@Arkadiy.mastin   I think in your particular case, there may be an issue with the aperture setting.  My guess is that you are using the Object Space NA aperture with a value equal to the NA of your fiber (looks like 0.09).  However, if this is indeed the case, then you are clipping the wings of the Gaussian beam when using the FICL analysis.  If you increase the NA of the aperture to say 0.15, you will see the coupling efficiency reported by FICL goes up significantly and closely matches that of POP. (Note: If the “System Efficiency” value is not close to one, then light is being lost prior to fiber coupling.)

In general, FICL should be okay as long as diffraction in collimated space can be ignored.  A good way to quantify this is based on the Fresnel number of the collimated beam at the output of the first lens as viewed from the input to the second lens.  If the Fresnel number is on the order of 10 or larger, then diffraction can be neglected.  The Fresnel number is given by Fn = r^2/(lambda*z), where r is the radius of the collimated beam leaving the first lens, lambda is the wavelength, and z is the propagation distance to the second lens.  In your case it looks like Fn is about 8, which is probably just large enough that diffraction is minimal, which is why the FICL and POP coupling values are comparable (again with an appropriate aperture setting).  In Paul Colbourne’s example where FICL and POP report fundamentally different values, Fn = 0.26, so diffraction cannot be neglected and FICL is no longer valid unless you use his clever skew ray scheme.

Here is my version of your system.

Aperture: Object Space NA = 0.09

Aperture: Object Space NA = 0.15