Five-Point As-Found That Fails High Across the Top
A 4-20 mA differential-pressure transmitter is ranged 0-250 inH2O and used for a level application. The site calibration procedure allows a tolerance of plus or minus 0.5% of span. You perform a five-point as-found check with a certified pressure source and a calibrated mA meter, increasing input only, at 0, 62.5, 125, 187.5 and 250 inH2O. Results:
- 0 inH2O: 4.00 mA
- 62.5 inH2O: 8.09 mA
- 125 inH2O: 12.16 mA
- 187.5 inH2O: 16.24 mA
- 250 inH2O: 20.33 mA
The transmitter was last calibrated 14 months ago and the preventive maintenance interval is 12 months. The unit is mounted on a column 40 ft up with no fixed ladder, and access requires a lift permit.
1. Compute the error at each point as a percentage of span, state which points fail, and characterize the fault.
Read the worked answer
Span = 20 - 4 = 16 mA. Tolerance = 0.5% of 16 mA = 0.08 mA (= 1.25 inH2O).
| Input (inH2O) | Expected mA | As-found mA | Error (mA) | Error (% span) | Result |
|---|---|---|---|---|---|
| 0 | 4.00 | 4.00 | 0.00 | 0.00% | pass |
| 62.5 | 8.00 | 8.09 | +0.09 | +0.56% | fail |
| 125 | 12.00 | 12.16 | +0.16 | +1.00% | fail |
| 187.5 | 16.00 | 16.24 | +0.24 | +1.50% | fail |
| 250 | 20.00 | 20.33 | +0.33 | +2.06% | fail |
Characterization: zero is correct and the error grows linearly with input. The slope is 16.33/16 = 1.021, so this is a span (gain) error of about +2%, not a zero shift and not random nonlinearity. All four nonzero points exceed tolerance; 62.5 inH2O is over the limit at 0.09 mA against 0.08 mA allowed.
A gain error of this pattern usually traces to a drifting span trim, a sensor that has been overpressured, or an electronics drift on an aging unit.
2. What adjustment do you make, and what must you verify afterward? Include why the single-direction check performed here is not sufficient on its own.
Read the worked answer
Adjustment: follow the procedure - set the LRV/URV, perform the sensor trim at zero and full scale (or the full multi-point trim if the procedure calls for it), then verify.
Before blaming the instrument, sanity-check the test setup. A consistent 2% gain error is large enough to be real, but a reference with an expired certificate, a leaking test connection, or the wrong deadweight tester piston would produce a similar pattern. Confirm the pressure source and the mA meter are in calibration and that you are not measuring static-pressure or temperature effects as calibration error.
Verification after adjustment:
- Five-point check in both directions (up and down) to expose hysteresis and deadband.
- Repeatability check at midscale - approach the same point two or three times.
- Zero return after the run.
- Record as-found and as-left data on the calibration sheet.
Why the up-only check is not enough: a one-direction sweep cannot detect hysteresis, deadband, or repeatability problems. Passing that check alone does not prove the instrument stays inside plus or minus 0.5% across the full error band - it only proves the up-scale curve is close. Many procedures deliberately perform as-found in the direction that will reveal the worst case.
Also: the unit drifted 2% in 14 months against a 12-month PM interval, and it was apparently not serviced on schedule. Investigate why the PM was missed, consider shortening the interval or replacing the transmitter, and note that this work needed a lift permit - plan it with enough lead time that the PM is not deferred again.
3. After reinstallation at a true level of 50%, the local display shows 50.0% but the DCS shows 47.8%. List a fault-isolation sequence.
Read the worked answer
A 2.2% gap between the local display and the DCS means something between the transmitter terminals and the DCS tag is losing or misinterpreting signal. Work outward from the process:
- Confirm the true process value independently - a sight glass or a second gauge. If the real level is not 50%, the DCS may be right and the display may be wrong.
- At the transmitter, measure the loop current in series with a calibrated meter and compare to the local display. Agreement means the transmitter end is good; write the number down.
- At the DCS input card, measure the same current in series at the card terminals. Any difference between step 2 and step 3 points to loop resistance, a parallel device, a loose terminal, or a ground fault diverting current.
- Check the analog input configuration: range (0-250 inH2O), 4-20 mA, engineering units, and any damping or filter left enabled. A card scaled for a different range is a common cause of a clean, repeatable offset.
- Loop de-energized: check insulation resistance from each leg to ground and look for a second ground on the shield.
- Check terminations for tightness and corrosion, and look for a series diode or isolator in the circuit.
- Only after the current loop is proven to deliver 12.00 mA to the card do you consider a card calibration or a DCS scaling correction.
Note that 12.00 mA should read 50.0%. The DCS showing 47.8% corresponds to about 11.65 mA - a definite loss of signal, not a rounding issue.