This simulator wraps a fiber around an adjustable mandrel. Change mean bend radius, number of turns, wavelength and the fiber calibration, and observe guided power, outward loss markers, loss per turn and the power left after the bends.
• A 3D bend fixture with a reference light source, an adjustable mandrel, a guided fiber winding, outward leakage markers, a radius gauge with turn reference and a bend-loss power meter. • Eight controls: radius, turns, wavelength, calibration choice, reference loss, radius decay scale, radius sweep and marker toggle. • Eight readouts: current radius, calibrated loss per turn, total bend loss, total loss with baseline, output power from 1 mW, power assigned to bend loss, power retained and bent fiber length. • Four guided experiments, a verification bench, Curves & measurements and a Learn & assess tab.
In a tight bend the guided field at the outer side can no longer keep pace with the fiber, so some power radiates away. The simulator uses an explicit empirical law: Lb,turn = A exp[(15 − R)/Rscale] (λ/1550)⁴, where A is the calibrated reference loss at 15 mm and 1550 nm (multiplied by 0.1 for the lower-loss option). Total bend loss is turns times the loss per turn, plus a 0.1 dB baseline, and output power is 1 mW · 10^(−Ltotal/10).
At the reference calibration, loss is 0.2 dB per turn and 0.8 dB across four turns, plus the baseline, and tightening by 5 mm multiplies the per-turn loss by e when the scale is 5 mm.
This is an explicitly empirical macrobend teaching model, not a Maxwell eigenmode solution, named-fiber calibration or standards-compliance check. The reference coefficient, radius decay and wavelength scaling are visible assumptions. Microbending, mechanical stress, fatigue, fracture and installation ratings are not predicted, and pitch and cross section are enlarged for visibility.
Bending distorts the guided field so that part of the power on the outer side leaks out instead of following the curve. The effect grows very quickly as the radius decreases.
Loss in dB accumulates linearly with turns. In the reference calibration, four turns at 15 mm give 0.8 dB in total plus a 0.1 dB baseline.
Longer wavelengths are less tightly confined, so they leak more readily. The model scales loss with (λ/1550)⁴ as a visible empirical assumption.
A fiber designed to confine light more strongly, modeled here by multiplying the reference coefficient by one tenth. Real fibers need their own measured calibration data.