When a factory keeps wire harness continuity results on paper or inside a tester, finding the evidence for one shipped unit can be slow. Which drawing revision and test program were used? Was the first attempt a failure? Was the unit repaired and tested again? A recording system should connect the physical harness, the actual test conditions and every test event. This guide explains how a Thailand-based harness plant can specify, buy, validate and operate that connection without confusing electrical continuity with other quality checks.
Define the electrical tests before designing the database
A continuity test checks whether terminals intended to be connected have an electrical path. A shorts test looks for unintended connections. A miswire test compares actual pin destinations with an expected connection map. Insulation resistance and high-potential tests address a different property and may require different equipment and safety controls. A single PASS on a tester screen says little unless the applicable recipe and test coverage are known.
| Test | Question answered | Evidence to retain |
|---|---|---|
| Continuity or connection resistance | Is the intended path present within the specified resistance range? | Pin pair, measurement or decision, limit and recipe revision |
| Shorts | Are independent nets unintentionally connected? | Nets examined, decision and threshold |
| Miswire | Does each wire reach the correct destination pin? | Expected map and observed discrepancy |
| Insulation resistance | Is measured isolation resistance above its specified minimum? | DC test voltage, duration, resistance in Ω/MΩ, minimum limit and applicability |
| Dielectric withstand or hipot | Is there excess leakage or breakdown at the specified test voltage? | AC/DC mode, voltage, dwell, leakage or trip current, breakdown result and limit |
| Fixture check | Can the adapter itself distort the result? | Fixture ID, inspection state and self-test result |
Cirris explains resistance thresholds in terms of intended and unintended connections. Its example settings are guidance, not universal acceptance limits. Establish limits from the customer drawing and product specification, allowing for cable length, gauge, terminals, contact resistance, fixture contribution and measurement method. A low-resistance application may call for four-wire measurement. For harnesses containing resistors, diodes or other components, simple pattern continuity may miss relevant defects; Cirris describes those component-testing limits.
IPC describes IPC/WHMA-A-620 as a cable and wire harness assembly requirements and acceptance standard. That description alone does not impose a universal MES field list, one test threshold or a single retention period. Confirm the edition, customer contract, product class and inspection plan that actually apply.
Three gaps that a recording system must close
The first gap is identity. A chronological list of tester results does not reliably identify the harness in front of an operator. Scan a serial number, unique lot position or other approved unit identifier before the test. Match it to the work order and part revision. If the process uses lot-level identification, define how an individual failed harness is isolated and returned to the lot record. Typing an identifier into a spreadsheet after testing creates an avoidable mix-up risk.
The second gap is test context. A part number can be tested under several drawing, program and threshold revisions. Preserve the revision used at the moment of testing; do not allow a later master-data update to change the interpretation of historical events. Save the fixture and tester identity as well. A result without its conditions cannot be reproduced or defended in an audit.
The third gap is process context. Crimp monitoring and final electrical test establish different facts. A sound crimp process does not ensure that a wire reached the correct pin, and electrical continuity alone does not prove every aspect of crimp quality. Link the two histories through the same unit ID. Our article on harness crimp monitoring covers the earlier operation; this article focuses on the final electrical result and its evidence.
Give the tester, shop-floor terminal and MES distinct jobs
The tester is the source of electrical measurements and decisions. A shop-floor terminal identifies the unit, work order, operator and approved fixture, and guides the sequence. The MES or quality data platform stores immutable test events, supports search and controls the shipping gate. If ownership is unclear, an integration program may overwrite a tester verdict or silently substitute the server receive time for the test time.
A workable sequence is: scan the harness; check part and drawing revision; select and verify the approved test recipe; verify fixture and tester readiness; run the test; capture the native result; persist one event; and return a release or hold decision. Define in advance what happens if any identity or version check fails. This is a quality decision, not merely an interface error.
Actual connectivity depends on the installed equipment. A tester might expose a supported API, a result file, a serial interface or a manufacturer’s application database. Cirris Easy-Wire describes programmable tests, archived reports with measured values and shared network data. CAMI describes automatic data logging and review of archived results. These are examples of product capabilities, not a guarantee that every installed model, license and software version exports every field needed by your MES. Request real output samples before selecting the interface.
Specify the minimum event record
One attempt should create one durable event. Capture an event ID; unit or lot ID; part and drawing revision; work order; tester, station and fixture IDs; recipe ID and revision; start, end and receive times; overall verdict; item-level results; attempt number; and operator or station identity. Where the instrument exports values, retain the value, unit, limit and pin pair. Where it exports only a verdict, mark the unavailable fields as unavailable rather than inventing measurements.
| Field group | Examples | Design rule |
|---|---|---|
| Identity | event ID, serial ID, work order | Reuse the same event ID on retransmission |
| Configuration | part, drawing revision, recipe revision | Snapshot the version at test time |
| Equipment | tester, fixture, station | Distinguish fixture replacement from tester replacement |
| Time | started, ended, received | Store time zone and clock synchronization state |
| Result | verdict, item, value, unit, limit | Distinguish failed, skipped and not applicable |
| Retest | attempt, prior event, reason, approval | Never erase the first failure |
| Provenance | operator, source hash, interface version | Keep changes auditable |
This is an implementation model, not a statutory minimum. Adjust it to contractual evidence and retention needs. The test time and the server receive time must remain distinct. During a network outage, local buffering and replay with the same event ID should prevent duplicate counts. Set buffer capacity, disk-full behavior, retry interval and acceptable clock drift. Only display “record saved” after the server has durably acknowledged receipt.
Keep the first failure when a harness passes a retest
Consider a unit that fails continuity, is re-seated and then passes. Replacing its FAIL with PASS hides the failure mode and the rework. Append a new event for each attempt and link the events to a repair record. Keep the final shipping disposition separate from the electrical measurements. Define who may approve a retest, how many attempts are allowed, when repeated faults require quarantine and what happens after a fixture change.
Report first-pass yield and final-pass yield separately, with the denominator stated. A rising final-pass rate accompanied by more retests is not necessarily process improvement. Also check the fixture before concluding that a displayed bad pin is physically defective: location information depends on the recipe’s net map and fixture wiring. Worn contacts can generate apparent product failures.
Choose a tester–MES interface with failure behavior in mind
A file or CSV adapter is often feasible for an existing tester. It must handle partially written files, reused filenames, character encoding, clock drift, changes in column order and archive of the original. An API or event interface can provide a tighter handshake, but confirm whether it exports item measurements and recipe revision or only the final verdict. Direct reading from a vendor database requires explicit vendor support and change-control terms. Manual transcription belongs only in a time-limited pilot with reconciliation.
| Method | Suitable situation | FAT focus |
|---|---|---|
| CSV or file | Existing tester has stable output | Partial file, duplicates, missing files, revision change |
| API or events | Supported external interface exists | Authentication, acknowledgement, replay, field coverage |
| Vendor application | Multiple testers use one supported platform | Licensing, upgrades, export rights |
| Manual entry | Short pilot | Comparison with source, error controls, exit date |
Ask for native samples covering PASS, open circuit, short, miswire, cancelled test, communication loss, fixture fault and retest. “MES compatible” without these samples is too vague for a purchase decision. Map a skipped insulation test to “not performed” or “not applicable,” never to “pass.”
Control recipe changes and mixed revisions
Maintain an approved association among work order, part revision, recipe revision and fixture. At the station, verify what program the tester actually ran, not just what a terminal displayed. A program number can be reused, so a file hash, approval number or effective date may be needed. Keep historical test events bound to their original recipe. If engineering later re-evaluates old data, record a separate approved disposition with scope and reason.
Before release, test the recipe with known-good assemblies and challenge samples: open circuit, adjacent-pin short, swapped pins, rising connection resistance, missing fixture and any component-specific fault. Verify test voltage and current where embedded electronics are present. If insulation or hipot is required, specify it independently from low-voltage continuity, including safety controls.
During a drawing transition, old materials, fixtures and programs may remain in the same factory. Treat a mixed combination as a block unless an approved deviation states the affected lots, period and approver. After a successful run, inspect the saved event: a current screen with an obsolete recipe version in the record is not acceptable evidence.
Write an RFP that can be tested
Provide bidders with representative part numbers, pin counts, production volume, shift pattern, cycle-time target, installed tester models and software versions. List the fault modes to detect and who owns the limits. Define the event fields, search and export needs, retention policy, backups, access control, offline operation and recovery. State who builds fixtures, approves recipes and maintains the MES adapter.
Useful RFP questions include: Can the supplier provide real native outputs for each result and error case? Which measurement values, limits, pin pairs and versions can be exported? Can wrong part, duplicate serial, expired recipe and unapproved fixture block the start? How many results can be stored offline, and how are they replayed? Who notifies the plant of a schema change after a software upgrade? How are original records and amendments retained? What fixture verification and spare-parts support is included?
Compare full implementation cost, including fixtures, recipe migration, integration, cutover downtime, annual support and future part additions. There is no credible universal price per station. A quote based on actual output samples and two or three representative recipes is more useful. For the wider architecture and cost drivers, see our traceability system cost guide.
Use FAT and SAT to prove the evidence chain
At factory acceptance testing (FAT), run expected-good and deliberately faulty samples. Include wrong part and serial combinations, duplicate event delivery, outage with several tests queued, clock drift, interrupted runs, retests, near-limit values, missing fixture and recipe mismatch. Compare four views: the operator screen, tester source record, MES event and shipping gate. Agree the expected outcome for each case before the demonstration.
Site acceptance testing (SAT) repeats the workflow with actual Thai factory equipment, network, operators, shifts and products. Check barcode readability, fixture attachment, rework return, power restoration and support coverage. A successful FAT cannot establish that the site clock, fixture wiring and local workflow are correct.
Use measurable acceptance criteria: no missing or duplicated event in a defined sample set; replay does not increase counts; each challenge fault is stored with the expected code; first FAIL and later PASS remain searchable; an obsolete recipe cannot start; and an authorised user can retrieve a complete serial history within an agreed time. Set quantities and times from plant and customer needs, rather than treating illustrative values as universal contract terms.
Pilot, search and operate the system
Start with one representative part and one tester, then add variants, fixtures and shifts. For an agreed overlap period, reconcile paper or legacy results against the new system by unit, time, verdict, attempt count and shipping decision. Investigate every difference as a possible interface, business-rule or fixture problem rather than simply correcting the spreadsheet.
Demonstrate retrieval with a passed unit, a repaired unit, a cancelled test, a fixture change and a drawing transition. The search view should connect the chronological events, item results, recipe, fixture, work order and release decision. If screen and CSV counts differ, inspect filters, time zones and cancelled-event handling. A customer-facing report may omit internal details, but the source events and report generation time should remain available to authorised staff.
Define how production works if the MES is unavailable. Quality owners decide whether offline testing is permitted for each product and whether finished units stay on hold. If allowed, preserve native tester records, identify units locally, protect untransmitted data and reconcile event ID to unit ID after restoration. Equal total counts do not prove that results were attached to the right serials. Track first-pass yield, retest rate, fault modes, fixture-related failures, recipe mismatch stops and delayed receipts by part, revision, line and shift. These measures become useful only after master data and missing records are controlled.
Frequently asked questions
Does a continuity PASS replace an insulation test record?
No. The purposes and conditions differ. If the product specification calls for insulation resistance or hipot, retain that test’s conditions and result separately. “Not run” must not be presented as “pass.”
Must a harness tester–MES link retain every measured value?
Decide from failure-analysis, customer evidence, instrument capability, data volume and retention cost. If values are available, item-level values and limits support later analysis. Never manufacture a value that the installed instrument does not export.
May we delete an initial FAIL after a PASS retest?
No. Keep both attempts, the reason for repair and the approval governing final disposition. Separate first-pass and final-pass reporting.
Can a CSV-only tester be integrated?
Often, if its output can be reliably matched to the scanned unit and recipe. FAT must challenge partial files, lost files, duplicates, clock drift and recovery.
Is crimp quality evidence the same as continuity evidence?
No. They describe different process steps. Link them by unit ID but validate each required quality characteristic on its own terms.
Summary
A useful wire harness continuity record joins unit identity, drawing and recipe revision, fixture, test conditions, item results and every attempt. Procurement should begin with real tester output and fault samples. FAT and SAT should prove that records survive retries, version changes and factory outages without losing or misidentifying a harness. Preserve failures and approvals so the resulting history supports both customer evidence and process improvement.
If your Thailand plant is assessing how to connect an existing harness tester to MES, contact TOMAS TECH with the tester model, a representative result file and the unit identifier you need to trace. We can discuss requirements while the RFP or pilot is still being defined.
