Thermocouple Accuracy and Adherence to Critical

Standards

Andrew Bassett – Aerospace Testing and Pyrometry; Bethlehem, Pa.

Eric Yeager – Cleveland Electric Laboratories; Twinsburg, Ohio
A discovery by Thomas J. Seebeck in 1821 paved the way for the modern
thermocouple, and today it is the most widely used electrical sensor in
thermometry.
Fig. 1. Seebeck Effect
Fig. 2. Thermocouple circuit
Seebeck discovered that when two dissimilar metals are connected (“junctioned”)
together, a minute electric potential will be generated. This phenomenon, known as the
Seebeck Effect, is the basis of how a thermocouple functions. A common misconception
is that the thermocouple junction itself generates the EMF (electromotive force);
however, a junction simply joins the wires together. In fact, the EMF – measured in
millivolts – is generated along the length of the two materials as the temperature
gradient changes from one end to the other end of the thermocouple. Therefore, it is
critical to have thermocouple wire with consistent, homogeneous composition from one
end of the wire to the other end so that a uniform signal may be produced.

Thermocouple wires, also known as thermoconductors, are manufactured from specific

alloy compositions and will vary slightly in material composition between any two lots (or
“melts”) of the specific alloy. This inherent variance between melts is the initial source for
thermocouple calibration deviation and the reason all thermocouples possess some level
of error.

1
Error and Tolerance
The amount of thermocouple error is quantifiable and designated as thermocouple
accuracy (or tolerance). Tolerance is defined for the various thermocouple types in ASTM
E230, Standard Specification for Temperature-Electromotive Force (EMF) Tables for
Standardized Thermocouples, which specifies the allowable error by thermocouple type.
Error indicates how close temperature measured by a given thermocouple is to the actual
temperature value and necessitates the need to calibrate all thermocouple wire or cable.

Once calibrated, a thermocouple calibration report indicates the “initial calibration

tolerances” and informs the user what tolerance they can expect from a given
thermocouple or lot of thermocouples. Tolerances are typically available as two separate
accuracy grades: “Special” or “Standard” tolerances of error (think high-octane vs.
low-octane gas). They both get you from point A to point B, but one performs better and
gets you better fuel economy.

2
To determine the errors and homogeneity of a given lot of material, a sample is taken
from each end of the new, unused lot of thermocouple wire or cable. It is then
manufactured into thermocouples for calibration testing. Using a specialized laboratory
furnace, the test thermocouples are tested alongside a calibrated reference standard at
given intervals. The test results are analyzed by the laboratory and reviewed against
ASTM E230, Table 1 Tolerances on Initial Values of EMF vs. Temperature for
Thermocouples, to determine whether a lot of material falls within Standard or Special
tolerances or other industry specifications.

NOTE 1:
Tolerances in this table apply to new essentially homogeneous thermocouple wire,
normally in the size range 0.25 to 3 mm in diameter (No. 30 to No. 8 AWG) and used at
temperatures not exceeding the recommended limits of Table 6. If used at higher
temperatures these tolerances may not apply.

NOTE 2:
At a given temperature that is expressed in °C, the tolerance expressed in °F is 1.8 times
larger than the tolerance expressed in °C. Where tolerances are given in percent, the
percentage applies to the temperature being measured when expressed in degrees
Celsius. To determine the tolerance in degrees Fahrenheit, multiply the tolerance in
degrees Celsius by 9/5.

NOTE 3:
Caution: Users should be aware that certain characteristics of thermocouple materials,
including the emf-versus-temperature relationship may change with usage; consequently,
test results and performance obtained at the time of manufacture may not necessarily
apply throughout an extended period of use. Tolerances given in this table apply only to
new wire as delivered to the user and do not allow for changes in characteristics with use.
The magnitude of such changes will depend on such factors as wire size, temperature,
time of exposure, and environment. It should be further noted that due to possible
changes in homogeneity, attempting to recalibrate used thermocouples is likely to yield
irrelevant results, and is not recommended. However, it may be appropriate to compare
used thermocouples in-situ with new or known good ones to ascertain their suitability for
further service under the conditions of the comparison.

*The standard tolerances shown do not apply to Type E mineral-insulated,

metal-sheathed (MIMS) thermocouples and thermocouple cables as described in
Specifications E608/E608M and E585/E585M. The standard tolerances for MIMS Type E
constructions are the greater of ± 2.2°C or ± 0.75% from 0°C to 870°C and the greater of
± 2.2°C or ± 2% from -200°C to 0°C.
A
Thermocouples and thermocouple materials are normally supplied to meet the
tolerances specified in the table for temperatures above 0°C. The same materials,
3
however, may not fall within the tolerances for temperatures below 0°C in the second
section of the table. If materials are required to meet the tolerances stated for
temperatures below 0°C the purchase order shall so state. Selection of materials usually
will be required.
B
Special tolerances for temperatures below 0°C are difficult to justify due to limited
available information. However, the following values for Types E and T thermocouples
are suggested as a guide for discussion between the purchaser and supplier:

Type E, -200°C to 0°C, ± 1.0°C or ± 0.5% (whichever is greater)

Type T, -200°C to 0°C, ± 0.5°C or ± 0.8% (whichever is greater)

Initial values of tolerance for Type J thermocouples at temperatures below 0°C and
special tolerances for Type K thermocouples below 0°C are not given due to the
characteristics of the materials. Data for type N thermocouples below 0°C are not
currently available.

Initial values of tolerance for Type J thermocouples at temperatures below 0°C and
special tolerances for Type K thermocouples below 0°C are not given due to the
characteristics of the materials. Data for type N thermocouples below 0°C are not
currently available.

The laboratory produces a report for the material stating the tolerance limits and error
values of the lot of material at a given temperature. The reported errors are to be applied
to new, unused thermocouple materials prior to use and are commonly used as the
acceptance criteria for the thermocouples and/or material.

Since the thermocouple is simply two wires, there is no adjustment knob to dial the
material into any specific error or range. They are calibrated “as found,” necessitating the
need to apply the errors to correct the thermocouple to the exact temperature.
Thermocouple tolerance is represented in degrees and/or percentage of reading. The
percentage applies to the temperature of interest. For example, a special-limits tolerance
of ± 0.4% for a Type K thermocouple at 1000°F would be ± 4°F, whereas a tolerance of
± 0.4% at 2000°F would be ± 8°F.

Since there is no single thermocouple type that may be used in every application at every
temperature, different letter-designated thermocouple types have been established and
are manufactured to cover different temperature ranges and applications. Base-metal
thermocouple Types E, J, K, N and T are best suited for low temperatures and cryogenic
ranges. Noble-metal thermocouple Types S, R and B are better suited for elevated
temperatures and are less susceptible to “drift” when properly protected from the
environment.

4
Fig. 3. Calibration certification

5
6
Fig. 4. Survey thermocouple wire
Fig. 5. Thermocouple EMF vs. time

Thermocouple Accuracy
The typical way a thermocouple “fails” is not from physical breakage but rather from
losing its accuracy. By nature, a thermocouple gets more and more inaccurate as it is
used in temperatures over time. Once the accuracy requirement for the thermocouple is
no longer achieved, it is said to have “drifted out of tolerance.” As a thermocouple is
heated, changes start to take place to the metallurgical structure of the wire, including
chemical changes from the environment around the thermocouple. These reactions are
accelerated at elevated temperatures.

An essential factor to consider is the need to select a proper thermocouple configuration

that fits within your specific application’s operating range while not approaching the
upper limit of the thermocouple’s maximum operating capabilities. As changes in the
thermocouple chemistry occur over time and temperature, the thermocouple’s millivolt
signal deviates from its initial calibration, thereby creating drift. The result is that the
thermocouple, while producing a reading, will no longer accurately measure temperature
within its stated tolerance.

7
Once a thermocouple drifts out of its stated tolerance it cannot be adjusted or calibrated
to correct it. Drift is irreversible, and the thermocouple simply needs to be replaced.
Drifting may start within a few minutes or may take many months depending on how the
thermocouple is used. Therefore, tolerances are stated as “initial calibration.” The initial
calibration tolerance is valid for the first use. There is no guarantee the tolerance will be
maintained after that. This is the reason why a user’s application might require that they
use a thermocouple only once and dispose of it, while another application will allow a
thermocouple to last up to a year or longer.

Drift is caused by minor differences in the chemistry of the two dissimilar metals making
up the thermocouple assembly. It may seem insignificant relating to the metallurgy of the
wires themselves. Once a contaminant is introduced, however, the EMF output of the
thermocouple can be compromised. The effect of this reduction in EMF varies from
thermocouple to thermocouple due to the amount of signal produced by the different
thermocouple types.

Table 2 shows three different thermocouple types and the EMF difference between 500°F
and 501°F. The EMF output for a Type K thermocouple has a 23-microvolt difference
between 500°F and 501°F, whereas a Type B thermocouple has only a 2-microvolt
difference between the two temperatures. For reference, an AA battery that you find in
your TV remote produces just 1.5V.

As you can see, a minor change in EMF output of the thermocouple can greatly affect the
accuracy. The net effect of drift is the thermocouple reducing its EMF output. If used as a
control thermocouple, it may cause the furnace to call for more heat, which can
potentially produce an over-temperature condition on the load.

With the EMF output of thermocouples being such small signal values, it is essential that
the complete measuring system is properly maintained. This includes assuring any
8
connecting hardware – such as plugs, jacks and connecting wires – is in good physical
condition and all connections are tight, clean and free of oxidation. Even a small amount
of dirt or tarnish to terminal connections may have a detrimental effect on complete
system accuracy.

Fig. 6. Vacuum furnace thermocouple

Pyrometry Requirements and How Drift is Addressed

There are ways to ascertain how far a thermocouple has drifted to help determine the
remainder of its useful life. One way is to check the thermocouple in situ, or while it is still
in use. Some industry specifications or manufacturer requirements – AMS 2750F,
Pyrometry and Boeing BAC5621 – require this validation. This is achieved by inserting a
certified standard thermocouple as close as practical to the measuring junction of the
thermocouple (within 3 inches), measuring the temperature with an independent
standard instrument and then comparing the readings to see if the thermocouple is
operating within accuracy requirements.

Another method is to periodically remove the thermocouple and have it calibrated in an

accredited laboratory to determine whether or not it has drifted. This practice will aid in
establishing a preventive-maintenance schedule and assuring accurate thermocouples
were in service, thus producing an acceptable product or process quality. Utilizing these
methodologies allows the user to develop a history for a particular application and come
up with a predictive model for thermocouple life determination.

Fig. 7. Temperature uniformity survey (TUS)

9
Recent Updates to AMS 2750 Pyrometry Requirements and

Use
In AMS 2750 revision F, Pyrometry, released June 2020, many changes have been made
to address concerns pertaining to thermocouple use and reuse and what controls shall be
in place to protect from drift issues. This is specifically stated in Section 3.1.4.4: “User’s
procedures shall control the replacement frequency of thermal-processing equipment
sensors including limits on maximum life and/or number of uses based on supporting
data such as, but not limited to, SAT (system accuracy test), TUS (temperature
uniformity survey) and re-calibration and/or trend analysis.” The most significant updates
are in regard to accuracy requirements, sensor use and reuse, record keeping and
self-regulation, additional sensor types and calibration reporting requirements.

The first change to note is the initial accuracy requirements for all base-metal
thermocouples. AMS 2750F Table 1 requires that all base-metal thermocouples used for
testing – such as SAT or TUS sensors – as well as control, recording and load
thermocouples, shall now be ± 2.0°F (± 1.1°C) or 0.4% of reading, whichever is greater.

The initial accuracy requirements for noble-metal sensors – Types R and S – shall be
± 1.0°F (± 0.6°C) or 0.1% of reading, while Type B shall be ± 1.0°F (± 0.6°C) or 0.25% of
reading. These tolerance requirements for both base-metal and noble-metal
10
thermocouples match those as stated in ASTM E230, Table 1 Tolerances on Initial Values
of EMF vs. Temperature for Thermocouples as special limits of error. Additionally,
another major change to Table 1 is that base-metal thermocouples are no longer allowed
to be used as secondary standard sensors to calibrate lesser thermocouples, such as
testing thermocouples (SAT, TUS), load sensors and control/monitoring sensors.

Temperature-sensor reuse, as it relates to mitigating sensor drift, is another significant

change in AMS 2750F. Expendable base-metal test and load thermocouples K, T, E and M
(nickel/nickel-moly) now have specific restrictions on how many times they may be
reused. These thermocouple types, when used to perform an SAT or used as load
sensors above 500°F (260°C) and below 1200°F (650°C), will be limited to five uses. In
the same range, Types J and N are less restricted with an allowable 10 uses, and those
above 1200°F (650°C) are limited to a single use (Table 5 and section 3.1.7.3).

Sensor use restrictions, as stated in AMS 2750F, are to be considered the minimum
baseline for record-keeping and self-regulation requirements. Since each application may
exhibit unforeseen influences affecting accuracy, each application should be reviewed
and analyzed for the validity of use. Records will need to be maintained of the
accumulated sensor reuse (inclusive of batch number, temperature and use count) to
determine the actual usable lifespan without exceeding the stated requirements. The
onus is on the end user to define when the sensors that control thermal-processing
equipment shall be replaced and have the data to back this up.

Three additional sensor types were added to AMS 2750F that were not addressed in
previous revisions. Nickel-nickel/moly thermocouples have been added, designated as
Type M in AMS 2750F. Although Type M is not recognized as a sensor type designation by
ASTM or other international standards, these thermocouples have been in use for quite
some time as an alternative to Type K. Additionally, Type C thermocouples were added.
Comprised of tungsten-rhenium alloys, Type C is neither base-metal nor noble-metal
thermocouple but is considered a refractory thermocouple that is used primarily in
extremely high-temperature applications such as hot isostatic pressing (HIP) and
vacuums. Lastly, although they are not a thermocouple, RTD (resistance temperature
device) sensors were added as another device to be utilized in thermal-processing control
and monitoring. The RTD shall be platinum-type and meet class A requirements per
ASTM E1137, IEC 60751 or other internationally recognized standards.

The last significant change is to the reporting requirements for thermocouple calibration
certificates per AMS 2750F, which include the following (italicized items represent the
new requirements):

11
Fig. 8. Temperature uniformity survey (TUS) on a vacuum furnace
 Identification of the sensor, batch or wire/cable roll

 Sensor type (e.g., K, N, RTD, etc.)

 Date of calibration or recalibration

 Quantity or length represented in calibration report

 Identification if the calibration was initial or a recalibration

 The required calibration accuracy

 Identification of test sensor and test instrument used

 Nominal calibration temperatures

 Actual temperature readings under test

 Calibration technique referencing ASTM E220 or other internationally

recognized standard

 Correction factors or deviations/errors for each calibration temperature

including the average correction factor representing both ends for wire/cable
rolls

12
  Documentation shall clearly state deviation (error) or correction factor

 Statement of traceability to NIST or other internationally recognized standard

organization

 Identification of calibration agency

 Identification of technician performing calibration

 Approval of authorized agent for the calibration agency

 User quality-organization approval

Summary
Since their discovery 200 years ago, thermocouples continue to prove themselves as
reliable, repeatable and accurate temperature measurement devices when properly
monitored and controlled. The selection and application of thermocouples, in conjunction
with adherence to stringent industry standards, can be achieved when offsets are applied
and operational use is properly monitored and documented.

The consideration of quality standards held by a thermocouple manufacturer, and their

accreditations, is key in achieving compliance to industry standards. Within two years
from its release date, AMS 2750F will require that all pyrometry service providers hold
ISO 17025 accreditation. To assure compliance, an accredited field-service company may
serve as a valued partner to assist in overall system review, accuracy and documentation.
After all, pyrometry should be viewed as a complete system rather than an individual
sensing device.

13