Translating an ISO 17640:2018 ultrasonic weld-testing report requires more than translating the word ‘ultrasonic’. An echo is not automatically a crack, a probe angle is not a bevel angle, and a testing level is not a weld-quality grade. An inaccurate translation can relocate a suspected discontinuity, exaggerate a detection, erase an inaccessible region or announce acceptance that an authorised examiner never gave. This guide explains the language of manual ultrasonic testing of welded joints and shows how to preserve the report’s evidence across languages.
People searching for ‘ISO 17640 translation’, ‘ultrasonic testing weld report’, ‘UT probe angle 45 60 70 degrees’, ‘weld indication echo amplitude dB’, ‘testing levels A B C D’, ‘sound path versus defect depth’, ‘DAC TCG calibration translation’ and ‘NDT weld acceptance criteria’ generally need to interpret a certificate, manufacturing record, repair dossier or inspection note. The practical task is to keep geometry, units, equipment settings, observation and judgement in their proper places so buyers and engineers can act on a faithful document.
The official ISO catalogue lists ISO 17640:2018, edition four, as the published manual ultrasonic weld-testing document, confirmed in 2023 and awaiting a future revision. Its public scope mainly covers suitable fusion-welded metallic joints of thickness at least 8 mm, with object temperatures from 0 °C to 60 °C. This article interprets publicly described concepts and common report language; it does not reproduce the licensed text, certify inspector competence, prescribe an examination, or replace the applicable acceptance procedure.
Read this guide by task
Understand the examination · Translate the instrument and signal · Preserve evidence and judgement · Four-week learning plan · Frequently asked questions · Official sources and connected learning. This narrowly solves one report-translation intent.
Understand the examination and its boundaries
1. Manual UT and its actual scope
The published ISO 17640:2018 describes techniques, testing levels and assessment for manual ultrasonic testing of fusion-welded joints in metallic materials. Its principal scope covers low-attenuation materials, thickness at least 8 mm and object temperatures 0–60 °C. The technical scope should travel with the report.
Common translation error: A translated inspection certificate labels an automated phased-array examination ‘manual ISO 17640’ although the record refers to a different inspection procedure.
Better handling: Retain the exact examination technique, material and conditions, identifying any specially specified use rather than broadening the standard’s ordinary scope.
Practical example: A site report about a six-millimetre joint requires clarification before it is presented as standard-scope work.
The first question is not ‘does this sound technical?’ but ‘could someone locate the same evidence from both language versions?’ A physical measurement, a calculation, an operator observation and a contract decision each enter the chain of proof differently. Keep them separately identifiable instead of blending them into one apparently confident conclusion.
2. What ultrasonic testing actually detects
Ultrasonic testing introduces sound into a component and interprets returning signals from acoustic boundaries. An indication is a response requiring interpretation, not automatically a crack, porosity or failure.
Common translation error: The source lists an ‘indication’ and the translation calls it a ‘confirmed fracture’ even though no flaw-characterisation evidence exists.
Better handling: Use ‘indication’, ‘reflector’ or the justified defect description as the actual record warrants; avoid turning possibilities into identities.
Practical example: An echo at a certain depth may reflect a weld feature, a geometric boundary or a discontinuity.
A good bilingual record leaves its numbers anchored. A numeral without a unit, reference plane, identifier or condition can become misleading even if no digit was mistyped. Copying a numeric value faithfully therefore means copying the constraints that tell the reader what that number represents.
3. Difference from radiographic examination
Ultrasonic testing infers information from sound propagation and reflected signals; radiographic testing uses ionising radiation to form images based on attenuation. Their records do not share identical vocabulary.
Common translation error: A bilingual table copies radiographic terms such as image density and film sensitivity into an ultrasonic report.
Better handling: Preserve A-scan, probe movement, echo amplitude, calibration and sound path language for UT; reserve radiographic terms for separate examinations.
Practical example: A purchaser comparing a UT record and a digital radiography report needs to know that their evidence is different.
Technical documents often pass through purchasing, fabrication, inspection and repair departments. Each reader may focus on a different part of the report. Clear translation protects the specialist meaning while offering plain-language explanation only where it does not create a new claim that the source did not contain.
4. The identity of the weld
An examination refers to a particular joint, location, drawing, weld identification and material. Losing this identity can attach the result to the wrong fabrication item.
Common translation error: Two welds have adjacent numbers. The translator shortens the identifier and makes it impossible to determine which joint contains the reported reflector.
Better handling: Copy joint tags, drawings and revision numbers character-for-character before editing surrounding prose.
Practical example: The phrase ‘weld W-17 on spool S-03’ is evidence of location, not unimportant formatting.
Treat small words such as ‘possible’, ‘recorded’, ‘not evaluated’, ‘limited’ and ‘subject to’ as important engineering information. Removing uncertainty can convert a cautious statement into an unsupported assertion. Strong technical prose is specific about what it knows and equally specific about what it cannot yet decide.
5. Weld configuration and penetration
Full-penetration and partial-penetration joints have different geometric and examination implications. ISO’s typical scope focuses on full-penetration ferritic joints; special cases demand defined procedures.
Common translation error: The translation removes ‘partial penetration’ because the terminology seems awkward, and the reader assumes a fully penetrated root.
Better handling: Retain groove type, backing, root configuration and any special procedure reference, without certifying penetration not established by the examination.
Practical example: Weld geometry can generate echoes that must not be mistaken for product defects merely because a signal appears.
An effective reviewer can work backwards from every translated statement to a table cell, procedure note, drawing or authorised conclusion in the original. If an assertion cannot be traced to any of those sources, classify it as a translator inference and resolve it before releasing the bilingual report.
6. Material and attenuation
Ultrasonic energy can attenuate because of absorption and scattering. Material microstructure and section shape affect inspectability and the reliability of the chosen method.
Common translation error: A report’s warning about attenuation is translated as ‘low sound volume’ rather than a meaningful technical limitation.
Better handling: Explain that attenuation refers to loss of signal with propagation, and note any restrictions on how confidently the examination covers the joint.
Practical example: The same equipment settings may not transfer sensibly between an ordinary ferritic steel and a coarse-grained material.
The reader needs a stable terminology map throughout the document. Choose equivalents for indication, reflector, sound path, depth, probe index and acceptance status, then preserve those choices in diagrams and footnotes. Changes in wording can be useful in ordinary prose but dangerous when they create two names for one measured variable.
7. Wall thickness and thickness mapping
Nominal drawing thickness, measured section thickness and the ultrasonic path through an angle-beam examination may not coincide. The unit and source must remain explicit.
Common translation error: The translator confuses an 18 mm wall with an 18 mm sound path and incorrectly relocates a reflector on the cross-section.
Better handling: Maintain fields for actual thickness, nominal thickness and sound path separately, making any conversion method traceable.
Practical example: A flaw location expressed from the scanning surface should not be reported as its total sound path.
Read the narrative and the diagrams together. Inspection prose may say ‘near the root’ while an accompanying figure supplies the coordinate origin. A translation that omits the diagram caption can mislead even when the narrative appears accurate. Visual references are part of the source’s meaning.
8. Object temperature range
The ISO scope publicly identifies object temperatures from 0 °C to 60 °C for its normal application. Couplant behaviour and sound velocity can change with temperature.
Common translation error: The source states the test piece was 48 °C and the translation calls it ‘48% sound transmission’.
Better handling: Keep temperature in degrees Celsius and state whether it is a testing condition, product service temperature or later design limit.
Practical example: An examination at elevated temperatures outside ordinary scope needs a separately justified procedure rather than a silent translation.
When an expression remains ambiguous after consulting the documented method, ask the responsible inspection authority a narrow question. A short clarification tied to the exact report field is more useful than a creative translation. Preserve the reply and do not let an editorial change masquerade as a revised test result.
9. Longitudinal and transverse sound velocities
ISO’s public abstract cites reference velocities for steels, distinguishing longitudinal waves from transverse waves. They are different wave modes, not two confidence levels.
Common translation error: A translated calibration record interchanges 5,920 m/s longitudinal and 3,255 m/s transverse reference values.
Better handling: Preserve the wave mode with every velocity, check material-specific calibration and avoid assuming universal speeds for all metals.
Practical example: A wrong velocity may distort calculated distance, which can shift the apparent location of an indication.
The first question is not ‘does this sound technical?’ but ‘could someone locate the same evidence from both language versions?’ A physical measurement, a calculation, an operator observation and a contract decision each enter the chain of proof differently. Keep them separately identifiable instead of blending them into one apparently confident conclusion.
10. Probe type and frequency
A conventional probe is described by its transducer type, nominal frequency and other identifying information. Frequency affects penetration, resolution and sensitivity but is not an automatic quality grade.
Common translation error: A 4 MHz probe is translated as ‘four-minute probe’, changing an equipment parameter into a duration.
Better handling: Retain MHz and the manufacturer’s identification. Explain frequency separately from probe angle and scanning speed.
Practical example: Before comparing two scans, check whether the equipment and probe configuration were genuinely comparable.
A good bilingual record leaves its numbers anchored. A numeral without a unit, reference plane, identifier or condition can become misleading even if no digit was mistyped. Copying a numeric value faithfully therefore means copying the constraints that tell the reader what that number represents.
Translate the instrument, geometry and signal
11. Probe angle and refracted beam
An angle-beam probe is used to introduce ultrasound at a specified refracted angle in the test material. Angle notation and the material in which it is defined matter.
Common translation error: A 60° probe is interpreted as a weld bevel angle rather than a refracted ultrasonic beam angle.
Better handling: Label ‘nominal refracted angle’ when appropriate and preserve the calibration context; distinguish mechanical wedge geometry from sound propagation.
Practical example: Two angled probes may interrogate different portions of a joint even when moved along the same scanning line.
Technical documents often pass through purchasing, fabrication, inspection and repair departments. Each reader may focus on a different part of the report. Clear translation protects the specialist meaning while offering plain-language explanation only where it does not create a new claim that the source did not contain.
12. Probe index and scan position
The probe index or exit point anchors position calculations for angle-beam scanning. Distance measured from this point differs from the edge of the probe casing.
Common translation error: The translator turns ‘index point’ into ‘serial number’, and an operator cannot reconstruct the scan geometry.
Better handling: Identify index point and scanning surface in words or a retained diagram, preserving measured offsets and origin.
Practical example: A change of coordinate origin can shift a reported indication despite leaving every numeric value apparently intact.
Treat small words such as ‘possible’, ‘recorded’, ‘not evaluated’, ‘limited’ and ‘subject to’ as important engineering information. Removing uncertainty can convert a cautious statement into an unsupported assertion. Strong technical prose is specific about what it knows and equally specific about what it cannot yet decide.
13. Sound path versus depth
Sound path describes travel distance along the beam. Depth is the perpendicular distance below a specified surface. For angle beams, these are generally not the same number.
Common translation error: A line stating sound path 50 mm becomes ‘defect depth 50 mm’, potentially relocating the indication.
Better handling: Keep the observed path and calculated depth separate, and do not reconstruct coordinates unless the source geometry supports it.
Practical example: A diagram can make the distinction clear: beam length is diagonal; depth is measured normal to the reference surface.
An effective reviewer can work backwards from every translated statement to a table cell, procedure note, drawing or authorised conclusion in the original. If an assertion cannot be traced to any of those sources, classify it as a translator inference and resolve it before releasing the bilingual report.
14. First leg, second leg and skip distance
Angle-beam sound can reflect from surfaces and travel along multiple legs. A signal location depends on the intended sound path and number of reflections, not just a display number.
Common translation error: The report notes a second-leg response but the translation describes an indication in the first leg, changing its inferred position.
Better handling: Retain leg and skip terminology, and preserve any drawing or calculation used for location.
Practical example: When a reviewer reads ‘two legs’, it must not become ‘two separate flaws’ unless the original supports that interpretation.
The reader needs a stable terminology map throughout the document. Choose equivalents for indication, reflector, sound path, depth, probe index and acceptance status, then preserve those choices in diagrams and footnotes. Changes in wording can be useful in ordinary prose but dangerous when they create two names for one measured variable.
15. Couplant and surface contact
A couplant improves acoustic transmission between probe and workpiece. Surface preparation, roughness and contamination can influence signal quality and examination reliability.
Common translation error: A translation treats ‘poor coupling’ as ‘weak weld bonding’, turning a scanning limitation into a metallurgical conclusion.
Better handling: Distinguish acoustic coupling at the inspection interface from weld fusion or bond integrity inside the component.
Practical example: A patch of coarse surface finish may require further examination without proving that the weld itself is damaged.
Read the narrative and the diagrams together. Inspection prose may say ‘near the root’ while an accompanying figure supplies the coordinate origin. A translation that omits the diagram caption can mislead even when the narrative appears accurate. Visual references are part of the source’s meaning.
16. Equipment calibration and verification
The instrument, probe, reference blocks and setup checks provide measurement traceability. A setup verification is not itself proof that every weld region was examined.
Common translation error: A calibration record from a reference block is copied into the product results table as a measured defect amplitude.
Better handling: Keep calibration, equipment check, operational scan and assessed indication in separate sections.
Practical example: A report can include an equipment verification value and a genuine weld indication with superficially similar numbers.
When an expression remains ambiguous after consulting the documented method, ask the responsible inspection authority a narrow question. A short clarification tied to the exact report field is more useful than a creative translation. Preserve the reply and do not let an editorial change masquerade as a revised test result.
17. Reference level and sensitivity
Sensitivity settings establish how signals are displayed relative to reference responses. Echo amplitude only has meaning with calibration and recording conventions.
Common translation error: A 6 dB change is translated as a sixfold change in amplitude with no consideration of the decibel scale.
Better handling: Retain dB and the direction of the adjustment, naming the reference response and setup where provided.
Practical example: If the original says ‘6 dB above reference’, keep the relationship to reference rather than reporting a bare percentage.
The first question is not ‘does this sound technical?’ but ‘could someone locate the same evidence from both language versions?’ A physical measurement, a calculation, an operator observation and a contract decision each enter the chain of proof differently. Keep them separately identifiable instead of blending them into one apparently confident conclusion.
18. DAC, TCG and DGS abbreviations
Distance-amplitude correction, time-corrected gain and distance-gain-size evaluation may appear in UT documentation. These are not synonyms, and not every report uses all three.
Common translation error: The translator expands every acronym as ‘dynamic adjustment control’, destroying the method vocabulary.
Better handling: Keep the verified expansion on first use, preserve the actual method and any calibration graph identifier.
Practical example: A multilingual form should not switch from DAC to TCG simply because the target language has a more familiar acronym.
A good bilingual record leaves its numbers anchored. A numeral without a unit, reference plane, identifier or condition can become misleading even if no digit was mistyped. Copying a numeric value faithfully therefore means copying the constraints that tell the reader what that number represents.
19. Scanning coverage
A stated coverage plan defines which surfaces and zones were interrogated. It does not mean the operator inspected every possible defect orientation with perfect detection probability.
Common translation error: The phrase ‘scanned from two sides’ becomes ‘100% defect-free weld’ in the translated summary.
Better handling: Describe access, scan direction and stated coverage independently of the evaluation outcome.
Practical example: Restricted access beside a stiffener should remain visible as a limitation, not disappear from the delivered report.
Technical documents often pass through purchasing, fabrication, inspection and repair departments. Each reader may focus on a different part of the report. Clear translation protects the specialist meaning while offering plain-language explanation only where it does not create a new claim that the source did not contain.
20. Testing levels A, B, C and D
ISO 17640 describes different testing levels associated with probability of detecting imperfections. A, B and C differ in examination requirements; D is reserved for special applications defined by specification.
Common translation error: A letter C in the test level field is rewritten as grade C of weld quality, confusing examination thoroughness with product acceptance.
Better handling: Translate testing level as a method or coverage designation, and separately identify any acceptance class used.
Practical example: The letter following ‘testing level’ is not a pass/fail mark awarded to a weld.
Treat small words such as ‘possible’, ‘recorded’, ‘not evaluated’, ‘limited’ and ‘subject to’ as important engineering information. Removing uncertainty can convert a cautious statement into an unsupported assertion. Strong technical prose is specific about what it knows and equally specific about what it cannot yet decide.
21. Evaluation level versus recording level
Reports may distinguish responses requiring evaluation from indications to be recorded. A response threshold does not itself define the severity or acceptance of the discontinuity.
Common translation error: A translator describes a recording threshold as the maximum permissible defect, making reportability identical to acceptability.
Better handling: Maintain each threshold’s function and report the relevant reference method without introducing a universal limit.
Practical example: Some responses may be documented for traceability even though subsequent evaluation finds them non-rejectable.
An effective reviewer can work backwards from every translated statement to a table cell, procedure note, drawing or authorised conclusion in the original. If an assertion cannot be traced to any of those sources, classify it as a translator inference and resolve it before releasing the bilingual report.
22. Echo amplitude and flaw size
Echo height is influenced by orientation, coupling, attenuation, beam geometry and reflector characteristics. It should not be directly translated into physical flaw length or height.
Common translation error: The source reports ‘40% full-screen height’ and a summary claims ’40 mm crack length’.
Better handling: Identify echo amplitude as a display or relative response measure; use flaw sizing only when performed with an appropriate stated technique.
Practical example: A small reflector facing the probe can give a stronger echo than a larger unfavourably oriented reflector.
The reader needs a stable terminology map throughout the document. Choose equivalents for indication, reflector, sound path, depth, probe index and acceptance status, then preserve those choices in diagrams and footnotes. Changes in wording can be useful in ordinary prose but dangerous when they create two names for one measured variable.
23. Indication length and location
Reported indication length may be determined through probe movement and evaluation rules. It is not automatically the true dimension of a physical crack.
Common translation error: An indication spanning 12 mm of scan movement is translated as a 12 mm-thick discontinuity through the wall.
Better handling: Keep the measurement direction, sizing procedure and positional reference, and explain any distinction between estimated and actual dimensions.
Practical example: A drawing annotation ‘length along weld’ must not turn into ‘penetration depth’ during localisation.
Read the narrative and the diagrams together. Inspection prose may say ‘near the root’ while an accompanying figure supplies the coordinate origin. A translation that omits the diagram caption can mislead even when the narrative appears accurate. Visual references are part of the source’s meaning.
Preserve evaluation, status and reporting integrity
24. Characterisation versus observation
The examiner may characterise a reflector as planar, volumetric, geometric or unclassified. The strength of that label depends on evidence and the procedure.
Common translation error: An ambiguous response is recast as confirmed lack of fusion even though the record calls for further evaluation.
Better handling: Retain qualifications such as ‘suspected’, ‘possible’ or ‘unclassified’ and avoid adding a defect name just to make prose specific.
Practical example: A high-risk indication can prompt further testing without allowing a translator to diagnose it.
When an expression remains ambiguous after consulting the documented method, ask the responsible inspection authority a narrow question. A short clarification tied to the exact report field is more useful than a creative translation. Preserve the reply and do not let an editorial change masquerade as a revised test result.
25. Acceptance criteria and ISO 11666 references
A method or assessment standard does not by itself provide every contract’s acceptance decision. Acceptance may depend on other applicable standards, specifications and inspection levels.
Common translation error: The sentence ‘tested to ISO 17640’ becomes ‘accepted to ISO 17640’ despite no identified acceptance criterion.
Better handling: State the test method and separately quote the actual acceptance document if the signed source names one.
Practical example: A procurement record needs to show both what examination was carried out and what rule determined acceptability.
The first question is not ‘does this sound technical?’ but ‘could someone locate the same evidence from both language versions?’ A physical measurement, a calculation, an operator observation and a contract decision each enter the chain of proof differently. Keep them separately identifiable instead of blending them into one apparently confident conclusion.
26. Result states and authorised sign-off
Detected, evaluated, recorded, rejected, accepted and repaired describe different stages of a quality process. The status should match what the report actually certifies.
Common translation error: A finding marked ‘pending review’ is translated into ‘passed’ because the table is ultimately part of a project completion package.
Better handling: Preserve status words and approval authority, especially before manufacturing release or repair decisions.
Practical example: A translator should not replace an examiner’s signature or a separate engineer’s disposition.
A good bilingual record leaves its numbers anchored. A numeral without a unit, reference plane, identifier or condition can become misleading even if no digit was mistyped. Copying a numeric value faithfully therefore means copying the constraints that tell the reader what that number represents.
27. Repair and re-examination
An original report and a post-repair examination may concern the same weld but different states of the work. A final summary must not erase the earlier history.
Common translation error: A record of an indication before repair is removed from the bilingual dossier because a later UT scan found no reportable response.
Better handling: Keep the sequence, dates, repair references and distinct examination identifiers.
Practical example: Traceability requires readers to see which evidence belongs to the original fabrication and which belongs to later rework.
Technical documents often pass through purchasing, fabrication, inspection and repair departments. Each reader may focus on a different part of the report. Clear translation protects the specialist meaning while offering plain-language explanation only where it does not create a new claim that the source did not contain.
28. Report date, edition and procedure revision
A report can cite the 2018 ISO edition, an internal technique revision and a test date in a different year. These are separate data fields.
Common translation error: The translator changes the standard reference to 2026 because the inspection happened in 2026, creating a citation to a non-existent edition.
Better handling: Retain the exact standard edition and the actual procedure version used on the specified examination date.
Practical example: The ISO catalogue also notes a draft replacement under development, but that does not retroactively update completed reports.
Treat small words such as ‘possible’, ‘recorded’, ‘not evaluated’, ‘limited’ and ‘subject to’ as important engineering information. Removing uncertainty can convert a cautious statement into an unsupported assertion. Strong technical prose is specific about what it knows and equally specific about what it cannot yet decide.
29. Uncertainty, limitations and dead zones
UT has sensitivity limitations near boundaries and in difficult geometries. Reporting these limitations is part of responsible evidence, not an admission to hide.
Common translation error: A report mentions a region inaccessible to the probe; translation says the full weld volume was tested without restriction.
Better handling: Preserve the inaccessible area, qualified exclusion and any recommended complementary examination.
Practical example: A transparent limitation prevents a downstream reader from assuming that an unexamined region was proven defect-free.
An effective reviewer can work backwards from every translated statement to a table cell, procedure note, drawing or authorised conclusion in the original. If an assertion cannot be traced to any of those sources, classify it as a translator inference and resolve it before releasing the bilingual report.
30. Bilingual table labels and numerical symbols
A sound translation of an A-scan result needs test item, probe, angle, reference, indication position, amplitude, length and evaluation fields to remain aligned.
Common translation error: A Word export wraps the ‘depth’ header over the ‘sound path’ column, so all values acquire the wrong interpretation.
Better handling: Audit the layout as well as words, especially axis labels, minus signs, decibel references, mm, MHz and degree symbols.
Practical example: For critical records, a second reviewer should reconstruct the result from the translated table without seeing the source.
The reader needs a stable terminology map throughout the document. Choose equivalents for indication, reflector, sound path, depth, probe index and acceptance status, then preserve those choices in diagrams and footnotes. Changes in wording can be useful in ordinary prose but dangerous when they create two names for one measured variable.
31. Disputes and clarification requests
Ambiguity can arise when a handwritten record, drawing annotation and electronic report disagree. The translator must distinguish correction authority from linguistic editing.
Common translation error: A project assistant chooses whichever number makes the acceptance outcome favourable and quietly updates the translation.
Better handling: Issue a targeted query with document ID, line, original phrase, proposed interpretation and requested confirmation.
Practical example: The final bilingual report should retain an audit trail of the authorised clarification rather than conceal it.
Read the narrative and the diagrams together. Inspection prose may say ‘near the root’ while an accompanying figure supplies the coordinate origin. A translation that omits the diagram caption can mislead even when the narrative appears accurate. Visual references are part of the source’s meaning.
32. Technical release checklist
Final review needs to prove the translation preserved the physical meaning, not merely that the sentences sound native. Check sign, scale, probe configuration, method, scope and decision authority.
Common translation error: A ‘clean’ translation omits sound-path qualifiers, examination level, restricted regions and reportable indications while retaining the overall word ‘passed’.
Better handling: Verify each title, weld ID, method reference, material, thickness, settings, data row, reviewer status and source limitation before issue.
Practical example: The reader should be able to infer what was measured, where it occurred, which rule was applied and what remains unknown.
When an expression remains ambiguous after consulting the documented method, ask the responsible inspection authority a narrow question. A short clarification tied to the exact report field is more useful than a creative translation. Preserve the reply and do not let an editorial change masquerade as a revised test result.
A four-week translation learning plan
Week 1: Inspect scope and specimen identity
Build a glossary of weld configuration, material, thickness, ultrasonic technique, testing level and probe terms. Use a mock drawing to match a weld number with a test record. Explain why an ultrasonic indication is not automatically a named defect. At the end of the week, ask another reader to identify exactly which fabrication item was tested.
Week 2: Reconstruct geometry and signals
Create a diagram showing sound path, depth, first leg, second leg and probe index. Translate a fictional table that contains MHz, degrees, mm and dB without changing what they modify. Use the source and target versions to reconstruct the reported position independently; discrepancies show where the language has changed the physics.
Week 3: Separate observation from acceptance
Write three paragraphs about a hypothetical response: what the examiner observed, what they evaluated, and what an authorised reviewer decided. Refer only to the named criteria present in the example. Replace overconfident words with properly supported qualifications and keep any inaccessible region visible in the translated summary.
Week 4: Release a traceable bilingual dossier
Audit an anonymised weld record line by line. Check method year, procedure revision, probe settings, calibration references, measured quantities, indication status, repairs and re-examination. Add a short query log for contradictory source fields. Have a second reader determine whether the target-language document supports the same decision, no more and no less.
Frequently asked questions about ISO 17640 translation
Does a strong ultrasonic echo prove a crack?
No. Echo amplitude depends on several factors and an indication may have different causes. Keep the observation distinct from any supported flaw characterisation.
What is the difference between sound path and depth?
Sound path follows the beam’s travel; depth measures distance below a specified reference surface. With an angle-beam probe those numbers can differ substantially.
Does testing level C mean quality grade C?
No. Testing levels describe examination requirements and detection considerations, not a pass/fail quality grade.
Can a translation replace ‘indication’ with ‘defect’?
Only if the source and applicable assessment truly justify that conclusion. Otherwise it prejudges the status of the reported response.
Is ISO 17640 the same as weld radiography?
No. Ultrasonic testing and radiography use different physical methods and reporting conventions. Related evidence can be compared only with appropriate technical context.
What if an examination was outside normal scope?
Keep any special procedure and qualification evident. Do not claim ordinary standard-scope application when the material, thickness or temperature requires special provisions.
Can I translate 6 dB as 6% more?
No. A decibel is logarithmic. Preserve the dB value and reference relationship unless a qualified calculation is documented.
Does ‘tested to ISO 17640’ mean a weld passed?
Not by itself. A method statement does not replace the separate acceptance criteria and the authorised assessment recorded for that weld.
Official reference and connected learning
Use the official ISO 17640:2018 catalogue for the standard’s current published status and public scope. To understand the distinct radiographic method, see the existing ISO 17636-2 digital radiography translation article. The wider translation architecture remains with Master Art of Translation, supported by Vocabulary Learning Hub and How English Works. This page does not replace those established owners.
A reliable ultrasonic-testing translation permits the same inspection decision to be traced through two languages. It protects the weld’s identity, the probe’s geometry, the instrument’s observations, and the exact boundary between reported evidence and authorised judgement.