The email is five lines long. Every line of it is correct, and none of it is a specification.
Nothing in that sentence is wrong. Four hundred microns through an eighteen-gauge needle is the pairing that every large published series converged on. One millimeter of protrusion is exactly what the 58-patient 1470 nm series locked with a plastic set-screw disc. Sterile packed is what a percutaneous disc procedure requires. A supplier could accept that email, manufacture to every word of it, and deliver an assembly that vents half as well as the one the buyer believed they had ordered — with the certificate of conformance to prove it was made as asked.
The reason is that a specification runs in columns, and that email has one. It has nominals. It has no tolerances and no test methods. “400 µm” is a nominal. “470 µm ±15 µm on the coated diameter, measured on a stripped sample, method stated on the drawing” is a line in a specification. The distance between those two sentences is what this article is about.
It is written for the people who order the tray rather than the people who use it: engineers at 810–1470 nm console makers, product managers at spine distributors, procurement teams building a custom laser fiber line for a market where the annual volume is small and the technical barrier is not. A PLDD fiber is one of the few surgical disposables where the customer’s own protocol — his needle, his protrusion, his fixation collar — is part of the product definition, and where a hundred units a year can still justify a private-label assembly. It is a market in which the specification document matters more than the price list, and in which almost nobody has one.
Four things have to be on that document. A gauge that means something, a protrusion that can be held, a collar that does not do anything it was not asked to do, and a pack that survives the chemistry it is about to meet.
“18 gauge” describes the outside of the needle. The lumen is a minimum, and nothing caps it.
Start with the number everybody writes on the order and nobody questions. Gauge is a size designation for the needle’s outside diameter. It is not a hole size, and the standard is explicit about which one it controls.
Under ISO 9626, tubing size is designated by nominal outer diameter in millimeters, with the corresponding gauge. An 18 G needle is therefore defined as a tube of about 1.20 mm across the outside. The dimensional table then gives an outer-diameter band — in the current edition of the standard that band is 1.15 mm to 1.35 mm for the 1.20 mm designation — and a minimum inner diameter of 0.79 mm. There is no maximum inner diameter in the standard at all. Read that twice, because it is the load-bearing fact of this section: a needle tube is only guaranteed to have a lumen of at least 0.79 mm, and the wall that decides how much bigger it gets is not part of the gauge.
| Designated size | Nominal OD | OD band (current revision) | Min. ID (ISO) | Typical ID, regular wall | Typical ID, thin wall |
| 18 G | 1.20 mm | 1.15 – 1.35 mm | 0.79 mm | ≈ 0.80 mm | ≈ 0.92 mm |
| 20 G | 0.90 mm | 0.85 – 0.95 mm | 0.56 mm | ≈ 0.56 mm | ≈ 0.64 mm |
| 21 G | 0.80 mm | 0.75 – 0.85 mm | 0.49 mm | ≈ 0.49 mm | ≈ 0.55 mm |
| 22 G | 0.70 mm | 0.68 – 0.75 mm | 0.39 mm | ≈ 0.39 mm | ≈ 0.47 mm |
| 23 G | 0.60 mm | 0.58 – 0.68 mm | 0.32 mm | ≈ 0.32 mm | ≈ 0.37 mm |
The wall convention has names — regular wall, thin wall, extra-thin wall — and the difference between them is larger than the tolerance. A regular-wall 18 G and a thin-wall 18 G differ by roughly 0.12 mm of lumen, which is a quarter of the fiber’s own coated diameter. Both are correctly described as 18 gauge. Both will appear on a purchase order as 18 G. Neither is wrong.
Connect that to the physics. The annular gap between the fiber’s coated diameter and the needle’s lumen is the exhaust path for the vaporized nucleus — the only engineered exit from a closed hydraulic compartment under pressure — and in the common pairings it varies by a factor of more than ten. That parameter, the one that decides whether the patient feels gas pressure in the disc, appears on no line of any commercial document in this supply chain. It is derived, computed from two numbers that live on two different datasheets. The coated diameters for every core in the range are on the bare fiber specification; the lumens are on a needle certificate nobody asks for.
So write “needle lumen, minimum ___ mm” on the order, not “18 G”. If the supplier cannot state the wall type and the minimum lumen, you have not specified the exhaust path. You have specified the puncture profile, which the fiber does not care about.
Tolerance stack-up: what actually moves, and what the tails really threaten
A custom laser fiber is four interfaces stacked inside each other, and every one of them has a tolerance. The fiber has a core, a cladding and a coating. The needle has a lumen. The collar has a bore. The connector has a ferrule. A published datasheet for a 400 µm low-OH step-index fiber looks roughly like this:
| Layer | Nominal (typical) | Tolerance (typical) |
| Core | 400 µm | ±8 µm |
| Cladding | 440 µm | ±4 µm |
| Polyimide coating | 470 µm | ±15 µm |
| Numerical aperture | 0.22 | ±0.02 |
Put those beside the ISO 9626 lumen numbers and you can compute the exhaust area one part number actually delivers. Take 400 µm fiber, nominal coated diameter 470 µm. Against a regular-wall 18 G needle with a minimum lumen of 0.79 mm, the annular area is about 0.32 mm². Against a thin-wall 18 G at 0.92 mm, it is about 0.45 mm². Same fiber, same “18 G needle”, 40 % apart on the parameter that controls venting — and the coating’s own ±15 µm widens the band further.
Two conclusions follow, and one of them is the opposite of what most buyers assume.
The tails do not threaten fit. Worst case is the smallest permissible lumen against the largest permissible coating: 0.79 mm against 0.485 mm leaves about 305 µm of diametral clearance — roughly 150 µm per side, which is more than the whole thickness of the polyimide. Nobody has ever failed to pass a 400 µm fiber through an 18-gauge needle because of a tolerance stack. The fear that drives buyers toward the largest needle the access will allow is real in the abstract and absent in the arithmetic. What the tolerance band threatens is not insertion. It is venting.
The numerical aperture is the tolerance nobody puts on the order, and it is a dose variable. NA 0.22 ±0.02 is a ±9 % swing in the emitted half-angle. In a contact geometry — which is what the disc is, the tip never leaves tissue — the cone angle changes the volume into which the energy is deposited at fixed power, and therefore the volumetric energy density. That is the same physical quantity the operator is trying to control from the console, and it arrived on the tray decided by a datasheet tolerance the purchasing department never saw. Divergence and its consequences are set out on the optical transmission reference; the purchase-order version of the lesson is that NA tolerance belongs on the drawing next to the core diameter.
Here is the sentence a supplier should be able to hear without flinching: buyers spend two weeks deciding between a 400 µm and a 600 µm core — a 2.25× difference in face irradiance that is also, conveniently, visible on a drawing — and then let the needle wall and the coating tolerance be decided by whatever is in stock. The decisions that get made are the visible ones. A custom laser fiber order is a good place to be less visible and more careful.
Protrusion is a dimension of an assembly, so specify the datum, not the gap
Ask for “protrusion 1.00 mm” and you have asked for something that cannot be cut. Glass can be cut to a tenth of a millimeter all day; a 400 µm fiber cleaved by hand and measured with a rule is ±0.3 mm on a good day and worse in a dim room. What cannot be produced is a protrusion that holds ±0.1 mm through assembly, sterilization, freight, and a nurse unwrapping a coil — because the protrusion is not a dimension of the fiber. It is the distance between two surfaces of an assembly, and the assembly is what the supplier is actually making.
That distinction changes what goes on the drawing. The correct form is a datum dimension, not a gap: protrusion 1.0 ±0.2 mm with the collar fully seated against the needle hub; needle length 90 ±1 mm. Written that way, the protruding length is the difference between two lengths that can each be held on a bench — the fiber’s cut length beyond its stop, and the needle’s length — and it is fixed by geometry when the collar bottoms out, so nothing in the design allows it to drift. Written the other way, the protrusion is a measurement somebody performs in the operating room on a device that is already committed, which is a quality check dressed up as a control.
There is a second length hiding behind the first. The strip length — how much coating comes off the proximal end of the working section — has to be protrusion plus margin, and its tolerance has to be looser than the protrusion tolerance, or the strip becomes the binding constraint on a part that was never meant to be difficult. Get the ordering wrong and the jacket remains on the section that enters the disc. A polymer jacket sitting against the cladding where cladding light is present is a localized heat source with an outgassing problem; that is a stripping-process question, not an optical one, and the tool and parameters are on the fiber stripper side of the catalog.
One line only, because it belongs to the clinical article: protrusion plus needle depth equals target depth, and a protrusion figure without a needle depth is a number without a location.
The collar has a material decision that almost nobody makes on purpose
The fixation collar in a PLDD set has three jobs the trade literature has described: it holds the fiber at the chosen protrusion so it cannot slip and laser inside the needle; it permits rotation; and it keeps a path open for gas to escape. Those are the three a design engineer will be asked about.
The fourth job is the one that gets missed, because it only appears in the operating room. A collar made of metal is a radiopaque object sitting at exactly the depth you are trying to image. PLDD guidance is fluoroscopic or CT; the level of interest is the disc you are about to treat; and a stainless-steel ring a few centimeters up the needle shank casts a dense artifact in precisely that region. On fluoroscopy it is a nuisance that can be worked around with a different angle. On CT it is a streak artifact with more authority than the anatomy beside it. A polymer collar — polycarbonate, PEEK, PEI, glass-filled grades where stiffness is needed — holds the same length to the same tolerance and is essentially transparent to the imaging that validates placement.
This is the part of the assembly that is genuinely bespoke, because it has to mate with a specific needle hub. That makes it the part that determines whether an order is a catalog sale or a real custom laser fiber program: it needs a machined batch or a mold, a first-article inspection, and a drawing that fixes its internal bore, its engagement depth with the hub, and its locking feature. It is also where the fixture meets the accessories that hold a needle steady in the first place — the surgical laser accessories side of the range.
Then there is tightening. A set screw is a point load on a thin glass column. Over-torque compresses the cladding under the screw tip, and the fiber then fails at the collar — not at the tip, where everybody is looking — as a progressive power drop with no visible cause at the business end. The strongest available fix is to design the lock so that length is set by a shoulder or a detent and the screw only takes up the remaining clearance, so that a torque spec is a convenience rather than the mechanism. If the lock really is torque-dependent, the assembly drawing has to say so, and the IFU has to repeat it, because the person holding the screwdriver is not reading the drawing.
You cannot sterilize the glass. You sterilize an assembly, and it is a hard shape
Here is the sentence that connects this article to the clinical one. The same annular channel that has to let vapor out during the case has to let sterilant in before it. A three-meter fiber coiled inside a sleeve, terminated at one end and passing through a needle at the other, with a sub-millimeter annular gap along a long, narrow, closed-ended path, is a textbook difficult sterilization geometry — and the difficulty is penetration, not materials.
That is why the answer is ethylene oxide, even before anyone mentions temperature. Steam depends on surface contact with condensing vapor; a long narrow lumen with a blind end traps non-condensable gas that acts as a cold spot, and no amount of cycle time reaches lethal conditions inside it. This is a known failure mode for narrow-lumen polymer tubing, and a coiled fiber assembly with a needle on the end is close enough to it that the material arguments become secondary. EO diffuses into spaces steam cannot purge, the materials tolerate it — cycles run around 45 °C — and the validation framework is mature: ISO 11135, with an overkill half-cycle approach in which the routine cycle is set at roughly double the exposure that inactivates the biological indicator, giving a sterility assurance level of 10⁻⁶.
Three things about EO belong on the specification, and two of them are the ones that get forgotten.
Residuals are a device limit, not a process preference. ISO 10993-7 sets the permissible residues for EO and its two relevant by-products, and the limits scale with contact duration. A disposable surgical fiber is a limited-exposure device — patient contact of 24 hours or less — and the current limits for that category are on the order of 4 mg ethylene oxide, 9 mg ethylene chlorohydrin and 44 mg ethylene glycol per device. A product that meets sterility and fails residuals is a non-conforming product, and residuals are one of the most common deficiency categories in EO-sterilized device submissions.
Aeration, not the cycle, is your lead time. The sterilization exposure is hours. The gas washes around it are longer. The aeration stage that brings residuals down to the limit runs from about a day to about two weeks depending on materials and load. Buyers who plan a launch date from the length of the sterilization cycle are planning from the wrong number.
Gamma is a legitimate alternative for one product in this range and a different conversation for another. Radiation sterilization is validated under ISO 11137, usually against a preset dose of 15 or 25 kGy, and it has a real attraction: no residues, no aeration, no humidity, a room-temperature process. The catch is that the dose that kills the bioburden is also a dose to the glass. Ionizing radiation creates color centers in silica — the E′ center and the non-bridging oxygen hole center are the two that get named most often — and those centers absorb, which is why an irradiated fiber goes dark. The absorption bands peak in the ultraviolet, around 214 nm and 258 nm in the work on gamma-irradiated silica, with tails that reach into the visible and the near infrared. Recovery is real but slow at room temperature; the literature shows essentially complete thermal bleaching only at temperatures far above anything a packaged device will see.
Two practical consequences, and they point in opposite directions:
- Composition decides radiation hardness. A pure-silica-core fiber with a fluorine-doped cladding is among the most radiation-hard fiber constructions there is; germanium-, phosphorus- and boron-doped cores are the ones that darken. An all-silica surgical fiber is the right side of that divide, which is what makes a validated gamma dose defensible at all.
- Wavelength decides whether it matters. The damage is UV-peaked with long tails. At 980 nm and 1470 nm, the tail contribution is small. At 630–750 nm — the photodynamic band that the diffuser fiber is built for — those same color-center bands sit much closer to the working wavelength, and the same 25 kGy that is unremarkable for a bare fiber is a different engineering question. One sterilization method, two products, two answers.
Which produces the specification line that this whole section exists for: state the sterilization method, the dose or the cycle family, the SAL, and require optical transmission to be measured after sterilization, not before. Most fiber datasheets in this industry are issued on unsterilized material, because that is what the fiber maker can measure at the draw bench. The buyer is then holding a transmission number for a device that no longer exists in that form. Post-sterilization transmission is the number that reaches the patient, and it is a line item, not a courtesy.
The pack is part of the device, and the coil radius is the industry’s quietest defect
Sterile barrier packaging has two functions that are easy to state and tedious to prove: keep the device sterile until the point of use, and protect it until the point of use. ISO 11607-1 covers materials, sterile barrier systems and packaging systems; ISO 11607-2 covers validation of the forming, sealing and assembly processes. The second one contains the requirement that trips up first-time buyers: process validation needs samples from three separate production runs, not one good run reproduced three times. A supplier who validates on a single lot has not validated the process; they have validated a day.
The test menu is standard, and its presence on a supplier’s report is a decent proxy for whether the packaging work is real: seal strength by ASTM F88, dye penetration by ASTM F1929 for channel leaks, bubble emission by ASTM F2096 for gross leaks, seal inspection by ASTM F1886, burst testing by ASTM F1140 or F2054, distribution simulation under ASTM D4169 or an ISTA profile such as 3A, and accelerated aging under ASTM F1980.
Accelerated aging deserves its own line, because the arithmetic is simple and the trap is not. ASTM F1980 uses the Arrhenius relation with a conservative Q10 of 2 — the reaction rate doubles for every 10 °C — at the 55 °C that industry actually uses. One year of shelf life simulates in about 52 days; five years takes about 260. The trap is that accelerated data supports a launch but does not replace real-time aging: the claim is only fully validated when a concurrent real-time study confirms it. Shelf life is a packaging claim, not a fiber claim, and a supplier who quotes a three-year shelf life should be able to show you a real-time study that is at least three years old.
Now the defect almost nobody specifies, and the one most likely to reach an operating room. A fiber is a glass filament under permanent surface stress, and it has a minimum bend radius — a long-term figure for storage and a tighter short-term figure for handling. For a 400 µm core in a 470 µm polyimide coating, long-term published figures run on the order of two hundred times the outer diameter: a bend radius in the region of 95 mm, or a coil diameter near 190 mm, for a fiber that spends its life coiled in a pouch. A pack that is slightly too narrow can therefore coil the fiber below its long-term radius, and the failure is exactly the kind this industry is worst at catching. It looks perfect. It passes a visual inspection. It has a hairline flaw in the cladding and a transmission loss that shows up as an underpowered case, or as a fiber that snaps when the scrub nurse pulls it from the coil.
The fix costs nothing and belongs on the packaging drawing: specify the coil diameter, not just the pouch size, and specify it against the long-term bend radius of the fiber actually being packed. Related, and now explicit in the 2019 revision of ISO 11607 as a usability requirement: the pack has to be openable and the contents presentable aseptically, which means the person opening it can see where to open, knows the technique, and can deliver the fiber without touching the sterile field. Where the peel corner is, is a design decision for a device that is opened under pressure by someone wearing two pairs of gloves.
The OEM question is not the price. It is who owns which variable.
A laser fiber is a component, and in the European framework it is normally an accessory: an article that is intended by its manufacturer to be used together with another device so that the other device can be used according to its intended purpose. That definition does more work than it looks like. It means the intended purpose the fiber carries is the console’s intended purpose, and the technical file that assesses it is the console maker’s file — not a document that a fiber supplier writes about itself.
Which is the real content of an OEM arrangement, and why a straightforward purchase order is a poor fit for this product. An OEM order transfers a set of obligations: the drawing, the process validation, the sterilization validation, the packaging validation, the labeling and IFU in the customer’s brand and documentation system, the lot record and traceability at the unit of use, the change-control route, and the complaint-handling path. That is why the supply relationship in this niche takes the shapes it does — a sub-assembly, a matched connector, a color-coded disposable line, a sterile single-pack, an IFU pack prepared under the customer’s brand. Those are not marketing tiers; they are different allocations of responsibility, and the OEM partnerships structure exists to make the allocation explicit before the first drawing is issued rather than after the first complaint.
Two commercial facts belong in the tender, and both are counter-intuitive.
The minimum order is set by the collar, not by the fiber. Fiber is cut to length, and length costs almost nothing to change. The collar is a machined or molded part with a hub-specific geometry, and it carries the tooling or setup cost. So “what is your MOQ?” is not really one question. The useful pair is: what is the MOQ on the collar, and what is the MOQ on the finished assembly? The second number is usually higher, and it is driven by a different cost center entirely — sterilization. An EO chamber has a load volume, and a load that is 12 % full still has to be run, aerated and released as a batch. For a low-volume surgical product, the sterilization lot size often sets the practical minimum order, and a buyer who understands that can negotiate a fill-the-chamber arrangement instead of a unit price.
The question that separates real suppliers from assemblers is unglamorous and takes ten seconds. Ask: which of these lines can you change without telling me? If the answer is “none”, the change control is real and the drawing is the truth. If the answer is “the needle supplier — we use whatever is in stock”, then the annular exhaust area is a variable owned by a company that has never heard of your product, and the buyer has just discovered that the parameter controlling gas trapping in the disc is outside the quality system that certified the assembly. That is the difference between a specification and a wish, and it is decided by a sentence, not a certificate.