Pressure gauge calibration: how to calibrate a pressure gauge (deadweight tester or comparator)

Pressure gauge calibration compares what a gauge reads with a more accurate, traceable pressure reference at several points across its range, first with the pressure rising and then with it falling, and records the deviation at each point. The reference is either a deadweight tester (known weights on a piston of known area) or a comparator: a hand pump or screw press feeding the gauge and a calibrated digital reference gauge at the same time. For a typical dial gauge, the free guideline DKD-R 6-1 asks for one preload to full scale, then five points including zero, read up and down, with the gauge tapped before each reading and a five-minute wait at the top (DKD-R 6-1, §7 and Table 1). Checking that a gauge reads zero, or comparing it against a better gauge, is a useful check, but not a traceable calibration.

This page covers both: what you can check yourself, and the full procedure a lab or in-house calibration room follows, with a worked example of a gauge that passes on the way up and fails on the way down.

Can you check a pressure gauge yourself?

You can check it, and on many dial gauges you can reset the pointer. Whether the result is a calibration depends on what you compare against and what you record:

Two checks are worth doing between calibrations:

1. Zero. With the gauge isolated and vented, a gauge-pressure gauge should read zero. That proves only that one point is right. An absolute gauge (a barometer, for example) does not read zero at atmospheric pressure and cannot be checked this way (DKD-R 6-1, §8.6.2).
2. Comparison with a better gauge. Connect the gauge and a more accurate gauge to the two legs of a tee and apply pressure slowly through the third (the same arrangement Dwyer describes for its Magnehelic gauges). Compare at a few points up and down. The better gauge should be several times more accurate than the one you check: about four times is the long-standing rule of thumb (Mensor; Fluke), and Ashcroft attributes it to ASME guidance.

Accuracy grades and classes: what “1 %” means

The tolerance of a dial gauge is set by its accuracy grade (US, ASME B40.100) or accuracy class (Europe, EN 837-1), and in both systems the percentage is of the span, not of the reading. On a 0–160 psi gauge, ±1 % means ±1.6 psi everywhere on the dial, so at 20 psi the gauge may be off by 8 % of the reading.

ASME B40.100 (current edition 2022) is a paid standard and we have not read it; the grades below are as published, consistently, by two manufacturers and a distributor (Ashcroft; WIKA data sheets; Boshart, which cites the predecessor standard, ASME B40.1):

Lower grades also exist; check your gauge’s data sheet. EN 837-1 expresses accuracy classes the same way, as a percentage of span: a class 1.0 gauge is good to ±1 % of span, a class 0.6 gauge to ±0.6 %, and precision test gauges reach class 0.1 (WIKA data sheets; DKD-R 6-1, App. B, E and F). EN 837-1 is also a paid standard: we give only the classes we could confirm in two sources. Use the grade or class marked on the dial or on the data sheet, and write down which standard it comes from.

The grade drives everything that follows: how good the reference must be, how many points and series you need, and the pass or fail limits.

What you need to calibrate a pressure gauge

• Reference standard

  A deadweight tester (pressure balance) or a reference gauge with a calibration certificate stating its expanded uncertainty

  Pressure balances and liquid manometers are the long-term stable references; electronic reference gauges are less stable and need shorter intervals (DKD-R 6-1, §3 and App. F). Handheld pressure calibrators (Fluke, Druck, Additel and others) are electronic reference gauges.

  DKD-R 6-1, §3; EURAMET cg-17, §5.3.1

• Accuracy ratio

  Reference uncertainty no more than one third of the uncertainty you aim for (DKD); a ratio of at least 2 (EURAMET); about 4:1 against the gauge tolerance is the long-standing rule of thumb

  For a ±1.6 psi gauge, a reference good to ±0.4 psi or better. Check yours with the TUR calculator.

  DKD-R 6-1, §3; cg-17, §5.3.1; Mensor and Fluke (4:1 rule of thumb)

• Pressure source

  A screw press or hand pump with fine adjustment, a vent valve and, for gas, a clean dry gas supply

  Hydraulic work uses a screw press and a priming pump; vacuum needs a pump with traps and isolating valves.

  EURAMET cg-17, §5.3.2

• Adapters and seals

  Clean fittings and seals for each thread, compatible with the medium

  Never put a gauge marked USE NO OIL (oxygen service) on an oil-filled tester. Liquid from glycerin-filled gauges can be drawn into the reference when you vent.

  Ashcroft (oxygen cleaning); AIGA SP 09/20; Fluke (contamination)

• Environment

  18 to 28 °C, stable within ±1 K during the calibration, recorded. With a deadweight tester, also record barometric pressure (and humidity, or estimate it)

  Let the gauge reach room temperature first; electronic gauges also need their warm-up time.

  DKD-R 6-1, §6; cg-17, §5.1

• Paperwork

  Data sheet (range, grade or class, mounting position), your procedure and a record form

  See how to write a calibration procedure.

Deadweight tester or comparator?

There are two ways to put a known pressure on the gauge. The method is the same comparison either way: the reading of the gauge against the pressure of the reference (DKD-R 6-1, §3).

A US government example of the two working together: the Bureau of Reclamation calibrated a master gauge against a deadweight tester at five or more increments, then calibrated working gauges against the master (USBR 1040-89, §3.1 and §9.2).

Corrections for a deadweight tester

Local gravity. The pressure is proportional to g, and industrial testers are often made for standard gravity, 9.806 65 m/s². Labs that calibrate pressure balances usually report results for standard gravity and say how to convert them to local gravity (cg-3, §6.2). Fluke’s example: in Phoenix, Arizona, local gravity is 9.794 74 m/s², more than 1200 ppm (about 0.12 %) below standard; by our arithmetic, a weight stamped 100 psi for standard gravity generates about 99.88 psi there. Fluke calls local gravity the most significant single source of uncertainty for a deadweight tester. To correct, multiply the nominal pressure by local g ÷ the g the tester was made for, or order the tester for your local gravity. Get local g from a survey, from NOAA’s surface gravity prediction, or, as a fallback, from the latitude formula in 40 CFR 1065.630, which ignores altitude and local geology (at 45° it gives 9.806 199 m/s²).

Temperature and air buoyancy. The effective area grows with piston temperature, and the air displaces a small part of the weights’ mass; both appear in the pressure equation (cg-3, eq. 3.1). That is why the piston temperature, room temperature, barometric pressure and humidity are recorded (DKD-R 6-1, §6). DKD notes that weather-driven changes in air density are normally negligible against the uncertainty of the piston area (DKD-R 6-1, App. A, note 3).

Head correction: keep the reference levels at the same height

Both guides ask you to minimize the height difference between the reference levels of the standard and the gauge, and to correct for it if it remains (DKD-R 6-1, §7; cg-17, §5.1). The column of medium between the two levels adds or removes pressure:

With oil, one inch of height difference is about 0.03 psi. That is negligible on a 0–160 psi grade 1A gauge (±1.6 psi), but on a 0–15 psi grade 4A test gauge (±0.015 psi) a 6-inch difference would cause an error more than ten times the tolerance. With gas, the effect is much smaller but grows with pressure, because gas density does.

Test points, series and readings

The number of points and series depends on how accurate the gauge is. DKD-R 6-1 defines three calibration sequences by the uncertainty you aim for, expressed as a percentage of span (DKD-R 6-1, Table 1):

For Bourdon tube gauges the wait at the top is 5 minutes in every sequence, and each pressure step takes at least 30 seconds before the reading (DKD-R 6-1, Table 1, notes, and §7).

EURAMET cg-17 has a similar ladder: a Basic procedure for a target uncertainty of 0.2 % of full scale or more (6 points, one cycle up and down, repeatability from two points measured three times), a Standard procedure (11 points) and a Comprehensive procedure (11 points, three cycles) for the most accurate instruments (cg-17, §5.2). Gauges used on both sides of zero need at least two points in the negative range (DKD-R 6-1, Table 1, notes).

For the example gauge, a 0–160 psi grade 1A dial gauge, sequence C applies: five points, one preload, one series up and one down.

Why up and down: hysteresis

A Bourdon tube and its movement do not follow exactly the same path when pressure falls as when it rises. The difference between the descending and ascending readings at each point is the hysteresis, and it is part of the result and of the uncertainty (DKD-R 6-1, §8.6.5; cg-17, §6.3). Here is the example gauge as found:

Read upscale only, this gauge passes: the worst ascending error is +1.4 psi at full scale. On the way down it reads +1.8 psi at 80 psi and +2.0 psi at 120 psi, outside ±1.6 psi. The hysteresis is 0.8 psi at mid-scale, half the tolerance on its own.

Mounting position

Calibrate the gauge in the position it is used in. DKD-R 6-1 requires the mounting position to be considered and stated on the certificate (§7 and §9), and cg-17 asks you to follow the manufacturer’s specification for mounting position (§5.1). Dwyer, for example, calibrates its Magnehelic gauges with the diaphragm vertical and says that for maximum accuracy they must be calibrated in the position in which they are used. If a panel gauge is mounted horizontally, calibrate it horizontally.

Step-by-step pressure gauge calibration procedure

The sequence follows DKD-R 6-1 (§5 to §7) and EURAMET cg-17 (§5.4), plus the as-found and as-left logic of any calibration procedure. Adapt it to your gauge and write your choices into your calibration procedure.

1. 1

   Identify and inspect

   Record ID, range, units, grade or class and mounting position. Look for damage to the pointer, threads, sealing surface and pressure channel, contamination, and a readable dial (DKD-R 6-1, §5).

   Damaged or bent pointer: record as found, tag out

2. 2

   Condition

   Let the gauge and the reference reach room temperature (18 to 28 °C, stable within ±1 K) and record it. Switch electronic gauges on for their warm-up time.

3. 3

   Mount

   Fit the gauge in its working position, with the reference levels of gauge and standard at the same height, or measure the difference for a head correction. Level a deadweight tester. Use clean, medium-compatible fittings.

4. 4

   Leak check

   Pressurize to the upper limit and isolate: the pressure must stay stable (cg-17, §5.4.1). Fix leaks before going on.

5. 5

   Preload

   Take the gauge to full scale and hold it, as many times as the sequence requires: at least 30 s each (DKD) or at least twice for 1 min (cg-17). Do not record.

6. 6

   Zero

   Vent and, after the gauge settles, set zero if the gauge allows it, then read zero immediately (DKD-R 6-1, §7).

7. 7

   As-found, ascending

   Approach each point from below without overshooting. Wait at least 30 s after the pressure change, tap a dial gauge lightly, read at eye level and record both the gauge and the reference.

8. 8

   Top of the range

   Read full scale, wait (2 min, or 5 min for Bourdon tube gauges), read again (DKD-R 6-1, §7).

9. 9

   As-found, descending

   Come down through the same points from above, again without overshooting, then read zero at least 30 s after complete venting.

   Any reading outside tolerance: record as found out of tolerance and open an out-of-tolerance report

10. 10

    Adjust only if needed

    Zero (pointer) or span (movement) per the manufacturer, and only with the owner's agreement; record what was changed (cg-17, §5.4.2.2).

11. 11

    As-left

    After any adjustment, repeat the whole sequence and record it as a separate set. If nothing was adjusted, the as-found results are also the as-left results.

12. 12

    Evaluate, record and label

    Calculate deviations, hysteresis and uncertainty, decide pass or fail with your decision rule, issue the certificate or record and label the gauge.

Calibration and adjustment are different operations: calibration compares the gauge with the reference, adjustment changes the gauge, and after an adjustment it should usually be calibrated again (VIM 2.39 Note 2; 3.11 Note 3). cg-17 builds that into its three steps: an initial check of the condition before any adjustment, the adjustment, then the full calibration (cg-17, §5.4.2).

Reading a dial gauge. Tap lightly before reading to release friction in the movement (DKD-R 6-1, §7). Read square to the dial: viewing at an angle shifts the apparent reading (parallax), which is why test gauges have mirror bands and why Fluke recommends always reading from the same position (Fluke, calibration tips). Estimate to a fraction of a division: DKD recommends 1/2, 1/5 or 1/10, with 1/10 only if the divisions are at least 2.5 mm apart (DKD-R 6-1, §8.6.1.1). For a digital gauge, the resolution is the last digit, provided the reading is stable within one digit when unloaded (§8.6.1.2).

Safety. Vent before connecting or disconnecting, wear safety glasses when applying pressure (USBR 1040-89, §6.1), and keep media separate: oxygen gauges are cleaned of hydrocarbons and marked USE NO OIL, because oil residue can ignite in oxygen (Ashcroft; AIGA SP 09/20).

Acceptance criteria and measurement uncertainty

Worked example: a dial gauge that fails on the way down

The example gauge, PG-204, is a 4.5-in Bourdon tube gauge, 0–160 psi, 1 psi divisions, ASME B40.100 Grade 1A. Its tolerance is ±1.6 psi at every point. It was calibrated against a digital reference gauge with an expanded uncertainty of 0.1 psi (k = 2), a test uncertainty ratio of 16:1. Each point shows two readings: up and down. The values are invented for this example.

• As found: two of ten readings are outside ±1.6 psi, both descending (80 and 120 psi). The gauge was out of tolerance: on falling pressure it read up to 2 psi high. Anything decided from its readings since its last good calibration needs an impact assessment.
• As left: after a span adjustment, the largest error is +0.8 psi (descending at 80 and 120 psi). The hysteresis is still there, 0.6 to 0.8 psi between 40 and 120 psi: adjusting the pointer or the span shifts the readings, it does not remove the friction and elastic lag behind the gap between the two series. A gauge whose hysteresis alone eats most of its tolerance should be repaired or replaced.

Uncertainty and error span

DKD-R 6-1 and cg-17 use the same model for an indicating gauge: deviation = indication − reference, with uncertainty contributions from the reference, the resolution, the zero deviation, the repeatability (when there are repeated series) and, for the mean of the up and down readings, the hysteresis (DKD-R 6-1, §8.3; cg-17, §6.2.1). For the example gauge, as left, at 120 psi:

Two rules from DKD-R 6-1 then apply. First, with sequence C (one series, so no repeatability), the uncertainty stated may not be smaller than 0.30 % of span: 0.48 psi here, which replaces the 0.44 psi calculated (§9.3). Second, the error span adds the expanded uncertainty and the size of the deviation; it is the largest difference to expect between a reading and the true pressure, and it supports a statement of conformity (§9.1.2, §9.1.3). At 120 psi the mean deviation is +0.5 psi, so the error span is 0.48 + 0.5 = 0.98 psi, inside ±1.6 psi. (For a conformity statement, DKD also sets a floor on the error span with sequence C, 0.60 % of span, 0.96 psi here; 0.98 psi is above it.) The worked examples in cg-17 (Example 1a) and DKD-R 6-1 (Appendix B) follow the same steps; our calculations reproduce both (see the tests of src/lib/pressure-gauge.ts).

Whether a reading close to the limit counts as a pass depends on the uncertainty and on your decision rule (Global ACI-TECH-1-002, formerly ILAC G8). The guard banding guide explains the options, and the uncertainty calculator builds the budget.

Low-pressure and differential gauges (Magnehelic)

Differential gauges reading inches of water, such as Dwyer’s Magnehelic, follow the same comparison with a few specifics from the manufacturer’s documents:

• Accuracy: ±2 % of full scale for most ranges, ±3 % or ±4 % for the lowest ones (Dwyer, Series 2000 data sheet). On a 0–1 inH₂O gauge, ±2 % is ±0.02 inH₂O, so the reference has to resolve and be accurate to a small fraction of that.
• Position: calibrated with the diaphragm vertical. Many ranges above 1 inH₂O can be used at other angles after re-zeroing, but for maximum accuracy calibrate in the position of use; the lowest ranges must be used vertical only (Dwyer).
• Zero: only with both the high and low pressure ports open to the atmosphere (Dwyer instructions).
• Calibration check: connect the high side of the gauge and a gauge or manometer of known accuracy to two legs of a tee with short tubing, and apply pressure very slowly through the third leg. If the accuracy is unacceptable, Dwyer recommends returning the gauge to the factory; its instructions also describe a field adjustment, but say recalibration by the user is not recommended (Dwyer instructions).
• Units: one inch of water (conventional) is 249.0889 Pa, about 0.036 psi (NIST SP 811, App. B.8). At these pressures a tiny leak changes the reading, so keep the tubing short and the connections tight.

Backflow test kits are also differential gauges; their calibration requirements are set by the local water authority or certification program, so follow those.

What to record on a pressure gauge calibration certificate

Whether you issue a certificate or fill in an internal record, keep:

• Identification: gauge ID or serial number, manufacturer, model, range, units, grade or class.
• Method: your procedure, or “DKD-R 6-1 sequence C” or “EURAMET cg-17 Basic procedure” if you follow them (DKD-R 6-1, §9).
• Set-up: reference standard and its traceability, pressure medium, mounting position, pressure reference level and any head correction, settings selected on the gauge (DKD-R 6-1, §9).
• Conditions: temperature; with a deadweight tester, also barometric pressure, humidity and the local gravity used (DKD-R 6-1, §6; cg-17, §5.4.2.3; cg-3, §6.2).
• Results: at every point, the reference pressure and the gauge reading, up and down, as found and as left, with deviation and hysteresis (DKD-R 6-1, §9, Table 8).
• Uncertainty: expanded uncertainty (or error span) per point.
• Decision: pass or fail against the stated grade or class, the decision rule, and any adjustment made.
• Who and when: technician, date, next due date.

Our guides to calibration certificates and calibration records cover each field. Print a matching sticker with the free calibration label generator.

How often to calibrate a pressure gauge

DKD-R 6-1 leaves the interval to the user and recommends, under normal conditions of use (App. F):

Many US programs start at 12 months: the Bureau of Reclamation calibrated its laboratory gauges on receipt and annually after that (USBR 1040-89, §4.2), and Ashcroft calls annual checks a rule of thumb, with the user setting the actual schedule. Whatever the period, recalibrate after an overload beyond the gauge’s limit, a repair, or rough handling that may have affected it (DKD-R 6-1, App. F), including a pressure spike (water hammer, for example) that may have exceeded it. Then let your as-found history move the interval: see calibration intervals.

When a pressure gauge fails calibration

An as-found result outside tolerance means more than “adjust the gauge”. In the example, the gauge read up to 2 psi high on falling pressure: a process held at a pressure limit, a relief setting checked against it, or a test pressure recorded from it may have been wrong since its last good calibration. Our guide to out-of-tolerance results walks through the impact assessment.

Common mistakes when calibrating a pressure gauge

Related guides

• How to write a calibration procedure: the general structure this procedure follows.
• Pressure gauge calibration hub: everything on pressure gauges in one place.
• TUR calculator: check that your reference is good enough for the gauge.
• Uncertainty calculator: build the uncertainty budget.
• Out of tolerance: what to do when a calibration fails.
• Calibration intervals: how often to recalibrate.
• What is calibration: calibration versus adjustment.
• Calibration label generator: print the calibration sticker.