The PDF arrives attached to a quotation, usually late in the evening, usually with a decision hanging off the other end of it. A logo at the top, a table with ten or twelve rows, and a column of values in micrometers and decibels. Everything in it can be checked. Very little of it can be enforced.
That gap is worth more to a buyer than any single number on the page. A spec sheet exists so that a stranger can compare two products without telephoning either supplier, and it does that job well. It was not written to be the document you point at when a pallet lands and something is wrong. Those are two different documents with two different authors, and on most fiber purchases only one of them carries a signature.
So this is less a decoding problem than a jurisdiction problem. Before you use a number, you need three things about it: which paper it lives on, what it is defined against, and who is answerable for it.
A spec sheet is not one document
Ask a supplier to describe a fiber and you get one PDF. Ask three departments of the same supplier and you get three papers, and they are not drafts of each other. They are written by different people, for different reasons, and they settle different arguments.
| The paper | Who stands behind it | Its language | What it can settle |
| Fiber datasheet | The supplier’s engineering function. Unsigned. | µm, NA, dB, GPa | Whether two products look different. It is a comparison document. |
| Sterile pouch label | The manufacturer, as a legal statement | Symbols, lot number, use-by date, sterilization method | Whether this unit was made and released the way the manufacturer says. |
| Quotation, drawing and purchase order | Both parties | Part number, dimensions with tolerances, quantity, packaging, sampling plan | What you may reject. The only paper with two signatures on it. |
Three terms get treated as though they lived on one label: a numerical aperture, a sterility assurance level, and a length in meters. In practice they live on three different papers. NA is a property of the glass and belongs on the datasheet. A sterility assurance level is a statement about a sterilization process, validated against a load and a cycle; it belongs with the sterilization records, and if it appears in print at all it appears on the pouch. And the length is a system-fit decision — a fiber supplied at 3 m and a handpiece specified around 2.5 m are two different projects — so it belongs on the drawing, where it can be reconciled with the handpiece it has to reach. None of the three can be checked against the paper the other two are printed on.
Which is why the first move on a receiving dock is not to find the number. It is to recognize the paper. One manufacturer’s own wording makes the point better than any standard: what is written on the quotation is what arrives on the pallet. That is the right standard. The customization page is where a supplier states what it is able to commit to, and the quotation is where that commitment stops being marketing and starts being an obligation.
Checkable is not the same as enforceable
Take the most-quoted number on any fiber datasheet. The numerical aperture looks like a measurement. It is a definition.
NA is defined from the far-field intensity distribution of light leaving the fiber, at the half-angle where intensity has fallen to a stated threshold. The threshold is not the same for every class of fiber — the standard’s own framework fixes both a threshold and a specimen length for each — so two datasheets can both be accurate and disagree, because they were produced under different conventions. The specimen matters as much as the threshold: the classes this test was written around are measured on a long length of bare fiber, and a delivery fiber is three meters of glass with a connector bonded to each end.
Then there is the geometry, which turns out to be the part nobody budgets for. The far-field condition requires the detector to be at least d²/λ away from the emitting face, where d is the emitting diameter and λ the wavelength. At 850 nm that works out as roughly:
- 50 µm core — 3 mm of standoff
- 200 µm core — 47 mm
- 400 µm core — 188 mm
- 1000 µm core — 1.18 m
The requirement grows with the square of the core. It is a few millimeters for a telecom fiber, and more than a meter for a one-millimeter energy delivery fiber — which is the point at which the measurement stops being a bench measurement at all. A numerical aperture quoted for a large-core delivery fiber was therefore produced on a setup most suppliers do not own and no receiving dock will ever reproduce. That is not a criticism of the number. It is a statement about what kind of number it is.
Good datasheets handle this by pairing the NA with the thing it implies. A large-core silica cable datasheet in circulation reads 0.22 ± 0.02 (full acceptance angle 25°), alongside 0.12 ± 0.02 (14°) for the low-NA variant. Look at those two pairs closely, because they are the same number stated twice: 0.22 is the sine of about 12.7°, doubled it is 25.4°, and the sheet says 25°. The 0.22 and the 25° are not two specifications. They are one specification with a sanity check built in. Which also means that a sheet quoting a half-angle beside a NA from a different source can be wrong by a factor of two and look perfectly reasonable — 12.7° printed where 25° belongs raises no eyebrows at all.
Now the number that looks most concrete on the whole page: core diameter. It is also defined rather than observed. For multimode fiber, core diameter is defined from the refractive index profile at a stated fraction of the index step between core and cladding, not at a visible edge — a fiber does not have a core edge the way a pipe has a wall. The measurement standard lists four methods for fiber geometry, and the only one based on mechanical contact determines the cladding, not the core. So your caliper measures the jacket, a micrometer measures the cladding, and nothing on a receiving bench measures the core.
This is the quiet reason ISO 13485 §7.4.3 asks for purchased product to be verified against documented purchasing information rather than against a datasheet. If your order says “400 µm core” and names no method, there is no coherent way to declare a delivered fiber non-conforming, because the two of you are not holding the same definition. The fix is one line long: name the method and the reference in the drawing. That converts a comparison figure into an acceptance criterion.
One more pattern worth knowing before it costs you a batch. Some lines on a datasheet are not independent. Published high-power silica cable sheets give a proof test of 100 kpsi with polymer jackets and 70 kpsi with polyimide. Those are the same promise at two levels: 100 kpsi is 0.69 GPa, which is the minimum proof stress the A1 multimode product specification sets, and 70 kpsi is 0.48 GPa, which is not. The jacket line is not cosmetic. It carries a mechanical rating and a temperature range with it, and a purchase order that changes the jacket to solve a routing problem has silently changed a strength figure. The bare fiber range shows the same coupling from the other direction: the same core sizes are offered in a silica-polymer and a silica-polyimide construction, and those two columns do not carry the same limits.
Bend radius hides a version of the same trap. One published datasheet gives 50 × cladding diameter for short-term mechanical stress and 150 × core diameter for use at high laser power. Two sentences with identical shape, two different references, and no way to compare them unless you noticed which diameter each one is relative to.
Why any of these numbers behaves the way it does is a separate subject, and this site has a whole set of articles on the physics of delivery. What follows is narrower: which of them you can hold a supplier to.
The part number is a compressed spec sheet
Before the PDF, there is the part number, and it is usually the only line that appears on the quotation, the box label and the packing list at the same time.
| Item number | What it states | What it does not state |
| HS-272/420-3.0 | 272 µm core, 3.0 m | What 420 refers to, or whether either dimension carries a tolerance |
| HS-365/550-3.0 | 365 µm core, 3.0 m | Same. The first column is headed Core; the second number is under no heading at all. |
| HS-550/750-3.0 | 550 µm core, 3.0 m | Same |
The holmium fiber range is a good test case, because its three item numbers contain two diameters and a length, and only one of those three fields is explained by the table beside it. Read that table quickly and you will find 550 doing two jobs: it is the core size of one item and the second field of another. Both are legitimate. A buyer who is decoding by pattern rather than by key can enter the wrong catalogue number into an ERP system and reconcile it against a correct-looking line, because the number on the box will match the number on the order.
The second field is genuinely ambiguous in the market, and not by accident. Core-to-cladding ratio is a named parameter with a small set of standard values — ratios of roughly 1:1.04 through 1:1.4 cover most of the catalogue ranges you will meet — but a part number’s middle field may hold a cladding diameter, a ratio, or a coating diameter depending on which family it comes from. That is a convention, not a rule, which means it can only be decoded from the supplier’s own key.
Two habits follow. Ask for the key rather than inferring it, and write the long form into the purchase order even when the catalogue number is present. Orders that carry both the part number and the dimensions spelled out in words are the ones that survive a part-number revision at the factory, and those revisions are rarely announced.
The seven lines, and what makes each enforceable
Custom laser fibers vary along a small number of axes — core diameter, length, connector, tip geometry, jacket, handpiece body, packaging — and a receiving check is only ever as strong as the weakest of them is written down. The bare fiber size table is a compact illustration of the problem: each core size is listed against a connector option and a construction, and those three lines are precisely the ones that have to appear on the order, spelled out, before anyone can check them at the door.
| Line | What it usually says | What makes it enforceable |
| Core diameter | one number in µm | the measurement method and the definition behind it, or the number is open to argument |
| Length | 3 m, or a figure with a tolerance | whether it is glass length or assembly length — the connector ferrule and strain relief make those two different numbers |
| Connector | SMA-905, FC-PC, ST, Luer Lock | the mating half, the ferrule material, and whether the interface is air-gap or potted. “SMA” alone is not a specification. |
| Tip geometry | flat, ball, radial, side-firing, diffuser | the geometry plus its acceptance criterion. For an end face that criterion is a visual standard, not a number. |
| Jacket | a polymer name | the temperature range and proof test that follow from that polymer, both of which move when the jacket moves |
| Handpiece body | included, or not | whether it is in the carton, and whether it is in the price |
| Packaging | sterile pouch, or bulk | the barrier system, the sterilization method, the lot-coding rule, and the source of the label’s symbols |
Three of those can be checked with a tape, a scale and a camera. The rest can only be checked against a document.
There is a faster route through a spec sheet than reading all of it. Find the lines that are not allowed to be blank, and treat their absence as the finding.
Maximum power is the row everyone assumes belongs to the glass. It belongs to the build. One published high-power cable table offers the same core diameters across five termination grades, and the maximum beam power climbs through 5 W, 30 W, 150 W, 300 W and 900 W as the connector goes from a plain SMA to a liquid-cooled D80. The glass never changed; the ceiling moved by a factor of 180. That is why a power figure means nothing on its own — it is a ceiling for the assembly the sheet happens to describe, and it has to be read again after any change to the connector. A range sold under one rating across several core sizes, such as the holmium fiber range at 120 W from 272 to 550 µm, has already told you that the rating is not a per-core-size property.
Most of what the third column refers to can be obtained before you order anything. Our downloads library holds the fiber datasheets, the connector mating checklist, the OEM capability pack and the single-use instructions for use, and it is deliberately organized by document type rather than by product, because these are four different kinds of paper. One habit worth forming: a datasheet describes a component, while an IFU instructs on a device. Where the two disagree about a limit, the IFU is not a second opinion. It is the document that governs the finished assembly.