A cable harness can fail an end-of-line test even when the harness is good. Worn mating connectors, oxidized pins, bent adapter contacts, loose fixture wiring, contaminated probes, an incorrect fixture map or a drifting tester can add resistance or create intermittent opens. The opposite risk is more serious: a miswired fixture or wrong test program can allow a bad harness to pass.
Procurement should therefore treat the tester, software, interface adapters, mating connectors, fixture wiring, probes, reference artifacts and test program as one controlled measurement system. The RFQ needs fixture drawings, contact-resistance budgets, replaceable-wear strategy, verification methods, calibration/traceability, golden-sample governance, maintenance limits and false-failure disposition. A continuity threshold alone is not a test-system specification.
This guide focuses on fixture architecture and measurement-system control at cable-harness end of line. It complements the cable-harness continuity, insulation-resistance and hipot guide, which defines electrical test content, and the battery-cable Kelvin resistance guide, which focuses on product resistance measurement. It does not prescribe live high-voltage test practice, certify a tester or replace the equipment manufacturer, metrology function or customer test authority. No statement here claims an unverified SINAWATTS tester, fixture, calibration laboratory, test capability, acceptance limit, certification, yield, stock, price, MOQ or lead time.
Direct answer: what should the RFQ require?
For each harness part number and test station, require:
- tester manufacturer, model, serial-control method, hardware/firmware and software revision;
- approved test-program number and revision tied to the harness drawing/BOM revision;
- complete fixture schematic from tester points to every DUT pin, shell, shield and accessory;
- adapter, mating-connector, pogo-pin, probe and fixture-cable part numbers;
- two-wire or four-wire measurement topology for each requirement;
- resistance budget separating tester uncertainty, fixture leads, mating contacts, switching and DUT allowance;
- continuity/open/short/component limits and expected test current or voltage where relevant;
- fixture verification method before release and at defined intervals;
- replaceable wear interfaces and rated/validated mating-cycle plan;
- daily checks, preventive maintenance and retirement criteria;
- calibrated standards/performance-check artifacts and traceability;
- false-failure troubleshooting and retest rules;
- golden-unit creation, protection, verification and permitted uses;
- operator mistake-proofing and connector keying;
- test-result traceability to station, fixture, program, operator, product and time; and
- a signed deviation schedule.
Do not accept “100% electrically tested” without identifying which tests, limits, fixture/program revision and measurement-system evidence produced the result.
Separate product failure from fixture failure
Use four status categories:
- Confirmed product failure: the defect follows the DUT on a verified station or is confirmed by an independent approved method.
- Confirmed fixture/test-system failure: a reference artifact, fixture verification or fault isolation shows the interface or tester caused the indication.
- Intermittent/undetermined: evidence is not yet sufficient to assign cause.
- Operator/setup error: wrong program, adapter, orientation, connection or step caused the result.
Do not automatically retest until a fail becomes a pass. Uncontrolled retest can hide intermittent harness defects or fixture wear. Define when one retest is permitted, what must be inspected first, and how the original fail remains in the record.
Track false-failure rate by station and fixture. A rising rate can indicate wear before hard failures appear. Track escapes and audit findings separately; a low false-failure rate does not prove the fixture map is correct.
Draw the complete test path
The electrical path can include tester relays, scanner backplane, adapter card, adapter connector, fixture cable, intermediate connector, replaceable mating connector and DUT terminal. Each interface contributes resistance and potential intermittency.
Create one controlled schematic that shows:
- tester point addresses;
- force and sense pairs where four-wire measurement is used;
- shield/shell connections;
- jumpers and common nodes;
- component simulation or loads;
- safety interlocks;
- fixture-identification devices;
- connector cavity names exactly as seen by the operator;
- mating-interface replaceable sections; and
- measurement boundary.
Use consistent “mating face” or “wire side” views. The harness pinout, cavity-view and wire-identification guide explains this drawing risk. A fixture wired from the opposite view can systematically validate the wrong pin map if the test program was learned from the same wrong fixture.
An independent design review should compare the customer schematic, harness drawing, fixture schematic and program netlist. Do not create the acceptance program only by “learning” one unknown sample without an independent expected netlist.
Budget contact resistance before setting limits
In a two-wire measurement, test-lead and interface resistance are included with the DUT. The threshold must allow measurement uncertainty and normal fixture variation without masking product resistance. In a four-wire measurement, force and sense leads separate current delivery from voltage measurement, but the Kelvin connection point determines which contacts and leads are excluded.
Keithley’s official Low Level Measurements Handbook, checked on 2026-09-28, explains that a two-wire low-resistance measurement includes both lead resistances in the result. That principle applies broadly; the handbook does not define an acceptance threshold for a harness. Keithley low-resistance handbook.
Build a resistance budget with actual boundaries:
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Contribution | Two-wire treatment | Four-wire treatment | Control evidence |
|---|---|---|---|
| Tester/relay path | Included or compensated per system | Depends on architecture | Tester specification/performance check |
| Fixture cable | Included unless offset | Excluded only beyond Kelvin split | Wire length/gauge and verification |
| Adapter contacts | Included | May remain inside measurement | Wear/cleaning and reference check |
| DUT mating contact | Usually included | Included if Kelvin split is behind it | Exact production-representative mate |
| DUT conductor/terminations | Intended product result | Intended product result | Drawing and limit |
| Temperature | Changes conductor and contacts | Still relevant | Defined conditioning/correction |
| Uncertainty/guard band | Must be reserved | Must be reserved | Measurement-system analysis |
Do not subtract one fixed “fixture resistance” forever. Contacts age and temperature changes. If compensation is allowed, define how it is established, maximum permitted offset and re-verification interval. A large offset can make a damaged fixture appear acceptable.
Place Kelvin points where the product requirement begins
Cirris’s CR user manual, checked on 2026-09-28, describes its four-wire method as using force and sense points and says properly configured fixtures can eliminate test cable/fixture resistance from the measured value beginning where each four-wire pair meets. It also notes that four test points are required per measurement. Those specifications apply to the named Cirris system and revision; they illustrate why fixture configuration defines the measurement boundary. Cirris CR user manual.
For a low-resistance harness, decide whether the mating contact is part of the product requirement. If the customer wants terminal-to-terminal resistance through the production contacts, place the Kelvin split so the mating interfaces are included appropriately. If the requirement begins at a crimp barrel or dedicated sense point, document that geometry. Do not claim “lead resistance removed” without a diagram.
Four-wire topology does not eliminate:
- unstable DUT mating contacts inside the measured boundary;
- thermoelectric offsets at very low levels;
- wrong force/sense pairing;
- insufficient test current resolution;
- fixture heating;
- contact damage from probes; or
- test-program mapping errors.
Use a representative standard and reversal/offset method where the tester manufacturer requires it.
Design replaceable wear interfaces
The production DUT should mate to a replaceable connector or contact saver rather than wear an expensive tester backplane. Select mating components that represent the production interface without damaging terminals. Control contact plating, keying, polarization and retention.
Cirris’s current adapter-system page, checked on 2026-09-28, describes replaceable mating connectors on some adapters and says adapters act as replaceable wear items. It recommends aligning adapters squarely, using locks/cover plates and inspecting pins. Cirris adapter system.
Its 4200-series getting-started guide says contact resistance gradually increases with mating cycles, gives a product-specific statement that premium gold plating can maintain very low contact resistance over its stated cycle range, and tells users to inspect right-angle pins and use slide locks. Do not transfer the stated cycle/resistance expectation to a different connector. Use it as evidence that maintenance limits must be defined. Cirris 4200-series guide.
The fixture BOM should identify every wear item and replacement method. A replaceable connector must be keyed so it cannot be installed one cavity off. If soldered or crimped onto fixture wiring, define the post-replacement verification.
Verify the fixture independently from the DUT
Fixture verification should prove that every intended path exists, maps to the correct DUT cavity and is isolated from unintended paths. Use a probe, shorting block, known adapter or another manufacturer-approved method. Verify after fixture build, repair, connector replacement, mapping change and suspected damage.
Cirris’s official Verifying Your Test Fixturing article says its named systems’ fixture-verification function checks that each conductor is present, connected to the correct mating-connector pin and not shorted to another. It explicitly warns that a DUT cannot be guaranteed correct if fixture/program validity is not established. Cirris fixture verification.
Independent verification should include:
- remove the DUT and use the approved verification artifact/process;
- confirm tester point to physical cavity mapping;
- challenge adjacent pins and shells for unintended shorts;
- verify force/sense pairs and Kelvin split;
- flex fixture cables and connectors while monitoring intermittency;
- verify fixture identity/program interlock;
- save the result with fixture serial and revision; and
- apply tamper-evident control after acceptance where appropriate.
A golden harness alone cannot independently verify a mirrored map if it was built from the same wrong information. Use design-derived verification plus known defects/challenge samples.
Challenge detection with fault-insertion samples
Create controlled artifacts or switch boxes that introduce known opens, shorts, high-resistance connections, swapped pins and intermittent conditions. Verify that the test program detects each required defect and reports the correct cavity names.
For safety, do not improvise faults during high-voltage testing. Use equipment-manufacturer-approved verification artifacts and de-energized fixture design. Keep high-voltage verification separate from low-resistance fixture maintenance when the equipment requires it.
The challenge matrix should cover:
- open at first and last network point;
- adjacent and nonadjacent shorts;
- swapped wires that preserve simple continuity counts;
- resistance just below and above limit;
- shell/shield continuity and isolation;
- components, diodes or resistors in both orientations where applicable;
- intermittent connection under the approved flex/manipulation method; and
- wrong adapter/program identification.
Record expected and actual results. A test that finds an open but labels the wrong cavity can cause incorrect repair.
Detect and manage adapter resistance
Cirris’s official Testing Adapter Resistance article, checked on 2026-09-28, describes creating a shorting block/program and using continuous testing while wiggling the block to reveal intermittent adapter problems. It says a pass means contact resistances meet the program requirement and recommends replacing a failing adapter. This is a Cirris-specific method that illustrates a general procurement requirement: provide a documented health check for the replaceable interface. Cirris adapter-resistance test.
Define fixture health limits below the product rejection limit with enough guard band. A fixture should be removed before it creates repeated product false failures. Trend actual contact resistance where the system permits it.
Do not clean plated contacts with an abrasive or solvent unless the connector/tester manufacturer approves it. Cleaning can remove plating, spread contamination or leave residue. Record maintenance material, method and date.
Control tester calibration and intermediate verification
Calibration addresses the tester’s measurement performance; fixture verification addresses the interface and mapping. One does not replace the other.
Cirris’s calibration page, checked on 2026-09-28, says new Cirris testers include calibration under its stated ANSI/NCSL Z540-1 process and recommends verification at least annually. Its Performance Check Kit page says the named kits include NIST-traceable standards and that a failed verification requires service. These are manufacturer statements for Cirris equipment, not universal calibration intervals for every tester. Cirris calibration and Cirris performance check kits.
The RFQ should require:
- calibration interval justified by manufacturer, use, risk and history;
- calibration scope covering the used functions/ranges;
- traceability chain and certificate results;
- as-found and as-left data where required;
- intermediate performance checks using controlled standards;
- response when a check fails;
- impact assessment back to the last known-good check; and
- control of reference-artifact calibration and shelf life.
Do not place a “calibrated” sticker on a fixture unless the term is defined. Fixtures are often verified rather than calibrated. Identify the controlled characteristic and method.
Quantify uncertainty and guard band near the limit
For a tight low-resistance requirement, identify tester accuracy, resolution, repeatability, fixture/contact variation, reference uncertainty, temperature effect and operator/setup variation. Choose acceptance/guard bands through the customer-approved measurement decision rule.
Run a measurement-system study using representative low, nominal and near-limit samples across operators, stations and fixture replacements. A conventional gauge R&R percentage may not suit attribute-like intermittent detection; use a study that matches the risk and data. Preserve raw data.
Do not widen the product limit to make a noisy fixture pass. Improve contacts, use four-wire topology, control temperature, shorten fixture leads, service the interface or choose a better measurement system. If uncertainty consumes too much of the tolerance, the process is not capable of making reliable decisions.
Prevent false passes from learned programs
Automatic learning can speed program creation but must be reconciled to the authoritative design. A learned good sample can contain a hidden miswire. A fixture and sample built from the same wrong cavity view can agree with each other.
Require program verification against:
- released harness schematic and netlist;
- connector drawings with view definitions;
- BOM components and orientation;
- customer-defined unused pins and isolation groups;
- shield/shell connections;
- test limits and units; and
- fault-insertion results.
Use electronic approval and revision control. Restrict editing rights on production stations. The result record should store a program hash or revision, not only a friendly name.
Operator ergonomics are measurement controls
A fixture that invites angled insertion, cable peeling or unsupported harness weight will wear faster and produce intermittent contact. Provide alignment features, latches, strain relief and clear orientation. Prevent operators from pulling on wires to disconnect.
Cirris’s connector-life article, checked on 2026-09-28, explains that peeling long connectors can bend end pins and that locking adapter boards reduces receptacle wear. Again, the specific cycle claims belong to its components; the field lesson is to control insertion/removal. Cirris connector life guidance.
Review reach, force, visibility and pinch hazards. Locate reject indicators so the operator does not remove the DUT before failure data are saved. Interlock the correct adapter and program. Provide a safe parking location for probes and caps to avoid contamination.
Use a bounded worked comparison
Consider a hypothetical harness with a maximum allowed terminal-to-terminal resistance of 80 mΩ. The numbers illustrate evidence review and are not a recommended limit.
Bid A uses a two-wire fixture whose new leads/interfaces measure 35 mΩ and sets the tester at 80 mΩ. A compliant 60 mΩ harness could read 95 mΩ and fail. The bidder proposes subtracting 35 mΩ but has no ageing or daily verification. Its measurement system is not controlled.
Bid B uses four-wire measurement but places the Kelvin split behind the fixture mating connector, excluding a contact that the product requirement intends to include. Its reading is repeatable but measures the wrong boundary.
Bid C defines the requirement boundary, locates Kelvin points accordingly, budgets uncertainty, uses replaceable mating interfaces, verifies fixture mapping independently, challenges known defects, performs daily reference checks and trends adapter resistance. It retains original fails and requires fault isolation before retest. Bid C provides the strongest decision evidence; the customer/metrology owner must approve the final rule.
False-failure containment and retest
When a DUT fails:
- preserve the original result and waveform/details;
- inspect connector seating and visible damage without altering the product;
- run the approved fixture health/reference check;
- if the fixture fails, quarantine results since the last known-good verification;
- if the fixture passes, retest only under the approved rule;
- confirm intermittent failures using a defined manipulation method that does not damage the DUT;
- route the DUT for independent confirmation when cause remains unclear; and
- record final disposition and causal category.
Never erase an initial failure because a second test passed. Trend repeat-pass events. They can signal marginal product contacts or fixture wear.
If the tester/reference check fails, assess previously accepted product back to the last valid check based on risk. The Cirris Performance Check Kit page explicitly notes that a failed standard can affect confidence in testing since the last certification; the project quality system should define the actual containment window.
IPC acceptance scope and customer test requirements
IPC’s current document-revision table, checked on 2026-09-28, lists IPC/WHMA-A-620 Revision E, and IPC describes the standard as covering requirements and acceptance for cable and wire harness assemblies. IPC’s overview says it addresses materials, methods, tests and acceptance for harness assembly activities. IPC revision table and IPC/WHMA-A-620 validation overview.
Do not claim that IPC/WHMA-A-620 alone defines every customer test limit, fixture or program. The purchase drawing, customer test specification, class/addendum and contract determine requirements. State the exact revision and precedence. Individual training or a company listing must also be described accurately; neither automatically validates a specific fixture.
First article and production release
Release the test system only after:
- fixture wiring and cavity mapping are independently verified;
- program netlist is compared with the released design;
- known-defect challenges are detected and named correctly;
- resistance standards confirm relevant ranges;
- Kelvin boundaries are documented;
- repeatability is demonstrated near limits;
- tester calibration/performance checks are current;
- operator instructions and safety interlocks are validated; and
- result traceability is reviewed.
Run actual first-article harnesses through the released system and, where risk requires, compare with an independent reference method. Archive the fixture/program baseline before production.
Production records should include product serial/lot, result, measured values, failure details, station/tester, fixture serial/revision, program revision/hash, operator, time and environment where relevant. Store enough data to investigate drift.
RFQ evidence matrix
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Decision field | Buyer requirement | Supplier return | Release evidence |
|---|---|---|---|
| Test boundary | Pins/contacts included | Exact schematic | Drawing and program netlist |
| Tester | Model, range, accuracy and revision | Exact station | Datasheet/calibration |
| Fixture | Adapters, cables, wear contacts | Controlled BOM | Fixture drawing |
| Resistance budget | Product plus measurement system | Contributions/guard band | Analysis and capability data |
| Mapping | Correct cavity/test point | Verified map | Independent fixture verification |
| Detection | Opens, shorts, swaps, resistance, intermittents | Covered defects | Fault-insertion report |
| Maintenance | Limits, cycles, cleaning and replacement | Schedule | Health-check records |
| Calibration | Scope, interval and traceability | Current route | Certificates/check standards |
| Retest | Preserve original and isolate cause | Procedure | Sample failure record |
| Data | DUT/fixture/program/station traceability | Database/report | Example production result |
“Pass” without the system identity is weak evidence. Require the context needed to reproduce the decision.
Change control
Reopen test-system validation for harness drawing/BOM revisions, connector/terminal changes, fixture rewiring, mating-contact replacement, adapter changes, tester hardware/firmware/software updates, altered limits, changed Kelvin points, new compensation, calibration failure or a revised IPC/customer requirement. Also reopen when false failures, escapes or intermittent rates change materially.
Back up approved programs before updates. Validate migrated programs on the new software. Do not assume identical program names produce identical sequences across revisions.
Source boundaries checked on 2026-09-28
The Cirris adapter, manuals, fixture-verification, adapter-resistance, calibration and performance-check pages; Keithley low-resistance handbook; and IPC standard-scope pages linked above were checked on 2026-09-28. Cirris specifications and intervals apply to its named products. Keithley explains measurement principles but does not define a harness limit. IPC defines industry acceptance scope; the customer contract sets the actual test program. Final equipment selection, safety, uncertainty and acceptance rules remain with qualified parties.
Send the harness drawing, pinout, resistance limits and required test coverage for a structured RFQ. Include expected volume, connector mating constraints, traceability and customer standard revision so suppliers can return a fixture, verification and maintenance package rather than only a tester model. The actual supplier must confirm capability, price, quantity and lead time.
Buyer FAQ
Why can a good harness fail continuity testing?
Fixture lead/contact resistance, worn adapters, bent pins, contamination, poor seating, mapping errors or tester drift can create an apparent open or high resistance.
Does four-wire testing remove all fixture errors?
No. It can exclude lead resistance beyond the Kelvin split, but contact instability, wrong mapping, bad force/sense pairs, temperature and program errors remain.
Where should Kelvin points be placed?
At the boundary defined by the product requirement. Document whether production mating contacts are included; do not choose placement only for a lower reading.
Is a golden harness enough to validate a fixture?
No. If the golden harness and fixture share the same mapping error, they can agree incorrectly. Use design-derived independent verification and known-defect challenges.
How often should adapters be replaced?
Use connector-manufacturer data, validated cycle/use history, resistance trends and a defined health limit. Do not copy another adapter’s cycle rating.
Is tester calibration the same as fixture verification?
No. Calibration verifies measurement performance against standards. Fixture verification checks wiring, mapping, interfaces and condition. Both are needed.
Should a failed harness be retested immediately?
Only under a controlled rule that preserves the original fail and checks fixture health first. Repeated testing until pass hides risk.
What is a false-failure guard band?
It is the reserved margin between the product limit and measurement-system uncertainty/variation used under an approved decision rule. It should not silently widen the product specification.
What data should every test result contain?
Record DUT identity, values/result, failures, tester, fixture serial/revision, program revision/hash, operator and timestamp, plus relevant environmental or calibration state.
What changes require revalidation?
Fixture repairs, connector replacement, rewiring, program or limit changes, tester/software updates, new product revisions and any calibration or performance-check failure should trigger defined revalidation.