Coming to Grips
Fusion Splicing Guidance for
A
s Fiber-to-the-Home (FTTH) deployments The ITU-T G.652 specification allows
proliferate and voice, data, and video networks a deviation in MFD of 7.9 to 10.2 m,
converge, an increasing amount of optical fiber is which could result in a potential splice
being fusion-spliced. Once viewed as much art as loss of 0.3 dB. In reality, today’s typical
science, fusion splicing has become more routine MFD specification of 9.2 ± 0.4 m for
due to improvements in the fiber itself and the type G.652 single-mode fiber should
development of highly sophisticated and automated fusion contribute ≤ 0.033 dB splice loss at the
splicing equipment. opposite extremes of this spec.
Despite the advances in fiber and fusion splicing technology, However, if unlike fibers with
there are still many aspects of splicing that practitioners must differing MFDs are spliced (for
keep in mind. Differences in fibers, equipment, environment, example, when a G.652 fiber is spliced
and technique can yield different splice loss results. to a G.655 non-zero dispersion fiber or
Understanding fusion splice process capability and splice loss to the new generation of G.657 bend-
measurement will ensure that network owners, designers, optimized fibers), splice losses will
contractors, and technicians have realistic expectations of splice likely be higher than when splicing
loss, especially with the new generation of bend-insensitive one type of fiber to itself.
single-mode fibers that are becoming popular. Core-clad concentricity error can
also contribute significantly to splice
Fusion Splicing 101 loss in single-mode fiber. The typical
Fusion splicing permanently specification for core-clad
joins two optical fibers when no concentricity in today’s G.652 single-
additional changes to those mode fibers is ≤ 0.5 m. If 2 fibers, each
fibers are expected at that with this much core offset, are spliced
juncture. This is in contrast to such that the offset is in opposite
connectors, which are designed directions (total 1 m offset), a 0.056 dB
to allow quick re-configuration loss could be expected.
of fiber links. Using a piece of Between MFD mismatch and core
equipment called a fusion offset, maximum potential splice loss
splicer, the fiber ends are should be ≤ 0.089 (0.033 + 0.056) dB.
prepared, cleaved, and placed in However, this only represents intrinsic
alignment fixtures. At the press of a loss due to fiber geometry and does
button, the fiber ends are heated with not account for any additional loss
electrodes, brought together, and fused. introduced by the splice equipment,
The fiber parameters that most affect technique, or the environment.
splice loss in single-mode fiber are mode For example, in a recent splice
field diameter (MFD - the diameter of study performed by OFS on
the light-carrying region of the fiber) G.652 single-mode fiber,
and core-clad concentricity (the average splice loss of all
amount that the fiber core is off-center samples was quite low at
from the cladding). 0.02 dB. But the study did
Estimated loss from mismatches in yield some splices that
MFD can be calculated from the had losses of 0.1 dB or
following equation: higher. (See Figure 1)
Loss (dB) = 20 x LOG10 [(2 x MFD1 x MFD2) / It should also be noted
(MFD12 + MFD22)] that splice studies
1 COMPLIMENTS OF OSP® • M AY 2 0 0 8
With Loss
Single-Mode Fibers
performed in a lab, like those humidity, wind, and dust. The most requirement. A high loss splice in
referenced above, are usually done thorough method to determine the a link that meets its loss budget
in ideal conditions with state-of-the- proper setting is to perform a should not be of concern.
art, well-maintained splicing splicing trial where a dozen or more
equipment and cleavers. In reality, splices are performed and measured 3. Optical Time Domain
splicing is often done in haste, and by other means (as described below) Reflectometer (OTDR)
in less-than-ideal conditions, with and compared to the splicer’s An OTDR is a very useful tool
equipment that may be well used estimated readings. This provides a that provides a representative
and not performing at its very best. correlation between the splicer’s picture of the fiber and only
estimate and the actual results. If requires access to one end of the
Measuring Splice Loss such a trial is not practical, the re- fiber. It is the best method for
Once a splice has been made, it is splice criterion should at a minimum measuring the loss of an
usually measured to see how much be set at the mean splice loss individual splice. (See Figure 2)
loss occurs at that joint. There are specification for the job. Setting it any But a good understanding of
several techniques and approaches to lower could result in a rework how an OTDR works, and its
doing this. Which method, or without really knowing if the limitations, is necessary
combination of methods, is used rework is necessary. in order to avoid
will depend on various factors, confusion or debate
including the design, purpose, and 2. Optical Loss Test Set (OLTS, in determining
critical nature of the network, as also known as a Source and whether splice loss
well as any industry specifications Power Meter) requirements are
or standards that may govern such. A traditional light source and
There are three basic ways to power meter can be used to
obtain a splice loss reading, each estimate splice loss on a relatively
with advantages and limitations: short link containing one splice.
This is still an estimate, since an
1. Fusion Splicer Estimated Loss OLTS measures the loss over the
Fusion splicers can give you an entire link, which would include
estimate of the splice loss, but it is the fiber as well as the splice. If
exactly that: an estimate. Caution is there are multiple splices in a link,
urged in relying on estimated an OLTS will only report the total
splice loss to determine whether a loss of all the splices. It cannot
splice is acceptable. There are no distinguish between individual
standards governing splice loss splices. Only the average loss of each
measurements by a fusion splicer, splice can be determined.
and no means of calibration, so An OLTS is an economical and
the accuracy of such useful way to make sure a link
measurements can vary widely. meets a power budget
Although fusion splicer loss requirement, which of course is
estimates are often used for the the most important criterion. Even
initial accept/re-splice decision, if there are splices in a link that
where to set re-splice criterion might show elevated splice loss, it
depends on the splicer, the fibers, does not matter as long as the
the technician, and environmental overall link meets the link loss
factors such as temperature,
W W W. O S P M A G . C O M 2
F i e ld T e c h s , T ools , and t h e C h ang i ng O S P
divided by 2). 0.02 dB is the true loss individual splices in a link may be
of this splice. counterproductive.
Because of an OTDR’s inherent In general, a reasonable splice loss
benefit, the ability to measure a fiber spec should be established based on
optic link from one direction, it is the design of the link and should be
tempting to do only one-way an average of all splices. If
measurements to certify a link, individual splice loss limits are
including splice losses. However, if imposed, keep in mind that some
Figure 1. Graph of OFS Splice Study
certification is required, individual splices may not meet the
performed on G.652 single-mode fiber.
measurements should always be spec, due to fiber geometry
taken from both ends and results variation, equipment, environmental
met, as well as preventing averaged. (Many OTDRs do the conditions, technician expertise, etc.
unnecessary rework or delay. averaging for you.) There should be some allowance for
An OTDR uses back-scattered (i.e., It does no good, and serves no this. One splice that may not quite
reflected) light to measure loss over purpose, to perform certification meet the spec will certainly be offset
distance. Consequently, it can from one direction only. by several other splices that are well
sometimes be fooled into thinking Unidirectional results do not reflect within the spec.
that there is more or less loss than the true performance of a splice. In Avoid setting a unidirectional
there actually is. The reason is that fact, you can be misled into thinking splice loss specification. If
the amount of light reflected back is the splice is acceptable, or not unidirectional measurements are
affected by changes in MFD. If light acceptable, when the opposite is used at all, it should be done with
goes from a smaller MFD into a actually true. Imposing some knowledge of the fiber being
larger MFD, less light is reflected unidirectional specs leads only to spliced, along with the capability of
back, making it seem like the loss is extra work and headache. the equipment being used. If a
greater than it really is – known as Unidirectional OTDR unidirectional limit is too tight, it
exaggerated loss. Conversely, if light measurements are useful for will cause many unnecessary
goes from a larger MFD into a troubleshooting, or for first-pass remakes.
smaller MFD, more light is reflected assessment of splices to decide if any One strategy that can be useful in
back at this point and the OTDR will should be remade. As long as a the field is to set a unidirectional
think that power was actually gained reasonable and moderate limit is put threshold of 0.3 dB or higher, which
at the splice (often called a “gainer”). in place on unidirectional loss, this would trigger a remake. If, after
In reality, power cannot be gained, can save time. (See “Splice Loss two tries, the unidirectional
and some amount of loss occurs at Specifications” below.) The few measurement does not come down,
the splice joint. However, recall that splices that might read “high” can a mode-field mismatch is very
even with 2 fibers at the opposite be remade. But the majority that likely. The splice should be
extremes of the typical MFD spec, read fine or even slightly high will measured from the other direction
actual loss due to MFD mismatch likely be fine when they are to determine its true loss before
will be ≤ 0.033 dB. (See Figure 2) measured at the other end and bi- spending more time on remakes.
directional results are averaged.
Two-Way (Bi-Direc- Guidelines for Network
tional) vs. One-Way Splice Loss Planners and Designers
(Unidirectional) Specifications Industry specifications vary
Measurement Splice loss specifications are somewhat in the allowable loss for
The industry recognizes that the intended to ensure that the overall splices, but they agree on two major
most accurate measurement of link loss is within budget. System points:
actual splice loss is the average of a performance is dependent on overall 1. True splice loss is based on a
two-way (bi-directional) OTDR test. link loss, not individual splices. In a bi-directional average OTDR
This is documented in industry typical link, some splices will likely measurement.
standard ANSI/TIA/EIA-455-8-2000 be on the high side. However, splices 2. Splice loss specification is
Measurement of Splice or Connector on the low side will offset these. As based on the mean splice loss of
Loss and Reflectance using an OTDR. long as total splice loss (and cable all splices.
For example, if a splice measures and any other connection loss) To ensure the end result of a high-
0.18 dB from one end, and -0.14 dB doesn’t exceed the power budget, performance network at the lowest
(gainer) from the other end, the there won’t be a problem. Imposing installed cost, network planners and
average is 0.02 dB (0.18 + [-0.14] a stringent splice loss spec on designers should:
3 COMPLIMENTS OF OSP® • M AY 2 0 0 8
F i e ld T e c h s , T ools , and t h e C h ang i ng O S P
significantly. Set a generous Tony Irujo is Manager of Customer Technical
threshold for remake. Support with OFS. Tony has 15 years experience
Guideline 5. If using unidirectional in optical fiber manufacturing, testing, and
OTDR testing to qualify splices, do applications. He provides technical, sales, and
not break and remake excessive marketing support for OFS and the user
losses more than once without community. For more information, visit
performing a bi-directional OTDR [Link].
average to validate the finding.
Figure 2. OTDR Trace Tim West, RCDD, is Applications Engineering
The Bottom Line Manager with Superior Essex Communications LP.
Fusion splice loss is affected by He has 10 years of experience in communications
many factors, only some of which
• Set splice loss specifications based on the cable engineering and manufacturing. For more
overall requirements of the link. can be controlled or manipulated by information, visit [Link].
splice technicians. Variations in glass
• Set splice loss specifications based on the
mean splice loss of all splices. geometry, between fiber types, fiber
• Specify bi-directional average OTDR manufacturers, and even within the
tolerances of the same fiber type, can
measurement for splice loss measurement.
result in true splice losses exceeding
• Avoid specifying unidirectional OTDR splice
loss requirements. 0.1 dB, and unidirectional
measurements (gainers or
Guidelines for Splicing exaggerated losers) even higher.
Contractors and An OTDR is the best device to
Technicians use for measuring splice loss, with
Many advances have been made the true loss being the average of
in optical fiber and the equipment bi-directional measurements.
used to splice it, but fusion splicing Unidirectional measurements can be
remains a technical skill that can be misleading, sometimes showing a
affected by many different factors. gain in power or exaggerated loss.
To ensure quality splices with This is due to natural variations in
minimal rework, splicing MFD between spliced fibers and the
contractors and technicians should way the OTDR measures the
consider the following guidelines: backscattered light at these joints.
Guideline 1. Follow the applicable Splice loss specifications should be
equipment manufacturer’s set with the total link power budget
guidelines for setup and in mind and be based on average
maintenance of all splice equipment. splice loss. This will ensure that an
Guideline 2. Maintain clean installation has realistic splice loss
equipment and a clean splice requirements and that the power
environment, being especially wary budget can be met with confidence
of windy and/or dusty conditions. while minimizing the time and
Guideline 3. Use the fusion splicer’s expense associated with splice
estimated splice loss reading as an remakes.
initial go/no-go evaluation of the
splice. Establish the limit at the same
dB value as the mean splice loss
value in the job spec. If the initial
splice does not fall within the limit,
repeat the splice no more than two
more times without performing a
bi-directional OTDR average to
validate the finding.
Guideline 4. If using unidirectional
OTDR testing to qualify splices, do
not break and re-splice gainers.
They typically will not change
W W W. O S P M A G . C O M 4
