Every nondestructive testing (NDT) result rests on an assumption nobody writes on the report: that the equipment was working. The UV-A lamp was bright enough, the yoke could still lift its weight, the flaw detector's screen was linear, the densitometer read true. Most of the time the assumption holds. Calibration tracking exists for the day it does not, and for the auditor who asks you to prove it held on every other day.
This guide covers what needs calibrating in a typical NDT shop, the daily and periodic process checks that sit beside calibration, how custody and send-outs fit in, and the question that decides whether a calibration program is any good: when an instrument fails, which jobs did it touch?
What needs calibrating
The short answer is anything that produces a number you rely on to accept or reject a part, or to show that the process was in control. In practice that list is longer than most shops expect:
- Ultrasonic instruments: screen height and amplitude control linearity, plus the whole system's calibration against a reference block before each examination. Thickness gauges belong here too.
- Magnetic particle equipment: bench-unit ammeters, timers and quick-break circuits, yokes (by lifting a dead weight), and Hall-effect gaussmeters and field indicators.
- Light meters: UV-A radiometers and visible-light photometers. These are the measuring instruments behind the most frequent checks in penetrant and magnetic particle testing, so a radiometer out of calibration quietly invalidates every lamp check made with it.
- Radiography: densitometers, with the step tablets and step wedge films they are checked against.
- Process equipment: drying-oven controllers and indicators, water pressure and temperature gauges, air pressure gauges on a penetrant line.
- Reference standards: ultrasonic calibration blocks, test pieces with known discontinuities, and in-house masters that other gauges are set against. These are not “calibrated” in the same way as an instrument, but they carry a certificate of material and dimensions, they wear, and they belong in the register.
Traceability is the thread through all of it. ASTM E1417-21 (7.8.4) expects equipment calibration to be traceable to NIST or another recognized national standard, where applicable. E1444-26 (7.1) requires a calibration system to ANSI/NCSL Z540.3, ISO 10012 or ISO/IEC 17025 for the meters, shunts, timers and radiometers used in verification, and E1417-21 (7.8.4.4) names the same three for its radiometers and photometers. A certificate that does not state its reference standard and its traceability is not much of a certificate.
The interval depends on the document
There is no single calibration interval for an NDT instrument. The interval comes from whichever document governs the work, and two documents can disagree about the same piece of equipment.
| Item | ASTM E1444-26 (MT, aerospace) | ASTM E1417-21 (PT) | ASME Section V (2025) |
|---|---|---|---|
| UV-A lamp intensity | Daily | Daily | Before use, after a power change, and at the end of the examination (T-676.4, T-777.2) |
| Yoke lifting check | Every 6 months | — | Each day the yoke is used (T-762) |
| Ammeter on magnetizing equipment | Every 6 months | — | At least yearly (T-761.1) |
| UV-A radiometer / light meter | Every 6 months | Every 6 months | At least yearly (T-761.2) |
| UT linearity (screen height, amplitude) | — | — | 3 months analog, 1 year digital (T-461) |
| Densitometer | — | — | At least every 3 months in use, with a check each shift (T-262) |
Our own selection from the editions listed at the end of this guide, paraphrased. It is not the full table in any of them: check every interval against your own copy and your procedure.
The yoke row is the one to notice. A shop doing aerospace magnetic particle work to E1444 and code work to ASME Section V can hold the same yoke to a six-monthly dead-weight check on one job and a daily one on the next. The AC lift figure is the same 10 lb; the DC figures and leg spacings differ, and the clock differs most of all. Your equipment record has to know which document each job runs under, or the stricter interval gets missed.
Some documents set no interval at all. ASTM E317-21, the practice for evaluating pulse-echo instrument performance without electronic test equipment, says in 5.7 that it neither recommends nor implies a frequency: that is for the user to decide, based on the application, the environment and the equipment's stability. It also says (1.6) that it is no substitute for calibrating the system for a given material. When a document leaves the interval open, your procedure or written practice must close it, and an auditor will ask where.
Intervals can move with evidence. E1444-26 lets the maximum time between most verifications be shortened or lengthened when real reliability data supports it, and E709-21 (20.2) describes six months as the usual verification interval for magnetic particle equipment while leaving the actual frequency to the facility's written procedure. Extending an interval is legitimate; doing it without the data on file is a finding.
Process checks: the daily layer
Calibration happens a few times a year. Process checks happen every day the line runs, and in penetrant and magnetic particle work they make up most of the record. E1417-21 Table 1 and E1444-26 Table 1 list them: penetrant contamination and system performance daily, UV-A lamp intensity and integrity daily, wet particle concentration every eight hours or shift change, visible and ambient light weekly or quarterly, and so on.
A few details trip shops up in audits:
- Days the line does not run. E1417-21 (7.8) bases its frequencies on a facility operating daily. A shop that runs less often may reduce the daily and weekly checks accordingly, but they still have to be done before parts are processed; other checks keep their full-time frequency.
- Battery-powered UV-A lamps. Both E1417-21 and E1444-26 ask for their intensity to be measured before and after use, and E1417-21 (6.7.4.1) wants the method and frequency for all UV-A lamps written into the procedure.
- Set points, not just ticks. A lamp check is a reading against a minimum irradiance at a stated distance. A yoke check is a lift of a stated weight at a stated leg spacing. Record the reading, not only “pass”, and the drift shows before the failure does.
- What does not need a record. E1417-21 marks several routine checks, and E1444-26 the daily UV-A lamp integrity check, as not needing to be recorded, and E1417-21 (7.8.4.8) says timers used for dwell times need not be calibrated at all. Knowing the exemptions stops a shop inventing paperwork it then fails to keep.
Ultrasonic work has its own daily layer. ASME Section V Article 4 (T-467) requires a calibration check at the end of each examination or series of similar examinations, when examination personnel change, and whenever any part of the system is changed. Those checks are the bridge between one calibration and the next.
Custody and send-outs
A calibration record says the instrument was good on a date. It does not say where the instrument went afterwards. Two more records close that gap.
Custody is who had the instrument, and when. In a field operation a flaw detector or a yoke moves between technicians and sites every week. If custody lives in a sign-out book, or in nobody's head, the question “who used this between March and May?” becomes an afternoon of reading timesheets.
Send-outs are the instrument's trips to a calibration vendor: which vendor, when it left, when it is due back, the cost, and the certificate that came back with it. While it is away it must be out of service, not quietly still on the shelf list. When it returns, the certificate's as-found readings matter as much as its as-left ones, because as-found is what tells you whether the instrument was already out before it went.
Reference chains belong here too. If working gauges are set against an in-house master, a failed master calls every gauge set against it into question, and the register should show that dependency rather than leave it to memory.
Out of tolerance: which jobs did it touch?
This is the part of a calibration program that matters most and gets the least attention. An instrument comes back from the vendor with an as-found reading outside tolerance, or fails a process check on the bench. Now what?
- Quarantine it. Tag it out of service so it is not signed out again while the question is open.
- Fix the window. The exposure runs from the last check or calibration the instrument passed to the day the failure was found. Anything inside that window is suspect.
- List the work. Every job the instrument went out on and every report written with it inside the window. ASTM E709-21 (20.2.1) calls it good practice for calibrated equipment to be traceable to the job it was used on, and this step is why.
- Assess the impact. Could the error have changed an accept or reject decision? A radiometer reading 5% high on a lamp that was well above the minimum may not matter; a UT sensitivity setting that drifted low may matter a great deal.
- Disposition and notify. Accept, reinspect or recall, and tell the customer where their work is affected.
- Correct the cause. Shorter interval, better handling, a different vendor, retraining. Record it as a corrective action in your quality system.
Some codes write part of this down for you. Under ASME Section V Article 4 (T-467.3), if a UT calibration check shows the sensitivity setting has dropped by more than its allowed margin, every data sheet since the last valid calibration check is void and the area it covered is reexamined. If distance range points have moved beyond their limit, the recorded indications since the last valid check are reexamined and corrected. That is the out-of-tolerance question with the answer already decided: the only hard part is knowing which data sheets they were.
What an auditor asks to see
- An equipment register with identity, calibration source and due date for every item in use.
- Certificates with as-found and as-left readings, the reference standard used, and traceability.
- Process-control records for the period, with readings against their limits and who did them.
- For any failure, the out-of-tolerance record: what was exposed, the assessment, the disposition and the corrective action.
- Where an interval was extended, the data that justified it.
For the wider quality system these records feed, see A QMS for an NDT company, and for how equipment follows a job from award to invoice, see Running NDT jobs end to end.
How TRACE handles it
Tool Control keeps the equipment register (serial, method, calibration source, NIST traceability, certificates and full check history) and a library of more than seventy recurring NDT checks, each carrying its governing standard and default acceptance criteria, which you can tighten to match your written practice. Daily checks fall due only on days the line runs, and readings are judged against their limits as they are typed, so a reading outside the limits stays marked even if someone logs it as a pass.
Custody and send-outs are records, not notes: sign a gauge out to a technician and the picker warns if they hold no current certification for the method; a send-out carries the vendor, RMA, expected-back date and cost, and the asset is out of service while it is away. A failed check opens an out-of-tolerance event and TRACE lists the jobs the asset was on and the reports written with it since its last good check, for you to rule in or out. It keeps the record; the assessment and the disposition stay your quality system's call.
Frequently asked questions
How often does an NDT instrument need calibrating?
As often as the document governing the work says, and the answer changes with the document. For example, ASME Section V (2025) asks for ultrasonic instrument linearity checks at least every three months for analog instruments and every year for digital ones, and annual calibration of magnetizing equipment with an ammeter. ASTM E1444-26 puts ammeter accuracy, yoke dead weight and radiometer calibration on a six-month cycle. ASTM E317 sets no interval at all and leaves it to the user. Your procedure or written practice should name the interval for each piece of equipment.
What is the difference between a calibration and a daily process check?
A calibration compares an instrument with a reference traceable to a national standard and records how far off it was, usually at a lab or against a certified reference, at a fixed interval. A process check is a quick, frequent confirmation that the system is working on the day: UV-A lamp intensity before inspection, a yoke lift, a system performance check on a known defect standard. Both are required; neither substitutes for the other.
What should happen when a gauge is found out of tolerance?
Take it out of service, record what it read and what it should have read, then work out what it touched: every job and report since its last good check. Assess whether that work could have been affected, decide what to do with it (accept, reinspect, notify the customer), and record the corrective action. Some codes spell part of this out; ASME Section V Article 4, for example, voids ultrasonic data sheets and requires reexamination when a sensitivity check shows the setting has dropped beyond its limit.
Does a timer on a penetrant line need calibrating?
Under ASTM E1417-21, timing devices used for dwell and similar processing times need not be calibrated. Magnetic particle equipment that uses a timer to control current duration is different: ASTM E1444-26 asks for that timer to be checked on its six-month cycle.
Checked against
- ASTM E1417/E1417M-21ε1, Standard Practice for Liquid Penetrant Testing, sections 6.6.2, 6.7.4.1, 7.8 and Table 1
- ASTM E1444/E1444M-26, Standard Practice for Magnetic Particle Testing for Aerospace, sections 7.1, 7.4 and Table 1
- ASTM E709-21, Standard Guide for Magnetic Particle Testing, section 20.2 and Table 2
- ASTM E317-21, Standard Practice for Evaluating Performance Characteristics of Ultrasonic Pulse-Echo Testing Instruments and Systems without the Use of Electronic Measurement Instruments, sections 1 and 5.7
- ASME Boiler and Pressure Vessel Code Section V, 2025 edition: Article 2 (T-262), Article 4 (T-461, T-467), Article 6 (T-676.4) and Article 7 (T-761, T-762, T-777.2)
This guide summarises the documents above in our own words to help you find your way around them. It is not a substitute for them: requirements change between editions and contracts add their own, so check every figure against your own licensed copy and your written practice before you rely on it.
Named for reference only. TRACE is an independent product: it is not affiliated with, endorsed by, sponsored by or accredited by ASTM International (publisher of E1417/E1417M, E1444/E1444M, E709 and E317), ASME (publisher of the Boiler and Pressure Vessel Code, Section V), the National Institute of Standards and Technology (NIST), NCSL International (publisher of ANSI/NCSL Z540.3) or ISO (publisher of ISO 10012 and ISO/IEC 17025), or by any other organization whose standards or programs it describes. Their names and marks are their own.