A fiber collimator converts the diverging output of an optical fiber into a parallel beam, and in reverse couples a parallel beam back into a fiber. It looks like a simple part — a lens in a tube with a fiber behind it — and it is less forgiving than it looks, because every one of its specifications is a match to something else in the system. Collimators are a documented device line in this manufacturer's optical communication programme, listed alongside lasers, isolators and circulators. The product page is fiber collimators; this guide explains how to arrive at the specification that page quotes against.
1. Fiber First: Mode Field Decides the Beam
The fiber is not an accessory to the collimator; it is half of the optical design. A single-mode fiber emits from a tiny, well-defined mode field, and the collimator's lens turns it into a clean Gaussian beam whose diameter is set by the lens focal length and that mode field. A multimode fiber emits from a core tens of times larger, producing a bigger, less tidy beam with different divergence behaviour — a collimator specified for one fiber type does not perform its catalogue behaviour on the other. State the fiber type (and for single-mode, the fiber model or its mode field at your wavelength) before anything else, because it constrains every later choice. The connector or pigtail format follows from the system the collimator plugs into; it is a mechanical match, but an unforgiving one, so name it exactly.
2. Wavelength: The Lens and Coating Are Band Parts
A collimator's lens is designed and anti-reflection coated for a band. Used outside it, two things degrade: the focal length shifts with the glass's dispersion, so the fiber tip no longer sits exactly at the focal point and the output is not quite collimated; and the coating's reflection rises, returning light into the fiber and losing it from the beam. Telecom systems live at their standard wavelengths and laser systems at their laser lines, and collimators are chosen for those numbers, not for 'near enough'. Dual-band and broadband designs exist for systems that genuinely work at two wavelengths — if yours does, say both, because a collimator correct at one line and approximate at the other is a specification, not a surprise to discover later.
3. Beam Diameter: Size It for the Optics Behind It
The collimated beam diameter is the collimator's headline output, and it should be chosen from the receiving optics backwards. Filters, isolators and switches have clear apertures; scan and focusing systems have pupil sizes; a second collimator at the far end of a free-space link has its own acceptance. A beam too small wastes the aperture and focuses worse than it should; a beam too large clips, and clipped light becomes stray light and loss. Beam diameter and focal length are the same choice seen from two sides — for a given fiber, a longer focal length produces a larger, more slowly diverging beam — which is why collimator catalogues are organised by focal length and beam diameter together.
4. Working Distance: Where 'Collimated' Stops Being True
No collimated beam stays collimated forever; it diverges at a rate set by its diameter and wavelength, and the working distance is the range over which it stays close enough to parallel for the system's purpose. Larger beams diverge more slowly, which is the second reason (after spot size) that long free-space links use large-beam collimators. For a two-collimator link, the coupling efficiency between them is the real specification: state the link distance, and the pair can be specified so that each collimator's output is what the other collimator couples best. Buying the two ends separately, each 'standard', is how free-space links acquire their unexplained decibels.
5. The Pairing Mistakes, Collected
Three errors to avoid. Specifying the collimator before the fiber, then discovering the mode field gives the wrong beam. Ignoring the return path — a collimator is reciprocal, and in sensing and interferometry the returning beam's coupling matters as much as the outgoing beam's shape. And treating collimation as a property instead of a match: a collimator is only 'correct' relative to a fiber, a wavelength, a beam diameter and a distance. State all five — fiber and connector, wavelength, beam diameter, working distance — in the first enquiry, so the quotation is written against your stated link rather than against a catalogue default. Where the collimated beam goes next also belongs in the message: into a beam expander, a filter set, or a second collimator across a gap are three different follow-on specifications.
Frequently Asked Questions
What are the five specifications of a fiber collimator?
Fiber type (and mode field), connector or pigtail format, wavelength, collimated beam diameter, and working distance. Together they are the inputs the enquiry asks for.
Why is my free-space link losing more light than expected?
Start with the matching: beam diameter versus the receiving collimator, wavelength outside the lens/coating band, or a link distance beyond the pair's comfortable working range.
Should I buy a matched pair for a link?
Yes — both ends are specified against the link distance, rather than combining two independently 'standard' parts.
Technical References
For the general collimator statements in this guide: RP Photonics Encyclopedia, Fiber Collimators (beam radius from focal length and fiber mode field, lens choice and working distance).
Testing a Collimator on Arrival
A collimator can be verified with standard bench equipment, and two preliminary measurements make a practical first check before acceptance. The first is beam diameter and divergence over distance: measure the spot at two or three distances along the beam and compare the growth with the specification — a beam that grows too fast is a reason to investigate, not a verdict: the collimation state, the fiber's position relative to the focal plane, the mode behaviour, aberration and the measurement setup itself can each contribute, and the check exists to find out which. The second is return coupling: launch the collimated beam into the receiving collimator (or back into its own fiber through a known path) and measure the coupled power against the launch power. That single number reflects the combined coupling performance of that test configuration; which items acceptance actually covers still follows the system's own specification. Record both at acceptance, and a later 'the link degraded' investigation starts from data instead of nostalgia. Neither test needs the manufacturer's laboratory — which is the point of buying a component whose specification is a match: the match can be checked where it is used.
One Enquiry, Five Lines
The guide can be compressed into the enquiry itself. Line one: fiber type and connector, exactly as the system names them. Line two: wavelength, or both wavelengths if the system works at two. Line three: beam diameter required, and what receives the beam. Line four: working distance, and whether the far end is a second collimator, an instrument, or free space onto a target. Line five: quantity, and whether the parts form pairs whose matching matters. Those five lines are what a configured quotation is written against; 'price for fiber collimator' alone leaves the configuration unstated. The component is small and the system depends on it entirely, which is why this guide ends by recommending those five lines rather than by recommending a product.