KEYSTONEIndustrial Services
Interactive trainer

HART transmitter calibration

A smart transmitter shows one measurement three ways — the applied input, the digital HART PV, and the analog 4-20 mA — and calibration means making all three agree. A sensor trim aligns the PV to the input; a D/A trim aligns the mA to the PV; ranging sets how PV maps to 4-20. Six transmitters, four views, including the two look-alike pairs you tell apart only by checking 0, 50 and 100 %.

By Francisco Padilla Interactive Technician level

Before you start — job brief

A smart transmitter shows the same measurement three ways, and a good calibration makes all three agree: the input the calibrator applies, the transmitter's digital HART primary variable (PV), and the analog 4–20 mA output. A sensor (input) trim aligns the digital PV to the applied reference; a D/A (output) trim aligns the mA to the PV; re-ranging (LRV/URV) sets how the PV maps onto 4–20. This device is a pressure transmitter ranged 0–100 inH2O = 4–20 mA, so a correct output is mA = 4 + 0.16 × percent.

  • Apply a known input; the digital PV should equal it and the mA should equal 4 + 0.16 × percent
  • PV disagrees with the applied input → a sensor-trim or transfer-function problem (the input link)
  • mA disagrees with the PV → a D/A-trim or ranging problem (the output link)
  • Always check at 0 / 50 / 100 % to tell an offset from a slope from a curve

A smart transmitter shows three views that must agree

A HART transmitter is not one measurement but three views of the same thing, and calibrating it means making all three agree. First is the applied input: the real pressure, temperature or level the calibrator puts on the sensor — the ground truth. Second is the digital PV: the transmitter's primary variable, read straight off the HART signal with a communicator or a master, in engineering units or percent. Third is the analog 4–20 mA: the loop current the control system actually sees. On the pressure transmitter in this trainer, ranged 0–100 inH2O = 4–20 mA, a correct output is mA = 4 + 0.16 × percent: 4 mA at 0 %, 12 mA at 50 %, 20 mA at 100 %. Apply a known input, read all three, and the transmitter tells you which one is lying.

Three adjustments, three different jobs

The three views are joined by two links, and each link has its own adjustment. The sensor (input) trim aligns the digital PV to the applied reference — you apply a known low and high input and trim the PV to match. The D/A (output) trim aligns the analog mA to the PV — you command the transmitter's 4 and 20 mA points and trim until a traceable meter reads exactly 4.00 and 20.00. And re-ranging (setting the LRV and URV) decides how the PV maps onto 4–20 — it changes the output scaling without touching either trim. Three tools, three jobs. The single most common calibration mistake is confusing them: adjusting the sensor when the mA is wrong, or re-trimming the output when the PV never matched reality in the first place.

Field note — which link is broken?

One question sorts every calibration fault in this trainer. Does the digital PV agree with the applied input? If it does not, the broken link is on the input side — a sensor/zero trim that has drifted, or a wrong transfer function — and no amount of output trimming will fix it. If the PV does agree with the input but the mA disagrees with the PV, the broken link is on the output side — a bad D/A trim or a range mismatch — and the sensor is innocent. Confuse the two and you "calibrate" the wrong thing: you trim a healthy sensor to hide an output error, or you redo an output trim to chase a sensor drift, and the transmitter comes back wrong the next time anyone checks it against a real input.

Offset, slope, or curve — check at 0, 50 and 100 %

When a link is broken, how it is broken tells you which adjustment to reach for, and you read that from the shape of the error across the range. A fixed offset is the same number wrong everywhere — a drifted zero on the sensor, or a D/A trim off by a constant few hundredths of a milliamp; it is there at 0 %, at 50 %, and at 100 % alike. A scaling (span) error grows: it is zero at the bottom of the range and largest at the top, the signature of a range or gain mismatch. A nonlinear error is right at both ends and worst in the middle — the fingerprint of a wrong transfer function, like a square-root curve left on a plain pressure measurement. That is why you never judge a calibration from a single midscale point: only by checking 0, 50 and 100 % can you tell an offset from a slope from a curve.

The two look-alikes

Two pairs of faults look identical at a single test point, and telling them apart is the whole discipline. When the PV is correct but the mA is wrong, a D/A-trim error and a range mismatch present the same at midscale — but the trim error is a constant offset at every point, while the range error grows toward the top of the span. Check 0 and 100 %: same milliamps off everywhere means redo the D/A trim; a gap that grows means re-range the analog. When the PV is wrong, a sensor-trim drift and a wrong transfer function look alike — but the sensor drift is a linear offset present even near the ends, while the √ curve is exact at 0 and 100 % and worst at midscale. Same symptom, opposite fix; the distinguishing reading is always at the ends versus the middle.

The discipline

Calibrating a smart transmitter is a fixed routine, not a feel. Apply a known input from a traceable calibrator. Compare the applied value, the digital PV and the analog mA at 0, 50 and 100 %. If the PV disagrees with the input, do a sensor trim or fix the transfer function; if the mA disagrees with the PV, do a D/A trim or re-range; if everything agrees, sign it off and resist the urge to adjust. Fix the broken link — and only the broken link — and every view of the measurement agrees again.