Sourcing
How to Audit a Fiber Optic Cable Factory: Checklist
Two suppliers answer your RFQ on the same afternoon. Both attach ISO 9001 certificates, both quote within three percent of each other, both promise eight weeks. Knowing how to audit a fiber optic cable factory is what separates them: one of the two extrudes the jacket you are paying for, and the other buys finished cable from a plant two provinces away and reels it under its own label.
Nothing in that paperwork distinguishes those two cases. An audit is a procedure rather than an instinct, and it runs on two tracks — a remote track you can start this week without booking a flight, and an on-site track you run once the numbers justify the trip. It ends with a score and a written verdict, not a feeling.
An Audit Is Evidence Collection, Not a Factory Tour
An audit is a bounded evidence-collection exercise run against a written standard. It is not a tour, and it is not a relationship-building visit with a lunch at the end of it. Before you book anything, decide what the audit has to settle:
- Can this supplier make my product to my specification, repeatedly, at the volume I need?
- Can it prove that with records that trace back to raw material, reel by reel?
- If a cable fails in the field in eighteen months, will I be able to reconstruct why?
An audit also has a fair-play side that buyers forget. A factory may legitimately decline to photograph a customer's mold, to name its other clients, or to open a room covered by an NDA, and it may reasonably ask you to sign its own confidentiality terms before showing you somebody else's design. What it cannot do is refuse to say why something is closed to you. Refusals with a reason are ordinary practice; refusals without one are findings.
A certificate states that a management system was audited on a given date, at a given address, within a stated scope — nothing more. Certification bodies audit systems, not shipments, and a system can be certified while a specific lot is quietly wrong.
The stakes are asymmetric, which is why the effort is worth it. The Uptime Institute's Annual Outage Analysis 2024 found that 54 percent of operators said their most recent significant outage cost more than \$100,000. Power and human error dominate that list of causes, but a cable that was never tested the way you assumed is a failure you cannot detect until it reaches the field. Set against one of those events, a week of document review and two days on site is cheap insurance.
The two tracks are not alternatives. The remote track is the filter and the on-site track is the confirmation. Run them in that order and you will spend your travel budget on the suppliers that deserve it. If nobody on your team has run an audit before, the questionnaire, the document list and the station route below can also be handed to an independent inspection firm — the scorecard is the deliverable either way.
This article covers the audit step itself — how to audit a fiber optic cable factory without confusing the plant with the paperwork. If you are still building the shortlist and weighing candidates against each other, the wider guide on choosing a fiber optic cable manufacturer is the right starting point.
How to Audit a Fiber Optic Cable Factory Remotely
A remote factory audit is a document-and-camera audit run before anyone boards a plane. It cannot prove process discipline — nobody reads a tension gauge over a video call — but it can disprove claims at almost no cost, and it usually decides whether the trip is worth funding. Four steps, in this order. Together they settle most of what a plant audit can tell you, without leaving your desk.
Step 1: Send a Written Pre-Audit Questionnaire
Send it as a spreadsheet or a form, with a reply deadline, and ask for evidence references rather than yes/no answers. The useful questions are structural:
- Which product families do you manufacture in-house, and which do you buy in or subcontract?
- How many production lines of each type do you operate, and what is each line's rated speed?
- Which standards do you test to (IEC, TIA, or a customer specification), and which test house issues the type-test reports?
- What are your calibration intervals, and who calibrates the OTDR and the extrusion gauges?
- What was your rework and non-conformance rate over the last twelve months?
Question 1 is the one suppliers answer vaguely. A manufacturer that subcontracts extrusion is not automatically disqualified, but you need to know before you audit the wrong building.
Step 2: Request the Documents That Are Hard to Fake
Ask for these at least two weeks before any visit. Discrepancies here can save you the trip entirely.
Document | What you are checking | A bad answer looks like |
|---|---|---|
Type-test report to the IEC 60794-1-1 generic specification and the IEC 60794-1-2 test-method series | An independent, accredited lab issued it; test parameters and pass/fail criteria are stated; the tested sample is identified | An "internal test report"; no lab accreditation reference; no sample identification |
In-process test records, last three months | Pass rates move; failures exist and are recorded | 100 percent pass, zero rework, no failures recorded |
Non-conformance reports and corrective actions | Root cause, action, verification date | No NCR log exists at all |
Reel-level traceability pack | Fiber lot, yarn lot, jacket compound lot, OTDR trace and reel serial all link to one reel | Serial numbers assigned after production; traces without bidirectional data |
Calibration certificates for OTDR, OLTS, power meter, diameter gauges | Dates in range, calibrating lab traceable | Equipment described as "just calibrated" with nothing attached |
Business license and export registration | Legal entity name and address match the entity on the quotation | Quotation from a trading company, production elsewhere |
The traceability pack is the single most revealing request. A manufacturer that can hand you a complete chain from fiber lot to reel serial has a functioning quality system whether or not its certificate is framed nicely. One that has to assemble the chain after you ask has told you something.
Step 3: Run the Video Walkthrough on Your Agenda
Remote work has one advantage over a plant visit: you can ask for the same thing twice, a week apart, and compare the answers. Use it. Do not accept a recorded factory tour — ask for a live call with a named engineer walking, and send the route before the call. Keep the camera on the floor, not on a conference table.
Ask to see the fiber store and its hygrometer reading; the pay-off tension display on a running line; the incoming inspection bench with a spool chosen by you at random; the reel yard where finished goods sit. If a station cannot be shown live, note it as unverified rather than assuming it exists.
One rule saves most remote audits: ask the operator a question directly, not the sales engineer. "What do you do if the diameter gauge drifts past tolerance?" takes ten seconds and tells you whether the person running the machine has been trained on it.
Step 4: Verify Every Certificate Against the Issuing Body
A PDF proves that somebody can produce a PDF. Every claim on a supplier's certificate slide has a register behind it, and checking five of them takes an hour.
Claim | Where to verify | What has to match |
|---|---|---|
UL listing (OFNP / OFNR jackets) | File number, company name, and the exact product category — not a similar one | |
CPR class (B2ca, Cca, Dca, Eca) | The manufacturer's declaration of performance, with the requirements set out by the European Commission | Class, AVCP system, notified body number, and the product family it covers |
ISO 9001 or TL 9000 | The issuing certification body's own register | Scope wording and site address — a certificate covering a sales office is not a factory certificate |
An "ISO/IEC 17025 accredited" test report | ILAC accreditation lookup | The lab that issued the report is accredited, and its scope covers the test performed |
Anatel, RoHS, REACH | The relevant national register or the certificate's own issuing authority | Holder name, model or family, validity |
Two habits make this step pay. First, check the site address, not just the company name — multi-site manufacturers hold certificates per site, and the site on the quotation is often not the site on the certificate. Second, compare the certificate scope against what you are buying. A TL 9000 or UL scope that excludes the product family you need is worth nothing on your order. For a worked example of the UL side of this, see how to verify a fiber cable supplier's UL listing.
Figure: the remote audit in practice — certificates, lot records and test traces are checked against the issuing body's register before anyone travels.
If the remote track leaves the main claims standing, book the visit. If it produces two or more critical findings, you have your answer without leaving the office.
Walk the Line: What to Check at Each Production Station
On site, sequence matters more than thoroughness. Walk the production flow in order — incoming materials first, finished goods last — because a station you visit out of sequence has had time to be rearranged for you. The route below is the same one used to make the cable itself, so it doubles as a check that the plant actually performs the steps it claims. It is also where most buyers stop treating how to audit a fiber optic cable factory as a question about equipment lists and start treating it as one about process control.
Figure: an on-site audit starts outside — confirming the manufacturer's sign sits on the building at the address given on the certificates, before you walk the line inside.
Station 1: Incoming Fiber and Materials
Ask to see the fiber store first, and check the environment rather than the brochure. Optical fiber belongs in a temperature- and humidity-controlled room; a working hygrometer on the wall is the cheapest proof that the room is managed. Pick a spool yourself, read its lot number, and ask for that lot's incoming inspection record immediately — not by email next week. Then look at the yarn and jacket compound certificates on file. "High-strength aramid" without a grade or a lot number is a label, not evidence.
Station 2: Coloring, Buffering, and Stranding
Watch a running line and write down three numbers: line speed against the ink manufacturer's recommended range, the pay-off tension reading during a spool changeover, and the excess fiber length for a loose-tube or stranded design. Tension that is set once at shift start and never monitored is the leading cause of latent fiber breaks — the cable passes final test and fails two years later. At the stranding bench, confirm that the interstices between tubes carry water-blocking yarn or swellable tape. An air gap inside the core is a path for water migration along the cable, and it will not show up in a short sample test. Then watch the bend radius the cable is actually held to: trays, sheaves and pay-off paths that kink the cable or coil it below its minimum bend radius leave micro-cracks that surface only in service, so a line that ignores its own bend limits is a latent-failure tell. For the full process chain these stations belong to, see how fiber optic cable is made.
Station 3: Jacket Extrusion
Extrusion is where a plant shows whether it controls a process or reacts to one. Ask for the barrel temperature profile and compare it with the compound supplier's data sheet; over-temperature degrades the filler in a fire-rated sheath, under-temperature gives poor dispersion. Check the cooling trough length against line speed, because a trough that is too short for the speed produces inconsistent jacket crystallinity. Then look at the diameter gauge: a laser micrometer should be logging continuously at two or more points, and the operator should be able to show you today's data with tolerance limits marked.
Look at the print legend on the finished jacket as well. A legible, repeating legend — manufacturer, cable type, fiber count, production year — is how your installer identifies a cable mid-span years from now, and it is the base of the color-code and identification practice covered by the TIA fiber standards. A plant that will not print a legend is telling you its cable is a commodity it never expects to see again.
Station 4: Assembly, Termination, and End-Face Quality
Most audits stop at the extruder. If you buy pre-terminated assemblies — patch cords, MPO/MTP trunks, FTTA waterproof assemblies — the value is added here and so are most of the field failures.
Ask three questions and expect physical evidence for each. First, end-face: is there interferometry or at least documented end-face inspection to a defined criterion, per polish grade, with records kept by production batch? Second, loss: what insertion loss and return loss limits apply to a mounted connector, and how many pieces per lot are measured — one sample out of a hundred is a claim, not a control. Third, sealing, for outdoor and FTTA assemblies: what immersion or pressure test is run on a waterproof assembly, and can the operator show you the fixture?
Walk the line with your eyes open for the cheap tells: assemblies on an open bench next to the extruder, dust caps piling up in one bin, adapters being re-used, and no separation between polished parts and the rest of the floor.
Station 5: Final Test, Reel Packing, and Sampling
Finish at the test bench and watch a complete test. For cable, that means bidirectional OTDR at both 1310 nm and 1550 nm, with launch and receive fibers long enough to push the near-end dead zone off the first splice and the far end. An operator who tests without a launch fiber is leaving the first hundred meters of every reel effectively untested, and the trace will look clean because the instrument cannot see it.
Then check the sampling posture rather than accepting "we test everything". If the supplier works to an acceptance-sampling scheme — the AQL tables most cable plants follow, standardized as ISO 2859-1 and ANSI/ASQ Z1.4 and explained from first principles in NIST's handbook section on lot acceptance — ask which standard, which inspection level, and which AQL value applies to which defect class: critical, major, minor. A supplier that cannot name its sampling plan is either testing everything or testing nothing, and it does not always know which. Ask how first-article approval works as well. The first article off a new construction — the piece your own engineer may witness being made and tested, which is the factory acceptance test (FAT) in its buyer-facing form — is where a design problem is cheapest to catch, and a supplier who cannot describe those steps is validating on your first order.
Before you leave, walk the reel yard. Cable stored outdoors without weather protection, reels that have been dropped from truck height, and forklift tines in contact with the jacket are all recorded defects, not housekeeping details.
Cross-Check Capacity Claims Against What You Can See
"800 km a day" is the hardest claim on any supplier questionnaire to dispute and the easiest to convert into arithmetic. Stop treating it as a headline and turn it into lines.
lines × line speed (m/min) × 60 × productive hours × yield = meters per day
A plant claiming 800 km per day of cable is describing roughly eight extrusion lines running two eight-hour shifts at 100 m/min, or fewer lines running around the clock. Both are possible; both imply something you can check on the floor. If your order mixes fiber counts or long reel lengths, ask which counts run without a changeover and whether your reel lengths come off one continuous run or are built up from shorter lengths — fiber count and reel length are the two variables that quietly set your unit cost. Count the lines. Ask what share of them were running on the day you visited. Ask the rated speed of the slowest one, because the slowest line sets your realistic delivery date.
Claim | What you can observe | What it implies |
|---|---|---|
"800 km/day" of cable | Number of extrusion lines, their rated speeds, shifts run that day | Whether the claim requires a plant bigger than the one in front of you |
"50,000 ends/day" of assemblies | Polishing and crimping stations, how many are staffed | Whether output depends on seasonal labor |
"Capacity for your volume" | Current order book, work-in-progress on the floor | Whether your order fits this quarter or the next one |
"Two weeks lead time" | Reels of your construction type in the yard, fiber in stock | Whether the lead time includes somebody else's production slot |
Two cautions. Daily nameplate capacity is not utilization — a figure can be accurate and irrelevant if the line it describes is booked out. And capacity is usually quoted for the plant as a whole, while the product you are buying may run on one line of six. Ask for the split by product family, in writing. DYS Fiber Optic, for example, publishes 800 km/day of optical cable across three manufacturing sites; the number is only meaningful once a buyer knows which site and which line their own order would occupy.
Log the Red Flags and Rate Their Severity
Collect findings continuously during the visit — photographs with timestamps, plus the lot or reel number each one relates to — and rate each one. A severity rating is what stops a serious issue from being buried under a list of housekeeping notes.
Finding | Why it matters | Severity |
|---|---|---|
Quotation issued by a trading entity; production at an unnamed plant | You cannot audit a building you have not been shown, and liability and origin become unclear | Critical |
Fiber lot on the floor differs from the lot on the certificate | Substitution is the classic failure mode, and it is invisible in final test | Critical |
No non-conformance log, or a claim of zero defects | Process control is not being measured, so it is not being managed | Critical |
Calibration expired on the OTDR or a measuring gauge | Every data point produced since expiry is suspect | Critical |
Jacket legend missing, illegible, or missing the fiber count | Traceability breaks in the field and at the splice closure | Major |
Yarn or compound certificates without grade or lot number | Strength and fire performance cannot be tied to a shipment | Major |
Reels stored outdoors unprotected, or visibly dropped | Latent mechanical damage that shows up as attenuation later | Major |
Social compliance questions refused | Labor practice exposure sits with your brand, not your supplier's | Major |
Tooling and mold ownership undocumented in an OEM or ODM program | Your tooling may be the supplier's asset, and your design may be sellable to your competitor | Major |
A second site or a non-China origin not disclosed before the visit | Origin and tariff exposure appear after the order, not before | Major |
Assembly area not separated from the cable floor | Contamination and handling damage at the end-face stage | Minor |
Photographs refused in a specific area | Usually a process the supplier does not want recorded | Note it |
Two critical findings at the same station mean you stop the audit there and finish the conversation in a meeting room. There is no value in completing a walkthrough of a plant you are not going to buy from.
Score the Audit Before You Negotiate
A written score turns a bag of observations into a decision your management can review. Weight the categories to match what you actually buy, then score each from 1 to 5 against the anchors you wrote into the checklist before the visit.
Category | Weight | A 5 looks like | A 3 looks like |
|---|---|---|---|
Documentation and traceability | 25% | Full reel-level chain, produced as a matter of course | Records exist but have to be assembled on request |
Process control and calibration | 20% | Continuous monitoring, in-range calibration, data retained | Gauges present, data available on request |
Test capability and evidence | 20% | Bidirectional OTDR and end-face records per batch | Test equipment present, record-keeping inconsistent |
Capacity and scheduling realism | 15% | Line-level capacity data that matches the floor | Plant-level figures only |
Certification and market access | 10% | Certificates verified at the site address, scope covers your product | Certificates valid, scope ambiguous |
Quality system and CAPA behavior | 10% | Root-cause analysis with verification dates | Corrective actions closed without verification |
Set the thresholds before the visit, because thresholds invented afterwards always bend toward the supplier you prefer. A workable default: 4.0 or above releases the order; 3.0 to 3.9 releases it conditionally, against corrective actions with dated verification; below 3.0 you do not award. If your product mix is assembly-heavy, move weight out of "capacity" and into "test capability" — for patch cords and MPO trunks, end-face and loss discipline is the whole risk. How much a certification is worth in that weighting depends on which system you are looking at; the practical differences between ISO 9001 and TL 9000 are worth understanding before you assign the points.
Close the Loop: Exit Meeting, CAPA, and Re-Audit
The last stage of a factory audit is not the report — it is what the factory does with the report. An audit that changes nothing on the floor was a visit. Two pieces of paperwork make it count.
The exit meeting, held on the same day, is where you read your findings back and separate fact from interpretation. Say what you observed; ask the supplier to correct anything you got wrong. Findings the supplier agrees to on the day are the ones that get fixed. Findings delivered as a surprise two weeks later get argued about instead.
The audit report, issued within five working days, should contain: scope and method, the evidence reviewed, the non-conformance log with severity ratings and photographs, corrective action requests with a named owner and a verification date, and the score. Keep the report factual in tone. "Diameter gauge calibration expired 2026-03-14; data from 2026-03-15 onward cannot be relied upon" carries more weight than any opinion about the plant's attitude.
Then enroll the supplier in a schedule rather than treating the audit as a one-off. An annual on-site audit plus a remote review each time a new product family is introduced, a line is added, or a site changes, is proportionate for most buyers. Where one supplier serves several markets, the surveillance calendar should also track which certificate has to stay valid for each market you sell into — a lapse in one scheme can stop a shipment into one country without affecting the others.
Re-audit is not a punishment. Buyers who never come back get the plant that was built for the buyers who do.
Decide: Release, Condition, or Walk Away
Three outcomes, and you should be willing to say all three out loud at the end of the visit.
Release. No critical findings, score at or above your threshold, evidence for every main claim. Place the order and diary the next surveillance audit.
Condition. Critical or major findings that are fixable and that the supplier accepts on the day. Award the order against a corrective action plan with dated verification, and hold first-article approval until the plan closes. Do not release a second order on the same score.
Walk away. Fiber substitution, an undocumented production site, a refusal to provide lot traceability, or expired calibration with no corrective plan. These are not maturity gaps; they are the failure modes the audit exists to catch. If the supplier's producing site is not the site you quoted from, and a second origin is part of the plan, work out which site's cost and lead time you are actually buying before you release anything — auditing the plant that will not run your order is the most expensive mistake in this whole process.
If you want to run this checklist against a plant rather than a brochure, note that DYS Fiber Optic builds cable and pre-terminated assemblies at sites in Huizhou, Mianyang and Hanoi, so we sit on the receiving side of exactly this table. Test reports and traceability records can be provided on request — tell us what your audit needs and we will confirm what can be released before a visit. Bring the scorecard with you.