PT100 vs thermocouple: choosing the right temperature sensor for autoclave cycle monitoring
Both sensors measure temperature. Both are used in industry every day. But steam sterilization does not ask for a sensor that covers a huge range — it asks for a sensor that is trustworthy inside a very narrow band, cycle after cycle, for years. That single requirement decides the argument.
Two completely different physics
A PT100 is a resistance thermometer. It is a platinum element whose electrical resistance rises with temperature in a way that is stable, documented and reproducible. You send a small current through it, you measure the voltage across it, you obtain a resistance, and you convert that resistance into a temperature using the curve of IEC 60751. Nothing else is required for the reading to be meaningful.
A thermocouple works on another principle entirely. Two dissimilar metals joined at one end produce a small voltage that depends on the temperature difference between that junction and the point where the wires are connected to the instrument. That is the whole trick, and it is also the whole problem: the measurement is relative, not absolute. To turn it into a temperature you must know, at every instant, the temperature of the terminal block itself. That is cold junction compensation, and it is one more thing that can quietly be wrong.
The signals are also of a very different size. A PT100 changes by roughly 0.385 ohm per degree near zero — a large, easy signal. A type K thermocouple produces around 41 microvolts per degree. Microvolt-level signals travelling along a cable in a technical room, next to contactors and motor drives, are far more exposed to electrical noise and to the quality of every connection along the way.
Accuracy where it matters: 121 and 134 °C
Steam sterilization lives at two plateaus. Everything that matters happens there, so the honest comparison is not which sensor covers the widest range, but which sensor is tightest at 121 and 134 degrees.
For a PT100, the tolerance classes of IEC 60751 are defined as a formula, so they can be calculated exactly at the temperature of interest:
| PT100 class | Tolerance at 121 °C | Tolerance at 134 °C |
|---|---|---|
| Class AA | ±0.31 °C | ±0.33 °C |
| Class A | ±0.39 °C | ±0.42 °C |
| Class B | ±0.91 °C | ±0.97 °C |
For thermocouples, the tolerance classes of IEC 60584 are expressed as a fixed value or a percentage, whichever is greater. In this band the fixed value dominates:
| Thermocouple | Class 1 tolerance at 134 °C |
|---|---|
| Type K | ±1.5 °C |
| Type N | ±1.5 °C |
| Type T | ±0.5 °C |
The gap is not subtle. A class AA PT100 is roughly five times tighter than a class 1 type K at the sterilization plateau. Type T narrows the gap and is genuinely good at low and moderate temperatures, but it still carries every structural weakness described below. When the whole point of the record is to show that a plateau was held, a tolerance wider than one degree eats into the very margin you are trying to demonstrate.
Wiring: where a PT100 can go wrong, and how it is fixed
A PT100 has one classic weakness: in two-wire mode, the resistance of the cable is added to the resistance of the element, and the instrument cannot tell them apart. Roughly one ohm of loop resistance shifts the reading by nearly 2.7 degrees. On a long run that error is not a rounding detail, it is a false record.
The remedy is standard and cheap: four-wire measurement. Two wires carry the excitation current, two wires sense the voltage directly across the element, and the cable drops out of the equation entirely. Cable length stops being a variable. Once you have done that, an ordinary shielded four-core cable is all you need, and the shield is grounded at one end only, at the cabinet.
A thermocouple has no equivalent escape route. Its weakness is structural rather than a wiring option you can select: the reference is the terminal block, and everything that happens to that terminal block happens to the reading. A cabinet whose internal temperature rises during the day moves the reference with it, shifting every value recorded.
Cabling: the cost nobody quotes
A thermocouple cannot be extended with ordinary copper. Any junction with a different metal creates a new thermocouple and a new error, so the run to the instrument must use matching extension wire of the same type, with correct polarity throughout. That wire costs more than copper, must be kept as a matched pair, and every terminal block along the way is a potential error. Get the polarity wrong and the fault does not announce itself — the reading stays plausible and drifts in the wrong direction.
A PT100 uses ordinary copper conductors, and with a four-wire connection the cable resistance drops out of the measurement entirely. On a retrofit where the converter sits a few metres from the probe, this is a practical simplification, not a detail. It also means a technician can source the cable locally instead of ordering a matched pair.
The difference that decides it: drift
A cycle record is only worth what it is worth in three years. Here the difference is about ageing, not initial accuracy.
Platinum is chemically stable and the element is not being consumed by the measurement. A well-installed PT100 in a protected thermowell drifts slowly and predictably, and when it is re-verified, it tends to be close to where it was.
Thermocouple wire ages differently. The alloys oxidise, and the very property that produces the voltage — the composition along the wire — changes with thermal cycling and with contamination. That is why thermocouples are usually treated as consumables in demanding applications. In a service context, replacing a sensor periodically because it drifts is a poor answer to give a customer who is relying on the record.
Repeatability matters as much as absolute accuracy here. A sensor that reads consistently, cycle after cycle, lets a technician see a slow change in the shape of the plateau. That signal is only visible if the noise floor of the measurement is well below the change you are looking for.
Where the thermocouple genuinely wins
None of this makes thermocouples inferior in general. They are the right choice in several situations that a PT100 simply cannot serve:
- High temperature. Depyrogenation tunnels, furnaces and burners run far beyond the useful range of a platinum element. Types N, R and S exist for exactly that.
- Very small thermal mass and very fast transients. A fine junction responds faster than a sheathed resistance element, which matters when you are chasing a spike rather than holding a plateau.
- Physically tiny or disposable probes. A thermocouple can be made smaller and cheaper, which is why it appears in load-penetration studies with many single-use points.
- Rugged, high-vibration environments where a wire-wound element would be at risk.
Permanent cycle recording on an installed sterilizer is none of those. It is a narrow band, a slow plateau, a fixed installation and a record that must stay credible for years. That is resistance thermometer territory.
Side by side
| PT100 | Thermocouple | |
|---|---|---|
| Principle | Resistance changes with temperature | Junction generates a voltage |
| Accuracy at 134 °C | ±0.33 °C (class AA) | ±0.5 °C (type T, class 1) |
| Long-term stability | Very good | Drifts more over time |
| Cold junction compensation | Not needed | Required |
| Extension cable | Ordinary copper | Matching type, polarity critical |
| Response time | Slower, especially in a thermowell | Very fast when the junction is fine |
| Typical use here | Permanent cycle recording | Validation studies inside the chamber |
What this means for a connected sterilizer
The chain used on a connected sterilizer follows directly from the comparison above: a four-wire class AA PT100 in a stainless steel thermowell, held in external contact on the drain line with thermal paste and a stainless clamp, feeding a PT100-to-RS-485 converter that publishes the value over Modbus RTU, then a gateway to the network. No hole is drilled, no fluid is entered, and nothing is connected to the electronics of the machine. Where the probe sits on that line changes the reading, which is why probe placement deserves its own treatment.
Two consequences are worth stating plainly. First, this sensor measures temperature and nothing else. Pressure appears in the record only when the sterilizer itself exposes it on a digital output that the gateway can read. Second, a temperature curve is evidence, not a verdict. It shows what happened during a cycle; interpreting it and releasing a load remain the decision of qualified personnel.
For the underlying physics in more detail — the resistance curve, self-heating, and the two, three and four wire arrangements — see how a PT100 probe works. For the wider picture of adding cycle records to a sterilizer you already run, see converting an existing autoclave.
The short answer
Use a thermocouple when the temperature is too high for platinum, when the probe must be tiny or disposable, or when you need to catch a fast transient inside a chamber during a validation study. Use a PT100 when a stable, absolute, repeatable number in a narrow band has to stand up years later. Permanent autoclave cycle monitoring is the second case, and the choice is not close.
Frequently asked questions
Which is more accurate at 134 °C?
Class AA PT100 at about ±0.33 °C, against ±0.5 °C for a class 1 type T and ±1.5 °C for a class 1 type K.
Do thermocouples drift?
More than platinum elements. For a permanent archive, that is the deciding factor.
When should I use a thermocouple?
For fine, fast sensors inside the chamber or a load during a validation study.
Can I extend a thermocouple with copper wire?
No. Matching extension wire of the same type is required, with correct polarity.
Does either measure pressure?
No. Both measure temperature only.
Talk to our team about your installation
IOT.AUTOCLAVEXPRESS.COM is a division of SCMDSD LLC (USA). The platform records and documents sterilization cycles. It does not determine the conformity of a cycle and does not authorise the release of a sterilized load; that decision remains with the qualified personnel responsible for it. All technical resources