Endface Geometry and Connector Reliability in the Outside Plant

Angled connectors exhibit two phenomena not found in non-angled connectors: First, the lower ferrule endface radius of angled connectors makes the connector more susceptible to permanent

PON Deployment of Angle

Polished Connectors:

Endface Geometry and Connector

Reliability in the Outside Plant

Abstract: Recent wide-scale deployment of APC connectors in the outside plant for FTTH

initiatives has forced connector manufacturers to consider reliability issues of these connectors Angled connectors exhibit two phenomena not found in non-angled connectors: First, the lower ferrule endface radius of angled connectors makes the connector more susceptible to permanent fiber withdrawal at elevated temperatures Second, any rotation of the ferrule in angled connectors will increase the apex offset of the connector These phenomena may result in loss of physical contact between the fibers of a mated connector pair This air gap may increase reflectance and insertion loss, reducing system reliability The air gap may also allow contaminants to migrate on the fibers Ferrule rotation and permanent fiber withdrawal need to be minimized for acceptable connector performance in the outside plant.

©2005 Optical Society of America OCIS codes: (060.2340) Fiber optic components

Introduction

Angled physical contact (APC) fiber optic connectors with zirconia ferrules have become the de facto standard in PON networks currently being deployed for FTTH initiatives The angle on the fiber in APC connectors produces low back reflections when the connector is not mated to another connector: Reflected light is directed into the fiber cladding because the endface angle is greater than the acceptance angle of the fiber The low reflection of an unmated APC connector provides an advantage for PON networks: A connector in the PON will usually remain unmated until service is required from that circuit This results in unused connectors emitting their signal into air, which can produce Frensel reflections as high as –14dB for connectors without an angle on the endface Service providers need to minimize system reflections so it won’t interfere with OLT card operation One solution is to use non-angled connectors and mate all unused connectors to a terminator This solution adds cost because terminators are an additional component Using APC connectors provides a cost-effective method to greatly reduce reflection issues caused by unmated connectors in the PON Using APC connectors eliminates the

PON Deployment of Angle Polished Connectors:

By Steven C Zimmel

Endface Geometry and Connector Reliability

in the Outside Plant

FTTH deployments are expected to install millions of APC

connectors into outside plant environments over the next

several years Up to this point, APC connectors have been

primarily used for central office and testing applications

The upcoming large-scale deployment of APC connectors

in the outside plant forces us to think about some of the

issues that are unique to APC connectors in extreme

environmental conditions This paper will consider two of

these issues First, it is necessary to understand the

importance of connector ferrule endface geometry

APC Endface Geometry and Outside

Plant Reliability

APC connectors provide two benefits First, APC

connectors produce fiber-to-fiber contact in mated

connector pairs Physical contact between fibers in mated

connectors is needed because it prevents air gaps from

forming between optical fibers Air gaps cause increased

reflectance and inconsistent insertion losses that will

reduce system performance and reliability In addition,

physical contact prevents contamination from migrating

onto the fiber cores The second benefit of APC

connectors is that they minimize reflectance If the radius

is polished at an angle such that the angle of the fiber is

greater than the fiber’s angle of acceptance, reflected light

will be directed into the cladding Using this method, one

can produce connectors with reflectance in the

–50dB range when unmated and as low as -80dB or better

when mated

Physical contact is realized by polishing a radius onto the

ferrule containing the potted optical fiber and by polishing

the fiber such that it is close to level with the ferrule

surface The key for physical contact to occur is to center

the apex of the radius over the optical fiber If the apex is

approximately centered on the fiber, physical contact is

guaranteed because the fibers will be the first objects to

contact each other when connectors are mated

The parameters used to describe the endface of an APC connector are radius, apex offset, fiber height, and endface angle These parameters are defined in the left side of Figure 1 Fiber height is the distance the fiber protrudes or recesses from the ferrule If the fiber is recessed too far into the ferrule, obvious air gaps will occur Apex offset of an APC connector is the distance from the apex of the polished ferrule radius to the center

of the fiber core when the endface is viewed at and an angle of 8° (viewed perpendicular to the plane of the

angled surface) If the apex offset becomes too large, an air gap will develop because the high point of the ferrule

on the radiused endface will be too far away from the fiber core per the right side of Figure 1 The endface radius

is applied to the ferrule by polishing the connector on a compliant surface If the radius becomes too large, the ferrule will become effectively flat This situation could cause local concave features that will introduce a gap between the fiber The radius is usually applied at an 8° angle The angle is in a plane that is 8° from perpendicular

to a plane that intersects the connector key 9° connectors are used, but are rare

APC endface geometry parameters are specified such that physical contact will be maintained over a large range of operating conditions For example, at high temperatures, pressure on ferrule endfaces caused by the spring in the connector may cause the fibers to permanently withdraw into the ferrule if the epoxy holding the fibers creep If the initial fiber height is too recessed into the ferrule, the fiber may creep too far back into the ferrule causing loss of physical contact Proper specification of these endface parameters is necessary to guarantee physical contact in the outside plant

Apex of the Radius

when viewed at 8˚

Apex offset

Air Gap Endface Radius

Fig 1 Endface Geometry of an APC Connector and of Two Mated APC Connectors

Several industry standards exist that address APC

connector endface geometry The purpose of these

specifications is to guarantee that physical contact

between different vendor’s connectors will not be lost The

most used standards are IEC-60874-14-6 and Telcordia

GR-326, Issue 3 These documents require apex offset to

be less than 50 microns, radius to be from 5mm to 12mm,

and fiber height to be ±100nm These three requirements

will allow physical contact to be maintained at

temperatures as high as 85°C with proper epoxy selection

The IEC is in process of updating endface requirements for

APC connectors The new requirements will reflect current

ferrule material properties and will tie all three properties

together into one function The new requirements are

expected to be published in 2005

The preceding discussion shows that proper ferrule

endface geometry is required for good system

performance Properly specified endface geometries

guarantee physical contact between fibers in mated

connectors Physical contact is needed so air gaps don’t

form between fibers in connectors, which will cause

increases in reflectance and insertion loss, reducing system

performance Now that endface geometry and the

importance of physical contact have been defined, we can

discuss issues concerning these properties in APC

connectors in the outside plant

The effort to maintain physical contact in APC connectors

deployed in the outside plant has two unique challenges

not present in non-angled connectors: First, APC

connectors are much more prone to large permanent fiber

withdrawals when exposed to high temperatures If the

fiber withdraws too much, physical contact can be lost

Second, ferrule rotation can create air gaps because of the

angle of the endface These phenomena must be

accounted for to guarantee proper connector

performance and reliability in the outside plant

Permanent Fiber Withdrawal, Radius, and Ferrule Cleanliness:

One manner that fiber height can change, causing fibers

to lose physical contact, is if the fiber permanently withdrawals into the ferrule This phenomenon occurs when mated connectors are exposed to elevated temperatures The force on mated ferrule endfaces transmits pressure to the fiber that is held in place with an epoxy This pressure may cause the epoxy and fiber to creep back into the ferrule at elevated temperatures If the pressure is greater than the epoxy bond strength, the creep will not recover and the fiber/epoxy will permanently withdraw into the ferrule inside diameter Figure 2 shows

an electron microscope scan of an connector before (left) and after (right) GR-326 environmental testing The region

on the left is the optical fiber, the region on the right is the ferrule, and the thin section in the middle is the epoxy Notice how the fiber is lower than the ferrule in the right photo This is the phenomenon of permanent fiber withdrawal

It has been observed that permanent fiber withdrawal is significantly larger in APC connectors than non-angled connectors that are subjected to GR-326, Issue 3 environmental testing We routinely observed APC connectors start a GR-326 test with a protruding fiber and finish with fibers recessed to a point that it no longer meets the –100nm requirement However, we rarely see this occur in non-angled connectors The question we must ask ourselves is why do APC connectors permanently withdraw so much more than non-angled connectors? Before we answer that question, we need to determine which of the GR-326 tests causes the withdrawal

An experiment was performed to determine which

GR-326 environment causes the most permanent fiber withdrawal The experiment consisted of six groups of 12 mated connector pairs (24 APC/SC connectors mated in

12 receptacles) Each group was subjected to either one week of one of the four GR-326 environmental tests, a –40°C cold age test, or an ambient age as a control group The connectors were measured for endface geometry using an interferometer, mated together in an adapter, and subjected to one week of one of the tests listed above After 1 week the connectors were removed from the chambers and allowed to rest at room temperature for one day Next the connectors were uncoupled and the endface geometry was measured The average, maximum, and minimum permanent changes in fiber height are shown in Table 1

Table 1 shows that thermal age and humidity age induced

the most average withdrawal and produced the highest

extremes Interestingly, the sample that went through

thermal cycle saw less change than thermal age Even

more interesting is that –40° age showed very little

change Prolonged exposure to elevated temperatures

induces the most fiber height change This is evident from

the fact that the tests that spend the most time at high

temperatures (thermal and humidity age) exhibit the most

change Thermal and condensation cycles are exposed to

high temperatures, but only for 1/8 of the test cycle

However, even though the cycling tests show far less

change, there is still a significant amount of withdrawal

This indicates that even short exposures to elevated

temperatures may cause the fiber to withdraw in APC

connectors

Experiments were then conducted to determine what

specifically about APC connectors induces more

permanent fiber withdrawal than non-angled connectors

Non-angled (UPC) and angled connectors (APC) differ in

two major ways: First, APC connectors have an endface

radius of 5mm to 12 mm and UPC connectors have an

endface radius of 10mm to 25mm These endface radii are

defined in IEC and Telcordia standards The smaller radius

in APC connectors will cause the pressure on the fiber to

be higher because the spring force is spread out over a

smaller contact area This higher pressure may cause the

fiber to creep more during GR-326 testing Second, many

APC connector manufacturers perform a secondary

grinding operation of the ferrule to change the chamfer to

geometries that aid in achieving low apex offsets This

operation is done before the ferrule is potted and may

introduce contaminants to the ferrule I.D

The experiment looked at two factors: Radius and ferrule

I.D cleanliness APC connectors were made with radii of

5-7 mm, 9-11mm, and 18-22 mm Ferrule inside

diameters were either used as provided or cleaned with a

steam bath and acetone The sample allocations for the

experiment are shown in Table 2 Groups are not equally

sized because the data represents a summation of three

separate experiments evaluating the same two

parameters

Thermal Thermal Humidity Condensation Cold Temp Ambient

Table 1: Permanent Fiber Height Withdrawal for Various GR-326 Environments

Fig 2 Above: SEM Scan of a Non-Withdrawn APC Fiber Below: SEM Scan of a Permanently Withdrawn APC Fiber

All connectors were mated to another connector within an

adapter The connectors were then subjected to three days

of +85°C thermal age followed by three days of

+75°C/95% RH humidity age in order to induce the most

fiber withdrawal The endface geometry was measured

before and after with a interferometer Figure 3 and Table

3 illustrates the results of the experiment The experiment

produced three conclusions: First, radius had the largest

affect on permanent fiber withdrawal Second, cleanliness

has a smaller, but still significant affect Lastly, there is little

or no interaction between cleanliness and radius

We observe that the primary cause of withdrawal in APC connectors is that a smaller endface radius increases the pressure on the ferrule/fiber/epoxy bond because of the smaller contact area The higher pressure causes the fiber

to permanently creep into the ferrule at elevated temperatures Contamination within the ferrule adds to the amount of withdrawal by compromising the epoxy bond strength

The only physical difference between angled and non-angled connectors is the angle at which the connector is polished They are otherwise identical The reason that non-angled connectors withdraw less than angled connectors is that non-angled connectors are usually polished with a radius of about 18mm This is because standards specify radius for non-angled connectors to be 10mm to 25mm The same creep and withdrawal mechanisms occur in non-angled connectors However, because angled connector ferrules may have more contamination, and because angled connectors have smaller radii, we see more withdrawal in APC connectors

Permanent Fiber Withdrawal Experiment Results

0

10

20

30

40

50

60

70

80

90

100

110

120

130

140

150

160

170

5 7 9 11 13 15 17 19 21 23 25

Ferrule Not Cleaned Ferrule Cleaned

Fig 3 APC Connector Fiber Height Withdrawal Test Results

Radius Ferrule not

Cleaned

Ferrule Cleaned 5-7 mm 36 APC connectors 60 APC connectors

9-11mm 12 APC connectors 36 APC connectors

18-22 mm 24 APC connectors 12 APC connectors

Table 2: Design of Experiments Sample Allocation Table.

The results of this experiment suggest that APC connectors should be manufactured with higher endface radii Endface radius is controlled by IEC document 60754-14-6 and GR-326, Issue 3 These documents specify a range of 5mm to 12mm It is common for connector manufacturers to polish APC connectors at the low end of the radius specification However, as the experiment shows, shifting the average to the upper end

Average Permanent Fiber of the 5mm to 12mm range will significantly reduce the amount of permanent fiber withdrawal, reducing the risk of air gaps forming The results also suggest that ferrule cleaning before potting will reduce fiber withdrawal and the possibility of air gaps

Radius Ferrule not

Cleaned

Ferrule Cleaned

Table 3: Design of Experiment Results:

Ferrule Rotation and Apex Offset:

We now examine the issue of ferrule rotation within APC

connectors and the affect this rotation has on apex offset

and insertion loss The rotation is about the ferrule axis as

shown in figure 4

Fig 4 View of an APC/SC Connector from the Front

Apex offset is the distance between the center of the optical fiber and the apex of the polished ferrule endface radius when the ferrule is viewed at 8° If the distance between the apex and the fiber center is large, physical contact may be lost The air gap created by the large apex offset will increase insertion loss and reflectance Therefore it’s important to keep apex offsets low so physical contact

of the optical fibers is maintained Apex offset will change

if the ferrule is rotated about its axis because the endface

is at an 8° angle Physical contact results from the 8° planes of two mated connectors being parallel to each other If one rotates about the ferrule axis, the planes will

no longer be parallel This results in movement of the apex, which will increase apex offset

The change in apex offset as a function of ferrule rotation can be determined using coordinate system transformations Figure 5 shows the endface of the APC connector as it would be viewed by an interferometer along the z-axis That is, an interferometer looks at the radius normal to the x-y plane The ferrule axis is tilted q degrees from normal to the x-y plane and is in the y-z plane q is also the polish angle, usually 8° To simulate rotation about the ferrule axis, we rotate the points on the sphere q about the x-axis Then we rotate the sphere j degrees about the z-axis to simulate ferrule rotation Lastly

we rotate the sphere back by –q about the x-axis The origin remains at all times on the ferrule axis at the point where the optical fiber exits the ferrule

D

ect

io

ferru

le Rot ation

Ferrule (Axis is perpendicular to

Fig 5 Coordinate System for Calculating Ferule Rotation

Apex when viewed at 8˚

Apex offset

Y-Axis X-Axis Z-Axis (Negative)

Z-Axis (Positive) Axis of Ferrule Rotation

Generally, the new location for any point rotated about

the x-axis as defined in Figure 5 is given by [1]:

y” = y’ cosθ- z 0 sinθ (2)

z” = y’ sinθ+ z 0 cosθ (3)

where θis the polish angle θis also the angle that the

ferrule axis is from the z axis Generally, the new

location for any point rotated about the z-axis is given

by [1]:

x“ = x’ cosϕ − y 0 sinϕ (4)

y” = x’ cosϕ + y 0 sinϕ (5)

where j is the angle of ferrule rotation about the ferrule

axis Combining the above sets of equations for the

three rotations described, we get the following:

x = x 0 cosϕ − (y 0 cosθ − z 0 sinθ) sinθ (7)

y” = (x 0 sinϕ +(y 0 cosθ − z 0 sinθ) cosϕ) cosθ +

(y 0 sinθ + z 0 cosθ) sinθ (8)

z” = −(x

° sinϕ +(y

° cosθ − z sinθ) cosϕ) sinθ + (y 0 sinθ + z 0 cosθ) cosθ (9)

where

z 0−R − (R 2−(x 0−A 0x ) 2 −y 0 −A y0 ) 2)1 ⁄ 2 (10)

x, y, and z are the location of the new apex after rotation.

x o , y o , and z o are the location of the new apex before

rotation R is the ferrule endface radius A ox and A oyare the

x and y components of the original apex before rotation.

These can be easily found using the apex and bearing data provided by most interferometers:

A 0x = ApexOffset • cosβ (11)

A 0y = ApexOffset • sinβ (12) where β is the bearing of the apex in degrees, 0° corresponding to the direction of the connector key Equations (7), (8), and (9) represent any point of the endface radius after rotation given its initial position

To find the new apex after rotation we, need to recognize

that the new apex will correspond to the values of x 0and

y 0 that minimize z in (9) after the rotation If we solve for

x 0 and y 0 in (9) such that (9) is minimized, the resulting x 0 and y 0 will be the original position of the new apex We

can then use these values of x 0 and y 0to calculate the new

apex location, x and y, by inserting these values into (7)

and (8) Now that we have the new apex location, we can determine the change in the apex:

∆ apex [(x - A 0x ) + (y - A 0x ) 2 + (z - A 0z ) 2]1 ⁄ 2 (13)

and the new apex offset is simply:

ApexOffset Afterrotation = [x 2 + y 2 +z 2]1 ⁄ 2

(14)

Apex Change vs Rotation for Variouse Radii

Ferrule Rotation (degrees)

250

200

150

100

50

12mm

9mm

5mm

0

300

Fig 6 Apex Offset as a Function of Ferrule Rotation for Various Radii

We found the minimum value of z by varying x 0 and y 0

using the Microsoft Excel solver tool Figure 6 shows how

apex offset will change as the ferrule is rotated about the

z-axis for a variety of radii

The increase in apex offset caused by ferrule rotation may

induce an air gap between two mated connectors, which

will increase reflectance and insertion loss For 2.5mm

ferrules made of Yttria stabilized zirconia, the allowable

fiber undercut as a function of apex offset and radius is [2]:

h = 1988R -.795 - R • 10 6

+√R 2 • 10 6

- ApexOffset 2

• 10 3

where R is the endface radius in millimeters, apex offset

is in microns, and h is in nanometers The last term

accounts for thermal effects A negative value of h in

(15) indicates a protruding fiber and a positive h

designates a recessed fiber

Equation 15 shows us that as the apex offset gets larger,

the allowable fiber height gets more negative That is, as

the apex gets large, the optical fiber must protrude out of

the ferrule more to compensate for the gap the apex will

create If the actual fiber height of the connector is larger

than the value given by equation 15 after ferrule rotation,

an air gap will occur This air gap is:

gap = 0 if h 0 - h < 0

gap = h 0 - h if h 0 - h > 0 (16)

where hois the actual fiber height and h is the value of

magnitude of the wavelength of the signal The relationship between air gap and insertion loss as a function of the air gap length and the wavelength is:

IL + - 10log[1 -2L (1-cos(4πn 8 x)) ] (17) where

L = ( n g - n f

x is the air gap, n is the index of refraction, and λ is the wavelength Figure 7 shows an example of how insertion loss will vary as a function of ferrule rotation for an initial apex offset of 25 microns at a 0° bearing a 12mm radius, and an initial fiber height of 0nm

The graph is not centered about zero because the initial apex offset is non-zero Figure 7 shows that small ferrule rotations can cause APC connectors to loose physical contact and see variation in insertion loss In the example shown in Figure 7, a rotation of slightly over 2° will cause the apex offset to become large enough to induce an air gap Different radii, initial apex offset, and initial fiber heights will produce slightly different values It is important that APC connectors used in the OSP be designed such that the ferrule cannot rotate more than a few degrees The actual amount of allowable ferrule rotation will depend on the typical radius and fibre height of a connector Ferule rotation can easily occur in the field if an operator cleans the ferrule or removes the dust cap with a twisting motion If the ferrule doesn’t return to it’s initial

IL vs Ferrule Rotation

(Apex=25um, R=12mm, h=0nm)

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Ferrule Rotation (degrees)

Fig 7 Example of Insertion Loss as a Function of Ferrule Rotation

λ

n g + n f