Annealing, tinning and coating lines often run fast cycles. The six-point data rule can expose recording gaps that remain hidden on slower batch processes.
- Spare parts readiness: Survey-grade thermocouples, protection tubes, extension wire, connectors and recorders prevent a survey from waiting on a purchase order. Review furnace spare parts and accessories.
- System design: Industrial furnace systems depend on a chain: the furnace, instruments, recorder and quality system must all work together. A capable industrial furnace manufacturer designs for surveyability through suitable sensor ports, a documented work zone, accessible calibration status and zone symmetry. That is the practical definition of good thermal processing equipment: not the fastest ramp on the datasheet, but the machine whose evidence is easy to produce on a Tuesday afternoon in the middle of a production campaign.
Across sectors, the same logic applies to an aluminum melting furnace: compliance evidence and process control are connected, not competing objectives.
A four-phase roadmap to pyrometry as an asset
Treat the roadmap as a sequence of evidence, not a list of purchases. Each phase should leave the plant able to answer one more question from its own records without calling anyone.
Phase 1: Asset register and class verification (30–60 days)
- List every furnace, qualified work zone, assigned class and instrumentation type.
- Link each part number or process to the customer and process specification that governs it.
- Flag equipment operating outside its qualified zone.
- Start a controlled sensor-use log.
Phase 2: Procedure and record repair (1–2 quarters)
- Rebuild the TUS and SAT calendar, including extension tracking.
- Verify that correction offsets are applied in production as well as on calibration certificates.
- Make calibration status visible at the furnace.
- Pre-build survey racks and standardise required records.
Phase 3: Sensor and survey logistics (3–6 months)
- Establish batch traceability for expendable sensors.
- Set a single-use policy for critical surveys where appropriate.
- Train staff in sensor storage and handling.
- Maintain a dedicated survey kit and name a survey owner for each shift.
Phase 4: Systems and capital planning (6–18 months)
- Link pyrometry records to a compliance platform and, where applicable, the historian for evidence retrieval.
- Specify class, instrumentation type, sensor ports and documented work zone in every new furnace purchase.
- Plan instrumentation upgrades, such as digital recorders, time synchronisation to ±1 minute per hour and redundant sensors, against measured survey-failure costs. [2]
The sequencing matters more than the speed. Phase 1 is the only phase that cannot be skipped, because a plant that does not know which class each furnace is qualified to cannot tell whether its testing burden is correct, excessive or absent, and it will usually be paying one of those three prices without knowing which. Teams that buy software before the asset register exists end up with an immaculately documented version of the same problem.
An enduring partnership in thermal compliance
For more than 35 years, Continental Furnaces has developed thermal processing solutions for industrial customers. Our ISO-certified quality, customised furnace solutions and prompt service are built around lifecycle value and keeping downtime minimal.
Pyrometry readiness should shape furnace selection, survey access, instrumentation, spares and service planning from the outset. Discuss furnace selection, surveyability, spare parts and instrumentation with our engineering team through the request-for-a-quote page. Make compliance a production asset, and a strategic step toward sustained competitive advantage.
Technical references
- Furnace classifications and how they relate to AMS2750: https://www.heattreattoday.com/featured-news/furnace-classifications-and-how-they-relate-to-ams2750/
- AMS2750G to AMS2750H comparison, technical paper: https://docs.heattreat.net/2750GHComparison2024.pdf
- AMS2750H, CQI-9 and furnace specifications reference, Keith Company: https://www.keithcompany.com/ams2750g-cqi9-furnace-specifications.html
- Temperature uniformity survey general practice for CQI-9 and AMS2750, Thermal Processing: https://thermalprocessing.com/temperature-uniformity-survey-%e2%80%8atus%e2%80%8a-general-practice-for-cqi-9-and-ams2750e/
- Uncertainty assessment in temperature uniformity survey of equipment according to AMS2750, Thermal Processing: https://thermalprocessing.com/uncertainty-assessment-in-temperature-uniformity-survey-of-equipment-according-to-ams2750/
- AMS2750H and CQI-9: TUS evidence scrutiny, PhoenixTM: https://www.phoenixtm.com/2750h-cqi-9-tus-evidence-scrutiny/
- TUS module and AMS2750 documentation requirements, C3 Data: https://www.c3data.com/modules/tus
- Key changes in AMS2750G versus AMS2750H pyrometry requirements, Conrad Kacsik: https://www.kacsik.com/blog/key-changes-ams2750g-vs-ams2750h-pyrometry-requirements
- Top PRI nonconformances in heat treating, Conrad Kacsik: https://www.kacsik.com/blog/pri-top-nrcs-in-heat-treating
- The 2021 PRI top 10 NRCs in heat treating, Thermal Processing: https://thermalprocessing.com/the-2021-pri-top-10-nrcs-in-heat-treating/
- CQI-9 special process heat treat system assessment, AIAG: https://www.aiag.org/training-and-resources/manuals/details/CQI-9
- Ten tips to ensure accurate thermocouple measurement, Thermal Processing: https://thermalprocessing.com/10-tips-to-ensure-accurate-thermocouple-measurement/
- Understanding thermocouples and their use in thermal processing, Conrad Kacsik: https://www.kacsik.com/blog/understanding-thermocouples-in-thermal-processing
- Release of Thermal View survey software V5.7, PhoenixTM: https://www.phoenixtm.com/release-of-new-thermal-view-survey-software-v5-7/
- Purchasing a furnace: quality concerns to consider, Thermal Processing: https://thermalprocessing.com/purchasing-a-furnace-quality-concerns-to-consider/
- Heat Treat Radio 130: AMS2750 pyrometry best practices, Heat Treat Today: https://www.heattreattoday.com/heat-treat-radio-130-ams2750-pyrometry-best-practices/
