Toolchest · Datasheet

Accumath Toolchest — Thermocouple Reference Tables (NIST ITS-90)

Data file: accumath-thermocouple-tables.csv Columns: type, temp_C, emf_mV, source Rows: 1,211 data rows (plus header)

Basis and conventions

• Scale: International Temperature Scale of 1990 (ITS-90). All temperatures are ITS-90 temperatures in °C; all EMFs are in millivolts (1990 SI volt).
• Reference junction: 0 °C. Every EMF in the file is the thermoelectric voltage of the thermocouple with its reference (cold) junction at 0 °C and its measuring junction at temp_C. EMF is 0.000 mV at 0 °C for every type by construction.
• Source standard: NIST Monograph 175, Temperature-Electromotive Force Reference Functions and Tables for the Letter-Designated Thermocouple Types Based on the ITS-90 (Burns et al., 1993), as distributed in the NIST ITS-90 Thermocouple Database, SRD 60 (version 2.0). US government work — public domain.
• Step: every 10 °C for all eight types, including the noble-metal types R, S, and B (no 20 °C fallback was needed).
• Rounding: EMF rounded to 3 decimals, as in the NIST printed tables.
• Letter designations follow ISA/ASTM practice: K = nickel-chromium / nickel-aluminium, J = iron / copper-nickel, T = copper / copper-nickel, E = nickel-chromium / copper-nickel, N = Nicrosil / Nisil, R = platinum-13 % rhodium / platinum, S = platinum-10 % rhodium / platinum, B = platinum-30 % rhodium / platinum-6 % rhodium.

Coverage

Type	Class	NIST range (°C)	Rows	EMF at range ends (mV)
K	base metal	−270 to 1372	166	−6.458 to 54.886
J	base metal	−210 to 1200	142	−8.095 to 69.553
T	base metal	−270 to 400	68	−6.258 to 20.872
E	base metal	−270 to 1000	128	−9.835 to 76.373
N	base metal	−270 to 1300	158	−4.345 to 47.513
R	noble metal	−50 to 1768.1	183	−0.226 to 21.101 (at 1768 °C)
S	noble metal	−50 to 1768.1	183	−0.236 to 18.693 (at 1768 °C)
B	noble metal	0 to 1820	183	0.000 to 13.820

Endpoint convention: the grid is every 10 °C across each type's full NIST range. Type K's upper limit, 1372 °C, is off the 10 °C grid and is included as a final row. Types R and S are defined to 1768.1 °C; their last row is 1768 °C, the last integer-degree point of the NIST table (21.101 mV for R, 18.693 mV for S).

How the values were produced (route used)

The CSV values were computed from NIST's own published forward reference-function polynomials — route 2 of the build plan — and then verified against NIST's published tables at every single grid point.

For each type, NIST publishes a polynomial reference function E(t) = Σ cᵢ tⁱ (type K above 0 °C adds an exponential term a₀·exp(a₁(t − a₂)²)). These polynomials are not approximations of the tables; they are the defining reference functions of Monograph 175, and NIST's printed tables are generated from them. The forward coefficients were transcribed from the coefficient section of each type's file in the NIST SRD 60 distribution:

• https://its90.nist.gov/downloadFiles/type_k.tab.txt
• https://its90.nist.gov/downloadFiles/type_j.tab.txt
• https://its90.nist.gov/downloadFiles/type_t.tab.txt
• https://its90.nist.gov/downloadFiles/type_e.tab.txt
• https://its90.nist.gov/downloadFiles/type_n.tab.txt
• https://its90.nist.gov/downloadFiles/type_r.tab.txt
• https://its90.nist.gov/downloadFiles/type_s.tab.txt
• https://its90.nist.gov/downloadFiles/type_b.tab.txt

No third-party coefficient set (Omega or otherwise) was used; no value came from memory.

Polynomial subranges (°C)

Type	Subranges
K	−270–0; 0–1372 (with exponential term)
J	−210–760; 760–1200
T	−270–0; 0–400
E	−270–0; 0–1000
N	−270–0; 0–1300
R	−50–1064.18; 1064.18–1664.5; 1664.5–1768.1
S	−50–1064.18; 1064.18–1664.5; 1664.5–1768.1
B	0–630.615; 630.615–1820

Verification against the published NIST tables

The same SRD 60 files contain NIST's published 1 °C tables. The value at every 10 °C grid point of those tables (1,211 points) was compared with the computed value:

• 1,211 of 1,211 points agree exactly after rounding to 3 decimals (100 % match, all eight types).
• The largest difference between any unrounded computed value and its published table value is 0.00050 mV — i.e. nothing exceeds the maximum possible rounding distance of a 3-decimal table. There were no rounding-boundary (0.001 mV) discrepancies and no outliers.

Representative spot checks (computed = published):

Type	t (°C)	EMF (mV)		Type	t (°C)	EMF (mV)
K	−270	−6.458		N	−270	−4.345
K	−100	−3.554		N	−100	−2.407
K	100	4.096		N	100	2.774
K	500	20.644		N	600	20.613
K	1000	41.276		N	1300	47.513
K	1372	54.886		R	−50	−0.226
J	−210	−8.095		R	100	0.647
J	−100	−4.633		R	1000	10.506
J	100	5.269		R	1768	21.101
J	760	42.919		S	−50	−0.236
J	1200	69.553		S	100	0.646
T	−270	−6.258		S	1000	9.587
T	−200	−5.603		S	1768	18.693
T	100	4.279		B	0	0.000
T	400	20.872		B	100	0.033
E	−270	−9.835		B	500	1.242
E	−100	−5.237		B	1000	4.834
E	100	6.319		B	1820	13.820
E	1000	76.373

Accuracy note — what these tables are and are not

• These tables are definitional: on ITS-90, the reference function is the EMF–temperature relationship for the letter-designated type. There is no "measurement error" in a table value; the computed values above reproduce NIST's published tables to the last printed digit at every point checked.
• Physical thermocouple tolerances are a separate matter and are NOT included. How closely a real wire/probe follows the reference function is governed by the limits of error in ASTM E230/E230M (standard and special grades) and by calibration. Nothing in this dataset certifies any physical sensor.
• Type B quirk, visible in the data: near ambient temperature its EMF is slightly non-monotonic (0.000 mV at 0 °C, dipping to −0.003 mV at 20 °C, back to 0.000 mV at 40 °C). This is a genuine property of the type B reference function, reproduced here as published — one reason type B is not used for low-temperature measurement.
• For field use with a reference junction other than 0 °C, apply standard cold-junction compensation: add the table EMF at the reference-junction temperature to the measured EMF before inverting. (The inverse functions — EMF → temperature — are also published in Monograph 175 / SRD 60 but are approximate and are not part of this dataset.)