Process Temperature Monitoring: PT100 and PT1000 in Practice

Last updated: 3 September 2026

In short: Process temperature accuracy is decided by the probe class and the wiring, not by the transmitter’s resolution. A PT100 changes about 0.385 ohms per degree, so one ohm of lead resistance is a 2.6°C error, roughly six times a Class B probe’s own tolerance at 0°C. Three-wire connection exists specifically to cancel that.

A sensor datasheet that promises 0.1°C resolution is telling you the size of the steps it reports in. It is not telling you how close those steps are to the truth. Those are different numbers, and only one of them appears in most quotes.

This guide covers the three things that actually set process temperature accuracy on an industrial retrofit: which probe you fit, how it is wired, and where you put it.

Resolution is not accuracy

Start here, because it is the most common misreading of a process temperature datasheet. The Milesight EM500-PT100 datasheet quotes a resolution of 0.1°C and a three-wire connection, with range options of -200 to 50°C, -50 to 200°C, -50 to 500°C and -50 to 800°C.

What it does not quote is an accuracy figure in degrees, and that omission is correct rather than evasive. The transmitter measures resistance. How faithfully that resistance corresponds to a real process temperature is a property of the platinum element you attach to it.

So the specification that governs your process temperature accuracy is not on the transmitter datasheet at all. It is the probe’s tolerance class.

What the IEC 60751 tolerance classes actually give you

The process temperature standard is IEC 60751. A PT100 is defined by a resistance of 100.00 ohms at 0°C and a temperature coefficient of 0.00385 between 0 and 100°C, with tolerance bands set by class. Those bands are formulas, not fixed numbers, because error grows with distance from zero.

According to the published IEC 60751 tolerances, the three classes are Class AA at ±(0.1 + 0.0017|t|), Class A at ±(0.15 + 0.002|t|) and Class B at ±(0.3 + 0.005|t|), where t is the temperature in degrees Celsius.

Put numbers through those formulas and the practical difference appears. At 0°C the three classes give ±0.1, ±0.15 and ±0.3°C. At 200°C they give ±0.44, ±0.55 and ±1.3°C.

Class B is the default you get if you do not specify, and for a great deal of process temperature work it is entirely adequate. If you are proving a pasteurisation hold or a heat treatment soak, it is not, and the upgrade to Class A costs far less than the argument about whether the batch was compliant.

Why lead resistance ruins two-wire installations

This is the process temperature error that dwarfs everything else on a real site, and it has nothing to do with the probe’s quality.

The instrument measures resistance. In a two-wire connection it cannot distinguish the resistance of the platinum element from the resistance of the cable running to it, so the cable is silently added to the reading.

Work out the size of it. A PT100 changes by roughly 0.385 ohms per degree Celsius. Ten metres of 0.5 mm² copper, out and back, is about 0.68 ohms, which the instrument reads as an extra 1.8°C that is not there.

Against a Class B probe’s ±0.3°C tolerance at 0°C, a 1.8°C wiring error is not a refinement. It is six times larger than the thing you paid for, and it is always in the same direction, so it never averages out.

A three-wire connection solves it. The third conductor lets the instrument measure the lead resistance separately and subtract it, which is exactly why the EM500-PT100 specifies three-wire connection rather than two. Four-wire does the same job more completely and belongs in laboratory work.

PT100 or PT1000 for process temperature?

For process temperature work the difference is the base resistance, 100 ohms against 1000 ohms at 0°C, and the consequence is sensitivity to exactly the problem above.

A PT1000 changes about 3.85 ohms per degree, ten times more than a PT100. The same 0.68 ohms of cable therefore produces roughly 0.18°C of error instead of 1.8°C.

• PT1000 is more tolerant of long or thin cable runs and of two-wire installations where three-wire is genuinely impossible. It suits building services and distributed process temperature points.
• PT100 remains the industrial default, has the widest range of available probes and thermowells, and is what most existing plant and paperwork already assumes.
• Compatibility decides it more often than physics. A transmitter built for PT100 will not read a PT1000 correctly, so match the probe to the input rather than to a preference.

Thermowells, immersion depth and response time

You are measuring the probe, not the process temperature itself. Everything that sits between the two is an error source, and installation causes more bad process temperature data than probe selection ever does.

Immersion depth is the classic process temperature failure. A probe that barely enters the pipe reads a blend of process and ambient, conducted along its own stem. The usual guidance is to immerse to at least ten times the probe diameter, and more where the pipe is small or the ambient is extreme.

Thermowells protect the probe and let you replace it without breaking containment, which is essential on anything pressurised or hazardous. They also slow the response down, because heat now has to cross an air gap and a metal wall.

If response time matters, use a thermal compound or a spring-loaded probe to eliminate the air gap. If it does not, accept the lag and set your alarm delays to match, rather than chasing what looks like a slow-responding sensor.

Place the probe in flowing product, not in a dead leg. A stagnant pocket reads its own local temperature perfectly and tells you nothing about the process.

Orientation matters on horizontal pipework too. A probe entering from the top of a line can sit in an air pocket if the pipe ever runs part full, and a probe entering from the bottom collects sediment that insulates the tip over time. Entry from the side, angled slightly into the flow, avoids both and is the position most process temperature installations should default to.

Finally, label the tapping. A probe nobody can identify at the panel end is a probe nobody recalibrates, and an unlabelled point is the first thing to be disbelieved when two readings disagree.

A worked process temperature error budget

Errors add up, and the useful discipline is to write them down rather than trusting the largest number on the datasheet. Take a realistic case: a Class B PT100 in a thermowell on a 90°C hot water flow, wired 15 m back to a transmitter.

Probe tolerance. Class B at 90°C is ±(0.3 + 0.005 × 90), which is ±0.75°C. This is the only figure most people account for.

Wiring. Three-wire connection cancels the bulk of the lead resistance, leaving a small residual from mismatch between the conductors. Call it ±0.1°C. Had this been two-wire, 15 m of 0.5 mm² would have added roughly 2.7°C of one-directional error instead.

Installation. A thermowell with an air gap and marginal immersion depth pulls the reading toward ambient. On a hot line in a cool plant room that is a consistent under-read, and half a degree is optimistic.

Total. Roughly ±1.4°C once you are honest about all three, against a transmitter advertising 0.1°C resolution. The resolution was never the constraint.

The value of the exercise is that it shows you where to spend. Upgrading to a Class A probe here buys you 0.3°C. Fixing the immersion depth and the air gap buys you more, and costs nothing but doing it properly. That is the usual answer on a process temperature retrofit, and it is why installation quality beats component selection almost every time.

Battery life and reporting interval

A wireless process temperature point removes the cable, which is usually the expensive part of the installation, but it introduces a reporting interval you have to choose deliberately.

The EM500-PT100 quotes 10 years of battery life at a ten minute interval at 25°C, from a single 19000 mAh lithium thionyl chloride cell, with an operating range of -30 to 70°C on the transceiver and IP67 protection.

Ten minutes is the right interval for a tank, a store or a curing oven. It is the wrong interval for anything with a fast excursion you need to catch, because a ten minute gap can hide the entire event.

Use the threshold and change alarm features rather than simply polling faster. Reporting on a deviation catches the excursion without paying for it in battery on the 99% of readings where nothing happened. The device also stores 1,000 entries locally with retransmission, so a gateway outage does not put a hole in the record.

Frequently asked questions

What accuracy can I expect from a PT100 in practice?

Take the probe’s class formula, then add the process temperature wiring error and any installation error. A Class B probe at 100°C is ±0.8°C on paper, and a poorly immersed two-wire installation can easily double that.

Is two-wire ever acceptable for process temperature?

With a PT1000 over a short run, yes. With a PT100 over anything more than a metre or two, the lead resistance error becomes larger than the probe tolerance, so use three-wire.

Thermocouple or RTD for process temperature?

RTD below roughly 500 to 600°C, where it is more accurate, more stable and more repeatable. Thermocouples above that, and where you need a very fast response or a very small probe.

Do process temperature probes drift?

Slowly, and mostly from thermal cycling, vibration and contamination rather than age alone. Anything feeding a compliance record needs a documented recalibration interval regardless of what the datasheet promises.

Can one transmitter cover the whole range?

The EM500-PT100 is ordered against a range option, from -200 to 50°C up to -50 to 800°C, and is customisable between -200 and 800°C. Specify the range you need at order time rather than assuming one part covers everything.

The short version

Specify the probe class against what you actually have to prove, insist on three-wire for any PT100, immerse the probe properly, and choose a reporting interval that matches how fast the process can change. Do those four things and the transmitter’s resolution stops being the interesting number.

For related measurement work, our guides to cold chain temperature monitoring and vibration monitoring cover the same specify-then-install discipline. If the probe is feeding a control system, the Modbus gateway guide covers getting it there.

The Milesight EM500-PT industrial temperature sensor is priced and in stock. Add it to a quote and we will come back within one working day.