The most basic design of demountable connector comprises a ferrule of hole diameter 128–129 μm into which the fiber is bonded using an appropriate adhesive as in the figure below. In this most fundamental form the ferrules can be rotated through 360° within an alignment tube or adaptor.
This rotation allows any eccentricity to be fully explored as is discussed below.
In addition a gap of perhaps 5–10 μm is incorporated between the ferrule faces. This gap is intended to prevent damage to the fiber ends but has the disadvantage that it incurs both Fresnel loss and loss due to separation effects as discussed in Chapter 4. Also the gap creates reflections, which limits the achievable return loss to approximately 11 dB.
It is worth while to review the basic joint losses observed in practice as opposed to those provided in manufacturers’ data, which may have been generated using the interference-fit model or similar distortions. Obviously these losses operate in tandem with losses caused by basic parametric mismatches.
The ferrule hole diameter (FHD) is of considerable importance since the cladding misalignment achieved when jointing a fiber of cladding diameter d1, to another with cladding diameter d2 in an FHD of X will be seen from equation (5.2), in the worst case:
cladding misalignment = X – (d1 + d2)/2
For a fixed FHD the impact of using fibers with diameters at the lower end of the specified tolerance can be quite severe. For instance, for FHD of 129 μm the misalignment for two 122 μm fibers can be as much as 7 μm, which could result in a power loss of 0.85 dB for 50 μm core fibers and total loss for single mode fibers.
Two factors work against such losses being encountered. First the specification for cladding diameter is rarely fully explored, with 97% of all fiber lying in the range 123.5–126.5 μm. Taking due account of these factors limits the misalignment loss to approximately 3 μm, corresponding to a loss of some 0.30 dB for a 50 μm core fiber. Although this figure is considerably lower than the 1.85 dB shown above it is nevertheless much greater than would be predicted using the interference-fit model (0 dB).
Second, the professional termination of the connectors involves filling the ferrule with an adhesive, normally some type of epoxy resin.The fiber is guided through the adhesive and passes through the hole in the ferrule end face. For smaller fibers the adhesive serves to assist in the centralization of the fiber in the hole, thereby reducing the eccentricity.
One of the first optical connectors to be used in data communications was the SMA. This basic demountable connector exhibits random mated worst case insertion losses between 1.3 dB and 2.0 dB dependent upon the quality of the connector components themselves.
The fact that the ferrules within basic demountable connectors can be rotated through a full 360° implies that the impact of core eccentricity (due to fiber manufacturing tolerance) and cladding-based misalignment (due to fixed FHD) will inevitably surface in the form of rotational variations in insertion loss. This results in poor repeatability.
This undesirable rotational degree of freedom has since led to the introduction of keyed connectors, e.g. keyed or bayonet-mount connectors such as the ST and FC, which define the orientation of the ferrule face against another. This prevents rotation and ensures good repeatability. However, this repeatability is only guaranteed in a given joint since the inherent cladding-based misalignments are still present.
An example would be that ferrule A against ferrule B may achieve 0.5 dB insertion loss in a highly repeatable and stable fashion. Also ferrule C against ferrule D may achieve the same performance; however, the difficulty arises when ferrule A is measured against ferrule C or B against D. In these circumstances the insertion loss is unpredictable (but repeatable) although it will lie with the bounds of the limits of the random mated worst case calculation for the joint design.
As a result keyed connectors do not necessarily give better results for insertion loss but are perceived to perform in a more repeatable way; however, this is only true for a given joint.