Engineering
Calibration certificates and traceability: what they actually prove
Eventually someone who didn't take a reading asks where it came from. What answers them is the chain behind the sensor, which is decided before the sensor is chosen.
Computing America Hardware. Custom hardware for physical automation and monitoring.
Send the measurement
In short
- A bare number on a dashboard is not yet a measurement result, because a result is incomplete without a statement of its uncertainty.
- NIST states plainly that having an instrument calibrated at NIST is not enough to make a later reading traceable. Traceability belongs to the result.
- Ask any certificate for the accreditation scope, the expanded uncertainty with its coverage factor, and the as-found as well as the as-left results.
- If you tell anyone a number is traceable, you must be able to support the claim, and the person relying on it decides whether your evidence is enough.
- As-found values, the recalibration interval and which unit was in service exist at one moment, and if they weren't recorded then they can't be recovered.
A calibration certificate shows that an instrument was compared against a reference, on one day, under stated conditions, with a stated uncertainty and a traceability claim. It doesn't show that the number your system stored is traceable, and that matters the day someone outside the firm asks where it came from. Say a robotic cell measures parts and rejects those out of tolerance, and a customer disputes a rejected lot. With a chain behind the sensor the question can be answered from records. Without one it can't. Until that day the two systems look identical: both produce plausible numbers, both trend smoothly, and routine site tests don't include a challenge.
A number is not yet a measurement¶
The international guide metrology runs on is unambiguous. The GUM states that the result of a measurement is incomplete without a statement of the uncertainty (opens in a new tab). It splits that uncertainty into two kinds. Type A comes from statistical analysis of repeated observations. Type B comes from everything else: the calibration certificate, the manufacturer's specification and informed judgment about the installation.
So the uncertainty of an installed instrument is a budget you assemble, and you can't look it up. Much of it is Type B: the reference, the calibration conditions, how far the process differs from them, the time since, and what the installation contributes. A flow meter with a stated accuracy of half a percent, installed closer to an elbow than its manual allows, isn't a half-percent instrument in that pipe. Nothing on the screen says so.
Traceability belongs to the result¶
NIST defines metrological traceability as a property of a measurement result (opens in a new tab), one whereby the result can be related to a reference through a documented unbroken chain of calibrations, each contributing to the measurement uncertainty. NIST adds that merely having an instrument or artifact calibrated at NIST is not enough to make the user's measurement result traceable.
The number your system stored
Installation, process and drift add uncertainty here
Within: Yours to document.Your instrument
One certificate: this link, this day
Within: Covered by certificates.Accredited laboratory's reference
Its own certificate and stated uncertainty
Within: Covered by certificates.National metrology institute
Realizes the unit
Within: Covered by certificates.The SI unit
The reference every link has to reach
A certificate documents one link in the chain, and the number on your screen sits above the last link, connected only by your own records.
SourceChain structure from NIST's traceability policy and ILAC P10, both cited in the sources.
NIST's policy puts responsibility where it falls. Providing support for a claim of traceability "is the responsibility of the provider of that result", and assessing the validity of the claim is the responsibility of the user. When your system publishes a number somebody outside your organization relies on, and you say or imply it is traceable, you're the provider of that claim.
What to demand from a certificate¶
- Accreditation and scope. ILAC P10 (opens in a new tab) accepts an accredited laboratory "whose service is suitable for the intended use (i.e., the scope of accreditation specifically covers the appropriate calibration)". Check that the scope covers your range and quantity.
- Uncertainty. ILAC P14 (opens in a new tab) requires an accredited laboratory's calibration result to include "the associated expanded uncertainty U", with the coverage factor and coverage probability stated. Compare U against what your process needs.
- As-found and as-left results. Ask for the readings before any adjustment as well as after it. A certificate with only the as-left half hides how far the instrument had drifted.
What the chain has to record¶
| Record | What it has to hold | How it usually breaks |
|---|---|---|
| Reference | What the instrument was compared against, with that reference's own uncertainty and traceability | Calibrated against a second working instrument nobody can chain further. |
| Event | Date, conditions, as-found and as-left values, who performed it | Only as-left is kept. |
| Interval | When it's due again, and why | The vendor's default, never revisited against observed drift: wasting money or expiring silently. |
| Installation | Where the device sits and what its manual asks for | A commissioning photograph and nothing machine-readable, so the installation's contribution is in nobody's budget. |
| Identity | Which physical unit produced this record, surviving replacement | A device swapped and its tag reused, so the new unit inherits the old one's calibration history. |
Drift is a rate, and the interval is a decision¶
Sensing elements foul and age, and electronics shift with temperature. None of it raises an alarm, because an instrument reading eight percent low behaves exactly like one reading correctly.
A recalibration interval is a decision about how much undetected error the process will carry, and the as-found values are its evidence. ILAC-G24 (opens in a new tab) describes five methods for reviewing one. The simplest is its "staircase" method. At each routine calibration, "the subsequent recalibration interval is extended (or kept unchanged) if the deviation from the reference value is found to be within an appropriately defined percentage of the range between the maximum permissible errors", and reduced otherwise. You define that percentage for the instrument.
The software has a part too. If the historian carries the conversion and the applicable interval beside each value, a corrected constant can be applied to history. If the conversion lives only in code, the past keeps the wrong constant forever.
Where this stops, and what to do first¶
Not every number needs this. A tank level that starts a pump needs to be roughly right and reliably available, and a metrology program around it spends money on the wrong risk. The test is whether the number may have to be defended to someone who didn't take it, and whether being wrong costs more than the chain costs to keep. A complete chain also doesn't make a number true. A sensor sited where it can't see the phenomenon gives a perfectly traceable measurement of the wrong thing.
Before choosing the sensor, settle four things.
- Who might question this number, and what would they ask for?
- What uncertainty must it carry, and which instruments can meet that installed, not on the data sheet?
- How will as-found values, the interval and the unit's identity be recorded?
- Where does the conversion live, so a corrected constant reaches history?
Sources
- 1.JCGM 100:2008: Evaluation of measurement data — Guide to the expression of uncertainty in measurement (GUM) (opens in a new tab), JCGM / BIPM
- 2.Metrological Traceability: Frequently Asked Questions and NIST Policy (opens in a new tab), NIST
- 3.ILAC P10:07/2020: ILAC Policy on Metrological Traceability of Measurement Results (opens in a new tab), ILAC
- 4.ILAC P14:09/2020: ILAC Policy for Measurement Uncertainty in Calibration (opens in a new tab), ILAC
- 5.ILAC-G24:2022 / OIML D 10:2022: Guidelines for the determination of recalibration intervals of measuring equipment (opens in a new tab), ILAC / OIML
Next step
Send us the number you have to defend.
Tell us which reading somebody outside your firm relies on and which instrument produces it. We will tell you what the chain behind it would have to contain to survive being questioned, and which link you are most likely missing.
- Phone
- (214) 723-2510
- Reply
- A person replies, not a sequence: within one business day, from someone who would be on the engagement.
Packages
Start with a package
Four packages, each with its price on the page. Pick one and buy it online.
$199/month
Billed monthly
$182.42/month
Billed yearly
$499/month
Billed monthly
$457.42/month
Billed yearly
- Scammers stopped from sending email as you
- Your website and systems scanned every week
- Your email settings checked every week
- Everything in the 1 to 10 plan
- Covers up to 50 people
$199/month
after a one-time setup fee
- A site planned, designed and built for you
- Hosting and updates every month
$499/month
Billed monthly
$457.42/month
Billed yearly
$999/month
Billed monthly
$915.75/month
Billed yearly
- Your Google profile, fully set up
- A Google post every week
- 2 social posts a week
- Everything in Get found
- 4 social posts a week
- Your top 10 Google searches tracked
from$500/month
after a one-time setup fee
- Your data gathered from the tools you use
- A first analysis that answers your questions
- New findings every month
Related reading
Engineering5 min readReference frames in sensor data: relative to what?
What to record beside every reading so two runs can still be compared a year later.
Delivery4 min readHardware pilot projects: scope one to buy a decision, not a demo
Scope a pilot backward from the go or no-go it must produce. A demo that works under supervision settles nothing about the rollout.
Engineering3 min readIP65 vs NEMA ratings: what an ingress rating does not test
IP65 records that a sample passed a dust test and a water-jet test, and a NEMA Type is not its equivalent. What your cleaning regime does to the box is yours to specify.
