4.3 Five-Point Calibration, Adjustment and Records
Key Takeaways
Five points sample the quarters and midpoint as well as the endpoints. The work instruction determines the directions and settling conditions.
Compare rising and falling readings at the same input to check hysteresis. Compare repeated runs in the same direction to assess repeatability.
Convert tolerance using its stated basis, then apply the decision rule with the uncertainty of the actual test system.
Preserve as-found data before changes. Use comparable current as-found and prior as-left results to estimate change between tests.
Recheck both endpoints and the prescribed intermediate points after adjustment. Range assignment, sensor trim and analog output trim serve different purposes.
Five-Point Calibration, Adjustment and Records
A bench pressure test applies a known physical input and compares the instrument output with the expected response. The comparison establishes the measured error under the test conditions. An adjustment changes the response. Preserve the initial comparison before making that change so the two records answer different questions.
Choose the points and approach
A three-point check at 0%, 50%, and 100% samples the middle and both ends. A five-point check at 0%, 25%, 50%, 75%, and 100% also samples both quarters. This can expose deviations missed by testing the endpoints alone. The work instruction specifies the points, direction, conditioning, stabilization, and permissible input range. The Fluke 753/754 offers rising and falling test strategies, including a five-point rising-only test.
For a directional test, hold each input until stable. Record the reference input and the corresponding output. Starting at the lower endpoint, step up to the upper endpoint, then return down through the specified points. The upper endpoint can serve as the shared turnaround in the following example. Respect the instrument and reference limits throughout.
Input (% of configured span)
100 * turn and descend
75 o o
50 o o
25 o o
0 o o → time
rising falling
If an ascending setpoint is overshot in a test that requires a consistent approach, let the reference return to the approved restart point below the target, then approach in the rising direction. Some procedures call for repeating the series. Reading as pressure falls to the target gives a different approach history. Matching rise and fall observations at the same input reveals direction-dependent response. Comparing repeated observations taken with the same direction and conditions tests repeatability.
Convert the tolerance and evaluate the points
For a linear 4–20 mA output, compute the ideal current from the reference input and the configured range. The output span is 16 mA. Define signed error consistently as measured minus ideal:
Check the units and basis of the applicable tolerance. When it is specified as ±0.25% of output span, the permissible magnitude is . A tolerance based on reading, sensor range limit, or a composite specification needs its own calculation. Select a reference system suitable for the task and apply the stated decision rule. Pressure reference uncertainty, the current readout, and relevant setup effects all contribute to the uncertainty of this comparison. A guard band, if specified, changes the acceptance boundary. See ILAC G8 on decision rules.
Worked DP transmitter record
The range is 0–100 inH₂O with a linear 4–20 mA output. The example procedure allows ±0.50% of output span and uses simple acceptance at the tolerance limit, giving an error band of ±0.080 mA. The upper-end reading is shared at the turnaround.
| Applied DP (inH₂O) | Expected (mA) | Rising (mA) | Rising error (mA) | Falling (mA) | Falling error (mA) |
|---|---|---|---|---|---|
| 0 | 4.000 | 4.015 | +0.015 | 4.022 | +0.022 |
| 25 | 8.000 | 8.035 | +0.035 | 8.048 | +0.048 |
| 50 | 12.000 | 12.052 | +0.052 | 12.071 | +0.071 |
| 75 | 16.000 | 16.040 | +0.040 | 16.065 | +0.065 |
| 100 | 20.000 | 20.025 | +0.025 | 20.025 (turn) | +0.025 (shared) |
The largest absolute error is 0.071 mA and all observed points pass under the example rule. At 50 inH₂O the raw directional difference is , about 0.119% of output span. Among the interior points the largest directional difference is 0.025 mA at 75 inH₂O, about 0.156% of span. Apply a separate hysteresis limit or zero correction only if the procedure defines one. A rising-only check would have missed these directional differences.
Separate adjustment from verification
Consider a Fisher 846 I/P set to direct-acting full range. The endpoints are 4 mA input and 3 psig output, then 20 mA input and 15 psig output. The manufacturer's procedure calls for rechecking the endpoints. Where zero and span interact, as in its documented 12–20 mA split-range procedure, follow the specified sequence. On a linear characteristic the ideal offset adjustment shifts the line and the ideal span adjustment changes its slope. On real hardware a span change can also affect the low output. After adjustment verify both endpoints and the intermediate points required by the work instruction.
In a 4–20 mA smart transmitter, range configuration assigns the input endpoints to the output endpoints. Sensor trim adjusts the reported digital measurement against a known physical input. Analog output trim adjusts the current generator against a reference current measurement while the device commands fixed currents. In the earlier wet-leg example, the calculated −126 inH₂O at 0% level is the configured lower point. True sensor zero is equal pressure at the ports under the approved conditions. An absolute pressure sensor requires an accurate absolute reference for a low trim. Verify the complete response after adjustment or reconfiguration.
Record what the test demonstrated
- As-found captures the equipment condition before correction. Record the device ID, range and transfer function, reference IDs, test method, points, direction, readings, tolerance, decision rule and results.
- As-left captures the final verified condition. Record actions taken and repeat the required verification after any change. If the instrument meets the criteria without adjustment, record that disposition and follow the local procedure for reporting the unchanged result.
- Out of tolerance calls for disposition under the applicable procedure. Preserve the original results and notify the responsible personnel as required. Compare the current as-found error and previous as-left error at the same point and comparable conditions to estimate change over the interval. Consider service conditions, history and risk when reviewing test frequency.
For U.S. processes covered by OSHA 29 CFR 1910.119, paragraph (j)(4) requires documentation of each applicable inspection or test, including when it was done, by whom, the equipment identifier, the test description, and the result. Paragraph (j)(5) addresses equipment deficiencies outside acceptable limits. A transmitter's passing as-found result supports the assessment of that device at the tested conditions. Past loop behavior and safety interlock function require their own evidence.
A linear DP transmitter maps 0–200 inH₂O to 4–20 mA. The test allows ±0.25% of output span and compares observed error directly with that limit. At 100 inH₂O the measured output is 12.065 mA. What is the signed error and result for that point?
+0.025 mA, within the test limit
+0.065 mA, outside the ±0.040 mA limit
+0.065 mA, within the ±0.080 mA limit
−0.065 mA, outside the ±0.050 mA limit
A pressure test requires a rising approach to 12.0 psig and allows restarting from the previous point at 10.0 psig. The indicated reference reaches 12.4 psig by mistake. How should the technician obtain the rising reading?
Lower pressure to 12.0 psig and record the settled output.
Record at 12.4 psig and scale the measured error to 12.0 psig.
Record the reading on either approach as long as the reference matches 12.0 psig.
Return to 10.0 psig and approach 12.0 psig from below again.
At the same test point and comparable conditions, the previous as-left error was +0.010 mA and the current as-found error is +0.060 mA. What does the comparison establish?
The observed signed error at that point increased by 0.050 mA between tests.
The error increased at a constant rate throughout the interval.
The safety interlock performed as intended whenever demanded during the interval.
The current as-left error after adjustment must be −0.050 mA.
Sections you finish are checked off in the contents.