RTD vs thermistor: the choice your input card already made for you

The question always arrives the same way. Somebody has a water line to instrument, call it 10 to 30 °C, a handful of points, and a vendor has talked them into thermistors because they cost less and move a lot more ohms per degree in that exact band. On the sensor side, fair argument. Then comes the part that actually decides it: which Allen-Bradley card takes one?

So I went through the five RTD card manuals I keep coming back to and searched every one of them for the word thermistor. It is not there. Not in the 1746-NR8 manual, not the 1762-IR4, the 1769-IR6, the 1756-IR6I or the 5069-IY4. On Rockwell hardware, RTD vs thermistor gets settled at the rack, not at the sensor.

RTD vs thermistor: two materials, two curves

An RTD is a wire. Platinum usually, and pure, and its resistance climbs with temperature at a rate somebody wrote down as a constant. For the common curve that constant is 0.00385 ohms per ohm per degree C, which is where the 385 in Platinum 385 comes from. Rockwell prints it in the 1746-NR8 specifications in publication 1746-UM003. The curve bends slightly. Not much.

A thermistor is a ceramic. Sintered metal oxide, and its resistance falls off a cliff as it warms. Vishay's NTC application note 29053 gives the working form: R = R25 x exp[B(1/T - 1/298.15)]. That exponent is the whole story on both sides of the ledger. Far more ohms per degree than platinum will ever hand you, across a much narrower band, delivered as a curve that nothing in the rack straightens out for free.

And I still hear an RTD called a kind of thermistor. It is not. Different material, opposite sign on the coefficient, different math downstream.

Same axes, opposite behavior. The RTD vs thermistor difference starts in the shape of the curve, and everything downstream follows from it.

The number that ends most of these arguments

Every RTD vs thermistor argument I have been in eventually arrives at resistance mode. All of these cards have one. You give it ohms, it hands ohms back, and you do the temperature math yourself in the controller. So look at where the mode stops.

Widest direct resistance range for each card, from publications 1746-UM003, 1762-UM003, 1769-UM005, 1756-UM009 and 5069-UM005.

A 10k thermistor is 10,000 ohms at 25 °C. That is what the 10k means. Put one on a 1769-IR6 and it sits more than three times above the top of the widest range that module has. The channel comes back over-range, which each channel detects and flags on its own. Nothing subtle about it. The card cannot see the sensor.

Go the other direction, since that is the version that almost works. Low-value parts do exist, down around 2 kilohms nominal, and 2200 ohms at room temperature does fit inside a 3000 ohm range. Then remember which way an NTC moves. It climbs as it cools, so the cold end of your span is where the headroom runs out, and the number printed on the sensor at 25 °C tells you nothing about that end.

Excitation current narrows it again. On a 1762-IR4 or a 1769-IR6 the 3000 ohm range is not permitted at 1.0 mA at all; on a 1746-NR8 it survives but shrinks to 0 to 1200 ohms. One milliamp into 3000 ohms is three volts across the sensor, and that is the direction the limit comes from.

Where RTD vs thermistor actually gets decided. The Pt100 span sits comfortably inside the window, and a 10k part is not even in the picture.

RTD vs thermistor accuracy, using numbers that came with a document

Both sides talk past each other here. Thermistor people quote resolution, RTD people quote accuracy, and those are not the same claim.

The cards publish accuracy for a Pt100 385, and it lands in nearly the same place across three generations of hardware. Plus or minus 0.5 °C on a 1769-IR6 or a 1762-IR4 after autocalibration at 25 °C ambient. Let the module's own temperature drift and it widens to 0.9 °C, a figure the 1769-IR6 publishes at 60 °C ambient and the 1762-IR4 publishes across 0 to 55 °C. On a 1746-NR8, 0.5 °C at 0.25 mA excitation and 0.7 °C at 1.0 mA. The 5069-IY4 states it as 0.10 percent absolute accuracy at 25 °C, unchanged whether the channel runs current, voltage, RTD or thermocouple. Resolution on the older cards is 0.1 °C, repeatability 0.2 °C.

None of that asks anything of you. Pick Platinum 385 in the configuration, land the wires, done.

A thermistor will beat those numbers on resolution inside a narrow window. What it will not give you is interchangeability, and that is the part that costs money two years later. Standard R25 tolerance on the Vishay parts runs 1 to 5 percent, and the B-value carries a tolerance of its own on top, so the sensitivity itself moves from part to part. Two 10k thermistors off the same reel are not the same sensor.

The RTD escapes that because somebody standardized it. Rockwell publishes an RTD standards table in both 1746-UM003 and 1762-UM003, naming what each supported element conforms to: IEC-751 1983 with Amendment 2 from 1995 for the platinum 385 types, DIN 43760 beside it, JIS C1604 for the 3916 curve, SAMA RC21-4-1966 for the 10 ohm copper. The 1746-NR8 manual hangs a caution on it telling you to use RTDs conforming to those standards or expect reduced accuracy. Nobody could print the equivalent table for thermistors, and that is the real reason a brand change moves the reading.

Three ways people get a thermistor in anyway

None of these are wrong. Each one just moves work off the card and into your column.

Resistance mode with the math in the controller. Select a range, read ohms, run the B-equation or a Steinhart-Hart fit in ladder or structured text. It works, for a low-value part inside its range. You also own that linearization permanently, including the day somebody swaps in a different brand and nobody refits the coefficients.

A divider or bridge into an ordinary analog input. Cheap in hardware, since the analog card is usually already in the rack with a spare channel. Your accuracy becomes the reference voltage plus the resistor tolerances plus the sensor's own spread, and none of those three arrive with a published temperature figure.

An external transmitter with a 4-20 mA output. This is the one that actually gets installed. The transmitter owns the curve, the loop does not care about lead resistance, and any analog input in the plant can read it.

Say you have twenty-four points to bring in and empty analog channels available. A transmitter per point looks expensive next to a thermistor and two resistors, until you price the alternative honestly: an RTD card, or a week of somebody's time writing and validating curve math that then gets maintained for a decade. That is where the RTD vs thermistor decision usually lands, and it usually lands on the RTD.

The three paths, and the only thing that really separates them: who ends up owning the curve.

When the thermistor is the right call

I am not arguing the thermistor is a bad sensor. The claim is narrower: it is a bad general-purpose process input on a Rockwell rack.

It earns its place where small changes matter more than absolute truth inside a fixed band, and on an over-temperature trip that only has to act at one threshold.

Most of the thermistors already sitting in your plant fall in that second bucket, and they live in OEM equipment, where the builder designed the controller around one specific part and the curve is already in firmware. None of them are wired to a PLC card. They are wired to the board that shipped with the machine.

Sometimes one spec settles it outright. Standard Vishay NTC lines are rated for -40 to +125 °C, automotive parts to +150 °C. A Platinum 385 element on a 1769-IR6 covers -200 to +850 °C. If your process crosses either end of the thermistor's window, RTD vs thermistor was never a decision you had.

Questions I get asked about this

Can I wire a thermistor straight to an Allen-Bradley RTD card?

No. Not a 1746-NR8, 1762-IR4, 1769-IR6, 1756-IR6I or 5069-IY4. None of those manuals lists a thermistor among the supported sensor types, and the direct resistance modes stop between 3000 and 4020 ohms depending on the card. A 10k part is nowhere near that. Your options are a transmitter, a bridge into an analog input, or the curve math in your code.

RTD vs thermistor: which one is actually more accurate?

Depends which word you mean. For absolute accuracy with nothing added, the RTD, and the figure is published: 0.5 °C for a Pt100 385 on a 1769-IR6 at 25 °C ambient. For resolution inside a narrow band, a calibrated thermistor can do better. It will not hold that across a sensor swap, though, because R25 tolerance and B-value tolerance both move.

RTD vs thermistor for a water line at 10 to 30 °C?

RTD, if it is going into a Rockwell rack. The temperature range is not what decides this one, the input card is. A Pt100 covers 10 to 30 °C with room to spare and lands on a card you can buy off a shelf today.

Is a Pt1000 a thermistor?

No. It is a platinum RTD with a 1000 ohm element instead of 100, on the same 385 curve, and the standards table in 1746-UM003 has it conforming to IEC-751 and DIN 43760 exactly like a Pt100. One catch: on a 1769-IR6 the 1000 ohm Platinum types are not allowed at 1.0 mA excitation, only at 0.5 mA.

Check the card before you order the sensor

IQElectro stocks the Allen-Bradley RTD input cards for every platform named here: 1746-NR8 for SLC 500, 1762-IR4 for MicroLogix expansion racks, 1769-IR6 for Compact I/O, 1756-IR6I for ControlLogix, and 5069-IY4 for Compact 5000. All tested, all with a two-year warranty. And if RTD vs thermistor is still open on your end, send the catalog number of the card already in the rack and we will tell you what it accepts before you spend anything on sensors.