Coupling a laser into a single-mode fibre
Two mirrors put the beam on the axis of a lens, and the lens focuses it on the end of the fibre: the layout of every fibre launch, with the one condition a ray model can check, the acceptance angle.
Open in the canvas →Background
A single-mode fibre guides light in one spatial mode, a spot a few micrometres across. To launch a free-space beam into it, the beam has to be focused on the end of the fibre with a profile similar to that mode and with flat wavefronts[1]. The core is typically only a couple of microns in diameter and has to be hit, which asks for mechanics of sub-micron precision[1].
The layout is nearly always the same. Two mirrors steer the beam so that it arrives along the axis of a lens, and the lens focuses it on the fibre's end face. The end of the fibre sits on translation stages that set its position across the beam, with further controls for its angle[1]. The focal length of the lens is chosen so that the focused spot has the size of the fibre's mode.
One part of this can be said with rays alone. A fibre accepts light only within a cone around its axis, of half-angle arcsin(NA), where NA is its numerical aperture. A beam of radius r focused by a lens of focal length f converges at a half-angle of arctan(r/f). If that is larger than the acceptance angle, the outer part of the beam cannot be guided, whatever the alignment.
What this setup demonstrates
A 633 nm laser, 1 mW in a 2 mm beam, is folded by two mirrors onto the axis of a lens of f = 11 mm. The end of a fiber stands in its focal plane. The fibre's input NA is 0.12, an acceptance half-angle of 6.9°; the beam converges at arctan(1/11) = 5.2°. The fibre is set to no loss and emits from its far end into the same NA, onto a power meter.
The meter reads 1.00 mW. What that number means matters: the tracer couples every ray that lands on the end face of the fibre's connector within the acceptance angle, and here all 25 rays of the beam do. It is the condition on the angle, and nothing else.
Things to try, each of which the acceptance angle explains. Each one starts from the setup as it opens, so undo a change before making the next:
| One change to the setup as it opens | What happens | Power meter |
|---|---|---|
| Lens to f = 4 mm | The beam converges at 14°; the outer rays are refused | 440 µW |
| Laser's beam widened to 4 mm | It converges at 10.3° | 600 µW |
| Fibre's input NA lowered to 0.08 | The acceptance half-angle falls to 4.6°, below the beam's 5.2° | 840 µW |
| Mirror 2 turned by 0.5° | The beam crosses the lens 2 mm off its centre and the cone arrives tilted | 240 µW |
| Mirror 2 turned by 1° | The cone arrives outside the acceptance angle | no light |
The readings move in steps of 40 µW, one ray of the 25. Two changes together do not simply add: with the beam at 4 mm and the input NA at 0.08, the meter reads 440 µW.
The enlarged view shows the cone meeting the end face, and prints the waist an ideal Gaussian beam with this cone would have, 2.23 µm (λ/(π·NA) for the cone's own NA of 0.09). The tracer does not use that number. Whether it matches the fibre's mode is the question the Free space to fiber coupling calculator answers, with the efficiency lost to a mismatch in size, an offset, a tilt or a defocus.
The reading is not a coupling efficiency. The fibre here has no core and no mode: its entrance is the whole end face of the connector, 10 mm wide in the drawing, and any ray that meets it within the acceptance angle is coupled in full. A focus that misses the centre of the face by a tenth of a millimetre, or lies a few millimetres in front of it, still reads 1.00 mW here; a real single-mode fibre, whose core is micrometres across, would be dark. Pull the fibre's end 3 mm back from the focus and the reading does not change.
So the setup shows where the focus lies and at what angles the light arrives, and counts the rays inside the cone. The overlap of the focused beam with the fibre's mode, which sets the efficiency of a real launch and its tolerance to position, angle and focus, is not computed on the canvas, nor are the reflection at the end face or the aberrations of a real lens.