An RTD is one of the few sensors you can genuinely check with an ohmmeter and a known temperature, because its resistance follows a published curve. A healthy Pt100 reads very close to 100.00 ohm at the ice point and climbs by a predictable amount for every degree above it. Bench-testing means confirming that the element sits on that curve, that its lead wires are matched, and that it responds when the temperature changes. Doing this on the bench, before the sensor ever goes back into a thermowell, tells you whether the problem is the element itself or the wiring and input card downstream of it.
Bench-Test an RTD in one line: To bench-test an RTD, measure the element resistance with a good meter and compare it against the IEC 60751 table for its type, using an ice bath to nail the 0 C reference where a standard Pt100 should read 100.00 ohm. Confirm the three lead wires read nearly equal resistance to each other, then verify the element responds smoothly when you warm it in a dry block. A reading that is wildly high, near zero, or open circuit points to an open, shorted, or drifted element rather than a bad transmitter.
The core of a bench test is comparing measured resistance to the standard table for the sensor. A platinum RTD to IEC 60751 has a defined resistance at every temperature, anchored at the ice point where a Pt100 reads exactly 100.00 ohm and a Pt1000 reads 1000.0 ohm. Above that point the resistance rises by roughly 0.385 ohm per degree Celsius for a Pt100, so a reading of about 138.5 ohm corresponds to roughly 100 C. You do not need to memorize the whole curve; you need a printed or on-screen copy of the table for your sensor type and a meter accurate enough that the comparison is meaningful.
The single most reliable reference you can create on a bench is an ice-point bath: a slush of finely crushed ice and clean water, stirred, with the sensor tip immersed deep enough that the stem conduction does not pull the reading up. A well-made ice bath sits within a small fraction of a degree of 0 C, which means a good Pt100 immersed in it should read within a few hundredths of an ohm of 100.00. If it reads 100.3 or 99.6 the element has drifted, and how far it sits off the table tells you roughly how many degrees of error the sensor carries. This is the check that separates a truly accurate sensor from one that is merely alive.
For a two-point sense of the curve, warm the element in a dry-block calibrator or a stirred hot bath to a known temperature and confirm the resistance climbs to the value the table predicts. Two points, one cold and one warm, catch both an offset error and a slope error. If the ice point is perfect but the warm point is off, the element may be contaminated or strained; if both points are shifted by the same amount, you are usually looking at a lead-wire or connection problem rather than the platinum itself, which is why the lead-wire check below matters.
A three-wire RTD carries the lead-wire compensation the input card relies on, and that compensation only works if the three leads are electrically matched. On the bench you check this by measuring resistance between the pairs of leads. Two of the wires are commoned to one side of the element, so those two should read nearly identical low resistance to each other, ideally within a fraction of an ohm, while the pair that spans the element reads the element resistance plus the lead. Matched leads let the transmitter cancel the wire resistance; mismatched leads inject an error the card cannot remove and shift every reading.
A dead or dying element announces itself clearly once you know what to look for. An open circuit, meaning the meter reads infinite resistance or over-range, is a broken element or a broken lead and usually shows up in the transmitter as an upscale or downscale burnout indication. A reading near zero ohm, or far below the table value, is a short, often moisture in the head or a pinched lead touching the sheath. A reading that is plausible but consistently off the table by a repeatable amount is a drifted element, common after long exposure to high temperature or vibration, and it is the failure that fools people because the sensor still reads a temperature, just the wrong one.
When the sensor checks out but the loop still reads wrong, the fault is downstream, and a decade resistance box lets you prove that separately. Disconnect the RTD and connect a decade box set to a known table value, for example 100.00 ohm to simulate 0 C or 138.5 ohm to simulate about 100 C, in place of the sensor. If the transmitter or input card now reads the right temperature, the electronics are fine and the original sensor was the problem. If the card still reads wrong with a perfect resistance injected, the fault lives in the card, its trim, or its wiring, and no amount of sensor swapping will fix it.
Bench-testing an RTD is not an isolated event; it is the ground truth you use to interpret what a monitoring system has been showing you. When a temperature point on a SCADA screen has been reading a few degrees high for weeks, or has started jumping erratically, the bench test is how you decide whether the trend reflects a real process change or a failing sensor. A drifted element found in the ice bath explains a slow, steady offset in the trend; an intermittent open or a moisture short explains the erratic spikes. The historical trend and the bench measurement together tell a story neither could tell alone.
This is also where cloud monitoring changes the economics of when you bench-test at all. A platform such as Merobix trends every temperature point continuously, so a sensor that begins to drift or twitch is visible as a change in the recorded data long before an operator notices at the panel. Rather than testing every RTD on a fixed calendar, you can prioritize the sensors whose trends have started to misbehave, pull those, and confirm on the bench what the data already suggested. The monitoring system flags the candidates; the ice bath and the table confirm the diagnosis.
After you return a tested or replaced element to service, the same trend closes the loop. A sensor that read 100.3 ohm at the ice point and was replaced should show its offset disappear in the recorded data, and a formerly erratic point should settle into a clean line. Watching the post-repair trend confirms the fix actually landed on the right sensor, which is the kind of verification that is easy to skip in the field and easy to see when every point is being logged. The bench proves the element; the SCADA history proves the repair held.
A Pt100 rises about 0.385 ohm per degree Celsius above its 100.00 ohm ice point, so at a typical room temperature near 20 to 25 C it should read roughly 107.8 to 109.7 ohm. The exact figure comes from the IEC 60751 table for the measured temperature. If your reading is far outside that band the element has drifted, shorted, or opened, and comparing it to the table at a known temperature tells you which.
Substitute a decade resistance box for the RTD and set it to a known table value such as 100.00 ohm for 0 C. If the transmitter or input card then reads the correct temperature, the electronics are healthy and the original sensor was the problem. If it still reads wrong with a perfect resistance injected, the fault is in the card, its calibration trim, or the wiring rather than the sensor.
The third wire lets the input card measure and subtract the lead-wire resistance so long cable runs do not add temperature error. That subtraction assumes the leads are equal, so if one lead reads noticeably higher than the others the compensation is incomplete and every reading is offset. On the bench you confirm the two commoned leads read nearly the same low resistance to each other before trusting the sensor.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.