Thermocouple Type Table
| Type | Continuous (°C) | Short-term (°C) | Metal pair / role |
|---|---|---|---|
| K | 0 to +1,100 | -180 to +1,370 | Nickel-chromium / nickel-aluminium - the industrial default |
| J | 0 to +750 | -180 to +800 | Iron / constantan - older US installs, dry locations |
| N | 0 to +1,100 | -270 to +1,300 | Nicrosil / nisil - the drift-resistant K upgrade |
| R | 0 to +1,600 | -50 to +1,700 | Platinum-rhodium 13% - lab and kiln work |
| S | 0 to +1,600 | -50 to +1,750 | Platinum-rhodium 10% - the calibration standard |
| B | +200 to +1,700 | 0 to +1,820 | Pt-30%Rh / Pt-6%Rh - cold junction ignorable |
| T | -185 to +300 | -250 to +400 | Copper / constantan - cryogenic to cold-chain |
| E | 0 to +800 | -40 to +900 | Nickel-chromium / constantan - highest millivolt output |
A thermocouple is two dissimilar metal wires joined at one end: heat that junction and the open end develops a few millivolts that map to temperature. The letter - K, J, T, B and friends - names the metal pair, and the pair fixes everything in the table: how hot the sensor can live (continuous range), how far it can be pushed briefly (short-term), and roughly how many microvolts per degree it produces. K (nickel-chromium over nickel-aluminium) is the default of industry because it covers 0 to +1,100 degrees Celsius continuously and is cheap; T (copper over constantan) owns the cryogenic-to-mild end; R, S and B spend platinum to reach past 1,600.
Two subtleties make the table practical rather than trivia. First, the continuous-versus-short-term gap is about drift, not melting: pushing a K-type to 1,300 degrees repeatedly recalibrates it (the emf wanders), while a one-time excursion does no lasting harm. Second, B-type has a party trick - below about 50 degrees Celsius its output is nearly zero, so the reference (cold) junction that every other type must compensate for simply does not matter in its 200-to-1,700 degree working band. Choose by ceiling, atmosphere, and budget, in that order.
How to use
- Pick the ceiling first: working temperature under 300 degrees and needing accuracy near room conditions points to T; general industrial to 1,100 is K; above 1,100 leaves only R, S or B - platinum money.
- Match the extension wire to the type: each letter has a matched compensating cable and, under IEC 60584-3, its own color code - mixing extension leads between types silently rescales readings.
- Check drift expectations: N-type was engineered as the K upgrade (less chromium oxidation drift at high temperature), and any type repeatedly run near its continuous ceiling should be recalibrated or derated - the short-term column is not a lifestyle.
Frequently asked questions
What is the difference between K and J type thermocouples?
Both are cheap base-metal pairs, but they age differently and top out differently. K (nickel-chromium / nickel-aluminium) runs continuously to +1,100 degrees Celsius and is the industrial default; J (iron / constantan) tops at +750 continuous but outputs more microvolts per degree in its band, which older analog instruments liked, and it survives reducing atmospheres that poison K's chromium. J's iron leg rusts in damp air, which is why J lives mostly in older US installations and dry,oily machine environments while new designs default to K or the drift-resistant N. The dangerous mismatch is invisible: K and J extension wire colors differ by standard, but the connectors and probes look alike - plugging a J probe into a K-input meter is one of the most common calibration errors, and it shifts every reading by tens of degrees.
Why does a B-type thermocouple not need cold-junction compensation?
Because its output curve is flat near room temperature. Every thermocouple reading is the difference between the measuring junction and the reference (cold) junction, so instruments normally measure or simulate the cold-end temperature to correct the emf. B-type (platinum-30% rhodium over platinum-6% rhodium) produces so little voltage below about 50 degrees that the cold junction contributes essentially nothing across its +200 to +1,700 working band - the table shows its continuous range starting at +200 for exactly this reason. The trade-offs are the mirror image: B cannot read anything cold, its sensitivity (microvolts per degree) is the lowest of the family, and it uses the most precious metal. It is the specialist for kilns, glass and high-temperature furnaces where the cold end can simply be ignored.
Which thermocouple type is best for low temperatures?
T-type (copper / constantan), and it is not close. Its table row is the only one centered below zero: -185 to +300 continuous, usable short-term to -250, with the copper leg matching the copper leads of the instrument side, which keeps thermal conduction errors small at cryogenic setups. It also has the tightest tolerance of the base-metal types near room temperature (about half a degree, class one), which is why laboratories, food cold-chains and freezers standardize on T. K works below zero on paper (-180 continuous) but its output flattens and its tolerance loosens toward the cold end; E and N are hot-side specialists. For deep cryogenics (liquid nitrogen and below) thermocouples generally give way to platinum RTDs and diode sensors, but for anything from -180 upward, T is the honest first pick.
Can I connect two thermocouple types in series or swap probes between meters?
Series - yes, deliberately: stacking two identical junctions doubles the output (a thermopile), and a known reference junction in series can compensate another type, but this is instrument design, not field practice. Swapping probes between meters of a different type - the common accident - is a silent calibration error: the meter applies one metal pair's lookup curve and cold-junction coefficients to another pair's millivolts, and since K and J outputs differ by 10-20 percent across their band, readings land tens of degrees off with no alarm. The defenses are mechanical discipline (color-coded connectors matched to type under IEC 60584-3) and a one-line bench check: immerse the probe in ice water and expect the standard reading for that type near zero. If the numbers drift after any rewiring, suspect the extension wire type before the probe itself - extension cable is mislabeled far more often than the sensor is wrong.