5s
+1
Submission successful!
A thermocouple and a resistance thermometer do the same job by opposite means. One produces a voltage from a temperature difference between two dissimilar metals; the other changes its own resistance as it warms. That single difference cascades into everything a buyer compares: the temperature range, the achievable accuracy, the way the instrument is wired, and whether the reading has to be corrected for the temperature of the terminals.
Both families are standardised, and the standards are where the numbers on a datasheet come from. For platinum resistance thermometers that standard is IEC 60751, which covers platinum resistors and thermometers over −200 °C to +850 °C. For thermocouples it is the letter designation system, with IEC 60584-1 and ASTM E230 behind it.
| Resistance thermometer (RTD) | Thermocouple | |
|---|---|---|
| What changes | The electrical resistance of a metal element rises as it warms | A small voltage appears across a junction of two dissimilar metals, and grows with the temperature difference between that junction and the terminals |
| The physics | Resistance against temperature | The Seebeck effect, which is why a thermocouple is a voltage source and not a resistor |
| The element | Platinum in most industrial work, because the resistance against temperature relationship is close to linear and the metal does not degrade readily. Nickel and copper elements exist and are less stable over time | Two dissimilar metals, chosen as a pair. Base metal pairs are built on nickel, iron and copper; noble metal pairs on platinum and rhodium |
| Standardised by | IEC 60751, which fixes the resistance against temperature curve and the tolerance classes | IEC 60584-1 and ASTM E230, which fix the voltage tables and the limits of error |
The usable range is the first thing that separates them, and it is wider apart than most datasheets suggest.
| Resistance thermometer | Thermocouple | |
|---|---|---|
| Covered by the standard | −200 °C to +850 °C | Not one range: each letter type carries its own, from cryogenic work to above 1,700 °C |
| In ordinary industrial use | About −50 °C to 600 °C, depending on the sheath material and the element chosen | Up to roughly 1,260 °C for the common base metal types, and past 1,700 °C for the noble metal ones |
| At the cold end | Works well. The standard curve runs below zero, and cryogenic classes exist | Works, but the output at the cold end is small. One noble metal type reads the same at 0 °C and 42 °C, so it cannot be used there |
| Why the ceiling exists | The element and its insulation, and the accuracy of the resistance measurement at the top of the curve | The metals, the atmosphere and the conductor gauge, and not the pair alone |
The published practical figure for the resistance thermometer is the one worth repeating: between about −50 °C and 600 °C in normal service, set by the sheath and the element, not by the standard curve. A claim that a platinum resistance thermometer covers only up to 150 °C describes a particular assembly, not the technology.
The resistance thermometer is the more accurate instrument and the more stable one, and the reason is structural. A thermocouple is a voltage source measured against a reference, so its error budget includes the reference junction. A resistance thermometer is compared against a known current, and its error is largely a matter of the element and the wiring.
| Resistance thermometer | Thermocouple | |
|---|---|---|
| Typical accuracy | The tighter of the two. Tolerance classes run AA, A, B and C, and the class applies to the complete thermometer, not only to the element | For base metal types, commonly 0.75 % of reading or ±1.0 °C, whichever is greater |
| Stability over time | Drifts little, which is why platinum resistance elements are used as references | Drifts with contamination and with time at temperature, and is rechecked more often |
| Linearity | Close to linear over its range, so a simple relationship holds | Non-linear, which is why the instrument applies a polynomial instead of a straight line |
| Where the error hides | In the wiring, and in the current used to read the element | In the reference junction, the extension or compensating cable, and the metals |
One clause in the standard is worth knowing before a specification is written. Since the 2008 revision, the classification applies to the complete thermometer and is verified at the connection end, and not on the resistor on its own. A supplier quoting a class for an element is not quoting it for an assembly.
A resistance thermometer has to be read by passing a current through it, so the resistance of the leads is in series with the element and appears as temperature. Three connection schemes exist to deal with that, and the choice is not a preference.
| Connection | How it works | Where it belongs |
|---|---|---|
| Two wire | One lead to each end of the element. The lead resistance is measured along with the element | Where accuracy is not the point |
| Three wire | Two leads to one side of the element and one to the other, which cancels the lead resistance on the compensated side | The usual industrial arrangement, and the norm for long runs |
| Four wire | Two leads to each side, so the current is carried by one pair and the voltage sensed by the other | The most accurate, and the arrangement the standard expects where accuracy better than Class B is claimed |
Two published figures explain why the connection scheme matters so much. A compensation error of one ohm in the leads produces about 1 % of error on a Pt100 and about 0.1 % on a Pt1000, because the same absolute error is a ten times smaller share of a ten times larger element resistance. And to meet Class A or better, the standard expects a four-wire connection, because the permitted tolerance on the leads themselves makes a three-wire Class A claim impossible to verify on long cable.
The thermocouple has the opposite problem. It does not care about lead resistance, because the instrument is reading a voltage and not driving a current through the sensor, but it does care what the leads are made of. Any joint between unlike metals adds its own voltage, so the run from the head to the instrument has to be the matching extension or compensating cable, and the polarity has to be right at every joint.
| Resistance thermometer | Thermocouple | |
|---|---|---|
| What the leads affect | The reading, through their own resistance | The reading, through the voltages generated at every junction |
| What to do about it | Choose two, three or four wire, and match the element value to the run | Match the cable to the type, and keep the polarity correct |
| Reference junction | Not applicable, because the sensor is not a voltage source | Required. The terminal temperature has to be measured and converted into an equivalent voltage before the reading is valid |
| Resistance thermometer | Thermocouple | |
|---|---|---|
| Its characteristic difficulty | Self-heating, because a current has to be passed through the element to read it. The current warms the element slightly, so the reading sits above the true temperature | The reference junction, because the output depends on the difference between the measuring junction and the terminals |
| What makes it worse | A high excitation current, a small element, and poor thermal contact between the element and the sheath | A temperature difference between the two terminals, and a cable of the wrong type in the run |
| What the buyer does about it | Accept the manufacturer's stated excitation current, and give the element good thermal contact with the sheath | Hold the terminals at one temperature on an isothermal block, and use the matching cable |
| The sign of getting it wrong | The reading is consistently high and moves with the excitation current | The reading moves with the ambient temperature around the instrument, not with the process |
| Requirement | Which family, and why |
|---|---|
| Above about 600 °C | Thermocouple. The resistance element and its insulation are the limit, and the thermocouple ceiling is set by the metals instead |
| The tightest accuracy available | Resistance thermometer, on a four-wire connection, in a class matched to the need, not to the budget |
| Fast response to a change | Thermocouple. A thin film element is quicker than a wire-wound one, and both are slower than an exposed thermocouple junction |
| Heavy vibration or repeated thermal shock | Thermocouple. The element in a resistance thermometer is the part that suffers |
| Long cable runs on a budget | Thermocouple, where the extension or compensating cable is cheaper than the copper that a three or four wire element needs |
| Cryogenic work | Either, with a check on the class: the highest resistance thermometer classes do not cover the lowest temperatures |
| Where self-heating cannot be tolerated | Thermocouple, since it generates its own signal and needs no excitation current |
| A reference or calibration duty | Resistance thermometer, for the stability |
The resistance thermometer, and the gap is structural and not a matter of quality. Its error is dominated by the element and the wiring, while the thermocouple also has to measure the temperature of its own terminals and correct for it. The published comparison puts a base metal thermocouple at about 0.75 % of reading or ±1.0 °C, whichever is greater, against a platinum element that reaches a small fraction of that when the wiring is done properly.
The thermocouple, by a wide margin. IEC 60751 covers platinum resistance thermometers from −200 °C to +850 °C, while thermocouples reach past 1,700 °C with the noble metal types. Common base metal types reach roughly 1,260 °C. That is why furnace and flue work is normally a thermocouple duty.
How many conductors reach the element and therefore how much of the lead resistance is cancelled. Two wire carries the lead resistance into the reading. Three wire cancels it on one side and is the usual industrial answer. Four wire separates the current pair from the sensing pair and removes the lead resistance on both sides. The standard expects four wire where a class better than B is claimed.
The scheme can be the same, but the same lead error costs far more on the Pt100. One ohm of lead resistance is about 1 % of error on a Pt100 and about 0.1 % on a Pt1000, because the element resistance is ten times larger. On a long run, the larger element value buys back accuracy that would otherwise have to be paid for in copper.
Only the thermocouple. Its output is the voltage produced by the difference between the measuring junction and the terminals, so the terminal temperature has to be measured and converted to an equivalent voltage before the reading means anything. A resistance thermometer is not a voltage source, so it has no reference junction and no such correction.
The base metal thermocouple is cheaper to buy, and the resistance thermometer often costs less over the life of the installation because it drifts less and is rechecked less often. Two other costs decide the comparison in practice: the compensation cable a thermocouple needs for the run to the instrument, and the copper a three or four wire element needs for the same run.
The difference is one of principle, and everything else follows. A thermocouple makes a voltage from a temperature difference and is therefore wide-ranging, fast and rugged, and also dependent on a reference junction and a matched cable. A resistance thermometer changes its own resistance and is therefore accurate, linear and stable, and also dependent on how many wires reach the element and on the current used to read it. IEC 60751 puts the standard platinum range at −200 °C to +850 °C; the thermocouple goes past 1,700 °C.
Nickel and nickel alloy products have been made at Changzhou, Jiangsu since 2002, in a factory of 12,000 m² that holds ISO 9001 certification and is audited by SGS. Thermocouple wire and mineral insulated cable are supplied across the letter types: send the sensor type, the range, the environment and the cable run for a confirmed quotation.
The figures quoted are published values for standard products and are not acceptance limits. What a particular assembly achieves depends on the element or junction, the sheath, the wiring and the instrument together, and the classification is verified on the complete thermometer, not on the sensing element alone.
english
français
Deutsch
Italiano
Русский
Español
português
Nederlandse
ελληνικά
日本語
한국
中文
العربية
हिन्दी
Türkçe
indonesia
tiếng Việt
ไทย
বাংলা
فارسی
polski