What an MPO line actually needs
MPO assembly needs equipment that duplex assembly does not. On a tour, these are the stations worth asking to see:
| Station | Why MPO needs it |
|---|---|
| Ribbon fusion splicing | Twelve fibers spliced as a ribbon. This is where polarity and sequence errors originate |
| Multi-fiber polishing | Programmable polishing across a rectangular ferrule, not a cylindrical one. Twelve fibers have to finish coplanar |
| Interferometric end-face measurement | Radius, angle and fiber height measured across the whole ferrule, per position |
| Multi-channel optical switching | Required to test twelve cores individually rather than measuring the cord as one path |
| Environmental chamber | Temperature cycling on finished assemblies, where epoxy and jacket problems appear |
A line running duplex patch cords has the first two in a different form and usually lacks the last three.
Five questions for a video factory tour
These are answerable on a call, and the answers separate an operating line from a sales office:
- “Show me the polishing station and the fixture you use for a 12-fiber ferrule.” A multi-fiber fixture looks nothing like a duplex one. Someone reselling cannot show it.
- “When a customer does not specify a breakout ratio, what do you ship?” A real line has a documented default — for example, splitting a 12-fiber trunk into six duplex legs. No default means no standardised process.
- “How do you measure the twelve cores individually?” The answer should describe a channel switch or an equivalent method. “We test the cable” is a per-cord measurement, which is not the same thing.
- “What is your ribbon splicing equipment and who operates it?” Named process, named station.
- “Can you pull the test record for a batch you shipped last month?” Retrieval speed reveals whether records are systematic or assembled on request.
How to Read an MPO Test Report
A test report is the only quality evidence a buyer gets before the goods arrive. Most disputes trace back to a report that looked fine and said less than it appeared to.
The fields a complete report carries
| Field | それがあなたに伝えること |
|---|---|
| 挿入損失, per core | Twelve rows for a 12-fiber cord, not one figure |
| リターンロス, per core | Same granularity |
| 試験波長 | 1310/1550 nm single-mode, 850/1300 nm multimode |
| Test method | IEC 61300-3-4 挿入方法B, or the method used |
| Reference cord grade | The quality of the reference jumper sets the floor for what can be measured |
| 極性とファイバー配列 | Verified separately. This does not appear in a loss measurement |
| End-face inspection result | Per position, コードごとではありません |
| 長さ | As-built against ordered |
| Serial or batch identifier | Links the report to the physical cable you received |
| Test date | Establishes when, and lets you check it against the ship date |
What a thin report leaves out
Compare an incoming report against the list above and note what is missing. Each omission hides a specific risk:
| Omitted | What it can hide |
|---|---|
| Per-core breakdown, replaced by one figure per cord | One core out of twelve failing while the cord average still looks acceptable |
| 試験波長 | Measurement at the favourable wavelength only |
| Test method | A method that produces lower numbers than the one your own field tester uses |
| Reference cord grade | A poor reference cord makes a mediocre assembly measure well |
| Polarity verification | The most expensive field failure, and it is not implied by any loss figure |
| Batch or serial identifier | A report that cannot be tied to the cable in your hand |
| Test date | A report reused across batches |
How to ask for what is missing
Send this before the order rather than after delivery:
“Please confirm the factory test report will include per-core insertion loss and return loss for every fiber position, the test wavelength and method, the reference cord grade, polarity and fiber sequence verification, end-face inspection per position, and a batch identifier tied to the shipment. If any of these are not part of your standard report, tell us which ones so we can agree scope before production.”
The response is itself a data point. A supplier that tests this way answers in one line. A supplier that does not will either negotiate the scope or go quiet.
What MPO Requires That Duplex Doesn’t
This is where most supplier evaluations go wrong. Buyers apply a duplex patch cord checklist to a twelve-fiber assembly, and four things fall through.
Per-core, not per-cord
A duplex cord has one fiber per connector, so one measurement describes it. A 12-fiber MPO has twelve independent optical paths sharing one ferrule.
If a supplier reports one insertion loss figure per cord, ask what it represents. An average across twelve cores can sit comfortably inside specification while one core sits well outside it — and in a parallel link, that one core throttles the whole channel, because all lanes have to run simultaneously.
Per-core measurement needs a way to address each fiber position in turn. In practice that means an optical switch between the meter and the connector, stepping through the positions.
Lane-to-lane consistency
Beyond each core passing individually, the spread between the best and worst core matters. Twelve cores at 0.30 dB behave differently from eleven cores at 0.10 dB and one at 0.34 dB, even though both sets pass a 0.35 dB limit.
Parallel optics run every lane at once, so the worst lane sets the channel margin. Ask whether the supplier tracks the spread across positions and whether there is an internal limit on it. Many suppliers do not track this at all — the question tells you how the line thinks about parallel links, regardless of the answer.
Polarity and gender verification
No insertion loss measurement detects a polarity error. A cord built to Type A instead of Type B measures perfectly on every core and fails as a link. Verification is a separate mapping check, and it needs to be a named line item on the report.
Gender is visual and takes a second to confirm, but confirm it against the ports at both ends rather than against the last order. Getting either wrong means re-ordering, since neither is field-adjustable. ザMPO to MPO configuration guide covers how polarity and gender combine across a full channel.
End-face geometry across twelve positions
A duplex ferrule has one fiber to keep within tolerance. An MT ferrule has up to twelve, and they have to be coplanar enough that all of them make contact under spring force at the same time.
Interferometric measurement reports radius of curvature, 角度, fiber height and the height difference between adjacent fibers. That last parameter is the one to look for: a fiber sitting a few hundred nanometres low will underperform on every mating, and it does not show up in an average.
Typical limits for a 12-fiber MPO assembly:
| パラメータ | 限界 |
|---|---|
| 曲率半径, バツ | ≥ 2000 んん |
| 曲率半径, Y | ≥ 50 んん |
| APC angle | 7.85°～8.15° |
| 繊維の高さ | 1000 に 3500 nm |
| 隣接するファイバーの高さの差 | ±300nm |
Parameter definitions and how these feed into link performance are covered in the MPO パッチコード選択ガイド.
Quality System Signals Worth Checking
Certificates are worth asking for, and they are also the easiest thing to gesture at without substance. These signals are harder to fake and easier for a buyer to verify.
Inspection coverage, stated precisely. “We inspect every cable” と “we test every core of every cable” describe different operations. Ask which one, and ask what the sampling ratio is for anything not tested at 100%.
A separate quality area. A line where testing happens at the assembly bench has a different failure profile from one with a dedicated measurement area and controlled conditions.
Batch-level environmental testing. Temperature and humidity cycling on finished assemblies catches epoxy and jacket problems that an ambient bench test never sees. Ask whether it runs per batch or only during qualification.
In-house flame testing. A supplier claiming LSZH, OFNR or OFNP has to substantiate it somehow. In-house burn testing means the claim gets checked internally rather than only inherited from a material datasheet.
認証, and how it is presented. Ask for the certificate itself: issuing body, scope, the legal entity named, and the expiry date. Scope matters most — a certificate covering one product family does not extend to another, and one naming a different entity does not transfer.
On this point, our own position: material reports are available on request, and certification documentation is issued per order and destination market rather than published on the site.
If certification evidence up front is a requirement for your procurement process, ask us at enquiry stage and we will tell you what we can supply for your market before you commit to anything.
Lead Time, MOQ and First-Order Risk Control
What to expect
| 典型的 | |
|---|---|
| Standard production, a few thousand cords | 7 に 12 日々 |
| Volume production, above 20,000 個 | 15 に 20 日々 |
| サンプル | Free of charge, buyer pays freight, deducted from a subsequent bulk order |
| MOQ | No minimum where cable is in stock; custom configurations confirmed per specification |
| Custom lengths | 0.2 メートルから 300 m |
| Custom breakout ratios | Any leg count; default is an even split unless specified |
| OEM and private labelling | 利用可能; custom packaging carries its own minimum |
Peak-season and heavily customised projects run longer than the figures above, so confirm dates against your own installation schedule rather than treating a published range as a commitment.
Sizing a first order
The purpose of a first order is to find out whether the supplier’s process works before you have money at risk. Three principles, no fixed quantity:
Qualify with samples first. Order the exact configuration you intend to buy — same fiber count, 極性, gender, grade and jacket. A sample in a different configuration tells you the supplier can make something, not that they can make yours.
Keep the first production order inside what you can absorb. Size it so that if the batch is wrong, you can still deploy it elsewhere, return it, or write it off without the project stopping. That figure depends on your business, not on the supplier’s minimum.
Move the inspection point earlier. Most first-order disputes come from checking at delivery instead of before shipment.
Inspection checkpoints
| ステージ | What to require | Why here |
|---|---|---|
| Before production | Written spec confirmation: カウント, 極性, gender, 端面, grade, ジャケット, 長さ | Cheapest place to catch a mis-specification |
| First article | Test report on the first assemblies of the run | Catches a systematic build error before the full batch exists |
| 出荷前 | Full batch test report with per-core data and batch identifier | Last point where a rebuild costs nothing but time |
| On arrival | Spot-check against the report: 極性, gender, 端面, a sample of loss readings | Confirms the shipment matches the paperwork |
The arrival check is a verification, not a first inspection. If it is the only test in your process, every problem gets found at the most expensive point.
Send your configuration and quantity with your enquiry and we will quote against it, or ask for the test report format for your configuration before you order. Current standard configurations are listed on theMPO/MTP patch cord page.
When Something Fails
Agree the path before you need it. Four questions, answered in writing at quotation stage:
What counts as a failure? Out-of-specification loss, polarity mismatch, physical damage and count mismatch all get handled differently. Define the categories.
Who tests? A disagreement over whether a cord is out of specification usually comes down to different reference cords and different methods. Agree the method up front, referencing the same standard as the factory report.
What is the remedy? Replacement, credit or rework, and who pays freight in each case.
How far back does it reach? A batch identifier on the report lets both sides establish whether one cord or one batch is affected. Without it, every dispute becomes an argument about scope.
Fibconet assemblies carry a one-year warranty with service life well beyond that. For a failed unit, send the batch identifier from the test report with the fault description, and we will trace it against the production record for that batch.
よくある質問
How do I know if an MPO patch cord is good quality?
Ask for the per-core test report before shipment and check what it contains: insertion loss and return loss for every fiber position, test wavelength and method, reference cord grade, polarity and sequence verification, end-face inspection per position, and a batch identifier. A single loss figure for the whole cord does not tell you whether one core out of twelve is failing, and in a parallel link the worst core sets the channel margin.
Do MPO suppliers provide 100% factory test reports?
Some do, some sample. Ask specifically whether every core of every cable is tested or whether a percentage is sampled, since “we test every cable” can mean one measurement per cord rather than twelve. Fibconet assemblies are tested core by core for insertion loss, リターンロス, 端面, 極性, 繊維配列, 長さと見た目, with electronic reports issued after order placement.
What is the standard lead time for bulk MPO patch cord orders?
Standard production of a few thousand cords runs 7 に 12 日々, and orders above 20,000 pieces run 15 に 20 日々. Custom lengths, non-standard breakout ratios and custom packaging extend this. Peak season varies, so confirm dates against your installation schedule rather than relying on a published range.
What is the minimum order quantity for custom MPO patch cords?
Where cable is in stock there is no minimum. Custom configurations are confirmed per specification, and custom packaging carries its own minimum. Free samples are available for qualification with the buyer covering freight, and that cost is deducted from a subsequent bulk order.
How can I tell a real MPO factory from a trading company?
Ask to see the multi-fiber polishing fixture and the ribbon splicing station on a video call, ask what the line ships when a customer does not specify a breakout ratio, and ask them to retrieve the test record for a batch shipped last month. Equipment specific to multi-fiber assembly is hard to produce on request, a documented default indicates a standardised process, and retrieval speed shows whether records are systematic.
Should I require an ISO certificate from an MPO supplier?
Ask for the certificate itself rather than a statement, and check the issuing body, the legal entity named, the scope of products covered and the expiry date. Scope is where most certificate claims come apart, since a certificate covering one product family does not extend to another and one naming a different entity does not transfer to the company quoting you.
What should I check when the first shipment arrives?
Treat arrival as verification rather than first inspection. Spot-check polarity, gender and end-face condition against the factory report, and re-measure a sample of cores using the same method the report cites. Anything found at this stage should already have been caught before shipment, so a mismatch is as much a signal about the supplier’s process as about the batch.
Can I get a sample test report before placing an order?
Ask for the report format and field list for your specific configuration at enquiry stage, so you can confirm the scope of testing before committing. Fibconet issues per-core electronic reports after order placement and can confirm the report format for your configuration beforehand, though we do not publish sample reports publicly.
What does lane-to-lane consistency mean on an MPO cord?
It is the spread between the best and worst performing core on the same assembly. Twelve cores at 0.30 dB and eleven cores at 0.10 dB with one at 0.34 dB both pass a 0.35 dB limit, but they behave differently in a parallel link where every lane runs at once and the worst lane sets the margin. Ask whether the supplier tracks the spread across positions and whether an internal limit applies.
Who pays for a failed batch of MPO patch cords?
Agree this at quotation stage rather than after a failure. Define what counts as a failure, which test method settles a disagreement, whether the remedy is replacement, credit or rework, who pays freight in each case, and how a batch identifier establishes the scope of the problem. Without a batch identifier on the test report, every dispute turns into an argument about how many cords are affected.
