Abstract

The measurement of Polarization Dependent Loss hasgained a great deal of attention among componentmanufacturers.

This Application Brief discusses two differentmeasurement techniques, the Polarization Scanningtechnique, and the Mueller Method, and examinespractical implementation difficulties.

Polarization Dependent LossMeasurement of PassiveOptical ComponentsApplication Note

2

IntroductionFiber optic networks are the present and future medium ofchoice for high-speed, high-volume data transmission. Thegrowth in demand for greater data throughput requiresgreater bandwidth and smaller channel spacing.

The rapid development of fiber-optic network technology isdriven by dramatic advances in the design andmanufacture of both active and passive optical devices.The tremendous need for higher data transmission rateshas always driven the development of new opticalcomponents to the limits of existing technology.

The development and testing of new optical componentshas become more challenging and complex, for example:

• Channel spacing is constantly being reduced, sowavelength dependent measurements must beincreasingly accurate.

• The complexity of multi-channel test systemsincreases as the number of channels increases.

• The extension of optical data transmission into newspectral regions, such as the L-band, involves thedevelopment of both optical components, and theequipment required for testing them.

• The performance of DWDM systems is increasinglyinfluenced by the polarization of light wave signals.The increasing length of fiber links has focusedattention on new test parameters, such aspolarization dependent loss, which is a signaldistortion that accumulates over distance.

• Higher data transmission rates (10 Gbit/sec or40Gbit/sec) require shorter pulse duration. In thefrequency domain, this results in a broader spectrum.High transmission quality requires broader spectralareas of low polarization dependent loss, to avoidattenuation variations for different spectralcomponents.

In addition, due to the rapid growth in the fiber-optictechnology market, manufacturers must ramp upproduction volumes by increasing manufacturing capacity,and by shortening test time while not compromising testaccuracy.

This Application Note focuses on the evaluation of twopolarization dependent loss measurement techniques thatare suitable for deployment in the high volumemanufacture of passive optical components. Theadvantages and disadvantages of each technique isdiscussed. Finally, a typical applied measurement solutionis described in detail.

First, polarization dependent loss is briefly defined, and itseffects in fiber-optic transmission links described.

Polarization Dependent Loss �DefinitionPolarization dependent loss is a measure of the peak-to-peak difference in transmission of an optical component orsystem with respect to all possible states of polarization.It is the ratio of the maximum and the minimumtransmission of an optical device with respect to allpolarization states.

Polarization Dependent Loss, PDL, is defined as:

���

����

�∗=

Min

MaxdB P

PPDL log10

Equation 1: Definition of polarization dependent loss.

In Figure 1, the effect of applying all possible states ofpolarization to an optical component is shown. Thepolarization of the constant, and fully polarized, inputsignal is varied. As the polarization of the incident lightvaries, the output signal shows a corresponding change inpower.

Polarization dependent loss (PDL)

MIN

MAX

PP log 10PDL =

Device under testConstant power,100% polarized,scan polarization

TimePMIN

PMAXOutputInput

� PDL measures the peak-to-peak difference intransmission for light with various states ofpolarization

Figure 1: Polarization Dependent Loss of passive opticalcomponents.The output power variation is the result of the variation inthe polarization of the incident light wave signal.

3

Causes of Polarization Dependent Loss

The polarization dependence of the transmission propertiesof optical components has many sources. Some of themost common effects are:

• Dichroism

• Fiber bending

• Angled optical interfaces

• Oblique reflection.

Polarization Dependent Loss inoptical transmission networksAll the above effects appear in the standard opticalcomponents used in fiber-optic networks.

A typical structure of a fiber-optic transmission networklink is shown in Figure 2. The transmission link includes anumber of different passive and active components. Themost common passive devices that exhibit PDL includeoptical couplers, isolators, wavelength-divisionmultiplexers (WDM) and photodetectors.

State of the art WDM link

Data out

ADM

Data in

MUX

MUX

λλλλ1111

λλλλn-1

λλλλn

EDFA

EDFA

DEMUX

DEMUX

λλλλ1111

λλλλn-1

λλλλn

Splice / Connector

Transmitter

Gain Flattening

Receiver

Figure 2: Typical WDM link in fiber optic networks.

The polarization state that exhibits maximum loss (that is,minimum transmission) through one component is generallynot the same as for other components in the transmissionlink.

Furthermore, the polarization state is not maintained alonga fiber. The evolution of polarization along a fiber is of acompletely statistical nature and, in consequence, istotally unpredictable.

Even if the PDL axis of every component is aligned, thisdoes not correspond to the minimum or maximum effect onpolarization sensitive transmission. Since PDL effects buildup in an uncontrolled manner, PDL can lead to adegradation of the transmission quality of the fiber-optic

link, or even to a failure of the optical system. Therefore,modern fiber-optic communication systems requirecomponents with low PDL.

Consequently, the measurement of PDL has attractedenormous attention from component manufacturers. Theneed for PDL test solutions is accompanied by therequirements of short measurement time, high accuracyand high reliability.

In the following, two PDL measurement techniques aredescribed and evaluated for their suitability for modernhigh-volume manufacturing.

Measurement techniques

In the context of passive component testing duringcomponent manufacture, two techniques for determiningthe PDL of a device under test (DUT) are recommended:The Polarization Scanning technique, and the MuellerMethod. While the Polarization Scanning technique isfound suitable for PDL measurements at specificwavelengths, for many wavelength points in a broadwavelength range the Mueller Method shows clearadvantages. Both techniques deserve a more in-depthtreatment.

The Polarization Scanning techniqueThe Polarization Scanning technique is the fundamentalmethod for measuring PDL.

The DUT is exposed to all states of polarization and thetransmission is measured with a power meter. Themaximum and minimum transmission through the DUT candirectly be measured. The PDL can then be calculatedusing Equation 1.

DUT

∆P (Sin)S0 Optical Power Meter

Optical Source

11896A

0

10

20

30

40

50

60

0 10 20 30Polarization scan time (s)

Mea

sure

men

t Unc

erta

inty

(%)

HP11896A P. ControllerRandomize at rate=5Pwr mtr avg time=20 ms

Typical Setup for Polarization Scanning

Relation between PDL errorand Scanning Time

Figure 3: Setup of a PDL measurement using thePolarization Scanning technique.The graph shows how the PDL measurement uncertaintydepends on the measurement time.

4

Exposing the DUT to all states of polarization is all butimpossible. In practice, a number of polarization states aregenerated at a scan rate that is suitable for the powermeter averaging time. The longer a polarization scan takes,as the transmission through the DUT is obtained for morepolarization states, the smaller the uncertainty of the PDLmeasurement [1]. This is demonstrated by the graph inFigure 3. At some point, increasing the measurement timedoes not yield significantly improved measurementaccuracy. Here, where the polarization controller'srandomize rate is 5 and the power meter's averaging timeis 20ms, a measurement uncertainty of 5% requires apolarization scan time of 10s. Increasing the measurementtime to 20s, (that is, measuring over twice the number ofpolarization states) results in a measurement uncertaintyof 3%, an improvement of only 2%. Consequently,improving PDL measurement uncertainty must always beconsidered in the context of the affect on measurementtime.

A typical PDL measurement setup employing polarizationscanning is shown in Figure 3. The source produces nearlyfully polarized light. The 11896A Polarization Controllertransforms the polarization by means of four motorizedfiber loops. The movement of the fiber loops causes avariation in the birefringence of the fiber, which results invariation of the polarization state. The different rotationalspeeds of the fiber loops generate polarization states in apseudo-random manner. The 11896A PolarizationController provides eight different scan rates, where thefastest scan is denoted by rate 8.

Fiber loop polarization transformation

Stacked fiber coils

Input E-fieldfas

t

slow

fast

slow

fastslow

Input field at each coil decomposes into the fast slow axis of each coil

Reorientation of a coil changes decomposition of input E-field

Figure 4: Fiber loop polarization transformation, shown onan example with three fiber loops.

Setting the correct polarization scan rate with respect tothe averaging time of the power meter is critical. Thepolarization scan rate dictates how rapidly the polarizationof the light wave signal is changed. A faster scan rategenerates more polarization states in a given time interval,so might decrease the duration of a measurement.However, if the polarization scan rate is too fast withrespect to the averaging time of the power meter, results

are falsified. At faster scan rates, the power meteraverages over more polarization states; a maximum orminimum transmission could be averaged out. It is clearfrom Equation 1 that an error in maximum or minimumtransmission value directly affects the PDL value obtained.

Averaging time is also critical in terms of noise. The signal-to-noise ratio is proportional to the square root of theaveraging time. Clearly, choosing the optimum averagingtime is a trade-off between the quality of themeasurement in terms of noise and the measurement time.How averaging time affects the PDL results isdemonstrated in Figure 5.

The three measurement examples at various averagingtimes show that with a small averaging time, such as100µs, the quality of the measurement is degraded bynoise. On the other hand, a long averaging time provides novisible improvement of the measurement results.

PDL vs Avg. time

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1531 1531.5 1532 1532.5 1533 1533.5 1534 1534.5 1535 1535.5 1536

Wavelength [nm]

PDL

[dB]

100us10ms1ms

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

1533.5 1533.75 1534 1534.25 1534.5 1534.75 1535

100us10ms1ms

Figure 5: Polarization Scanning with three differentaveraging times: 100us, 1ms, and 10ms.

The Mueller MethodA different approach to the measurement of PDL is todetermine the Mueller matrix for the DUT. The technique istherefore known as the Mueller Method.

5

Mueller Method

DUTPn (Sn, λ)S0 Optical Power

Meter Optical Source

8169A

Pol λ/4 λ/2Sn

n=1..4

Application of four well-known polarization states to the DUT

Optical power transmission measured at these polarization states only

Transmission data are used to calculate Mueller matrix coefficients

PDL is obtained from the Mueller matrix coefficients

Single wavelength PDL measurement within few seconds

Figure 6: Principle of a PDL Measurement using theMueller Method.

The Mueller Method determines PDL by exposing the DUTto only four, but well-known, states of polarization. Thefour polarization states are chosen to be LHP (linearhorizontal polarized), LVP (linear vertical polarized), L+45(Linear +45 degrees), RHC (right hand circular). The PDL iscalculated from the transmission results.

This approach was first introduced in Reference [2].Additional Information is provided by Reference [3].

The PDL measurement procedure has two steps, areference measurement and the DUT measurement, asillustrated by Figure 7.

Step 1: M easure the pow er Pn at the input of the DUT ( for calibration )

DUTPn (Sn, λ)S0 Optical Pow er

M eter Optical Source

HP 8169A

Pol λ/4 λ/2Sn

n=1..4

(S n, λ)S0 Optical PowerM eter

Optical Source

HP 8169A

Pol λ/4 λ/2Sn

n=1..4nP~

Step 2: M easure the pow er P n at the output of the DUT

~

Figure 7: Measurement Procedure of Mueller method:Reference and DUT measurement.

First, the optical power at the four defined polarizationstates is measured. In the second step, the same fourpolarization states are applied to the DUT and thetransmitted optical power is measured.

The Mueller matrix describes the polarization and powertransmission properties of the DUT. The relationshipbetween an input Stokes vector and output Stokes vectorof a DUT can be written as:

inDUTout SMS ∗=

where MDUT is the Mueller matrix of the device.

The Mueller matrix is a 4x4 matrix. The four first-rowcoefficients of the Mueller matrix describe the powertransmission of a device, which is sufficient to obtain thePDL.

As stated previously, the reference measurementdetermines the power of the input Stokes vector. The DUTmeasurement yields the total power transmitted throughthe DUT. When measured for the four polarization states,a system of linear equations can be solved to determinethe desired coefficients of the Mueller matrix, as shown inFigure 8.

From these coefficients, the maximum and minimumtransmission can be derived, as shown in Figure 9, fromwhich the PDL can be calculated, as shown in Equation 1.

�����

�

�

�����

�

�

=

00

~~

1

1

1,PP

Sin

�����

�

�

�����

�

�

−=

00

~~

2

2

1,P

P

Sin

�����

�

�

�����

�

�

=

0

~0

~

3

3

1, P

P

Sin

�����

�

�

�����

�

�

=

4

4

1,

~00

~

P

P

SinLHP: LVP: L+45: RHC:

2121112

2121111~~

~~

PmPmP

PmPmP

−=

+=

4144114

3133113~~

~~

PmPmP

PmPmP

+=

+=

Input Stokes Vectors ( from DUT input power measurement - Step 1)

Output Stokes Vectors (from DUT ouput power measurement - Step 2)

Figure 8: With the power measurement results, a systemof linear equations can be solved.

����������

�

�

����������

�

�

−

−

���

�

�−

���

�

�+

=

����

�

�

����

�

�

114

4

113

3

2

2

1

1

2

2

1

1

14

13

12

11

~

~

~~21

~~21

mPP

mPP

PP

PP

PP

PP

mmmm

214

213

21211max mmmmT +++=

214

213

21211min mmmmT ++−=

Solving the equation system for m11..m14 yields:

minimum and maximumtransmission through DUT:

Figure 9: Calculation of minimum and maximumtransmission.

The four polarization states are synthesized by an 8169APolarization Controller. The polarization controller consistsof a polarizer, and two retarder plates (one quarter-wave,and one half-wave). All elements are rotatable; the axis ofrotation being parallel to the direction of light propagation.

6

The polarizer generates a linear polarization state, whichthe retarder plates transform into any other polarizationstate. Desired polarization states are obtained by settingthe retarder plates to specific angles.

The polarization controller exhibits a polarizationdependent loss. The PDL of the polarization controller isspecified to within ±0.03dB. The reference measurementrecords the absolute power at each of the four polarizationstates. If the system is unchanged, each polarization statehas the same output power during the DUT measurementas during the reference measurement. Hence, powervariation across polarization is taken care of by thereference measurement.

Advanced PDL measurement

Requirements for maximum accuracy

Regardless of which measurement technique is used, toachieve the highest possible accuracy, its setup must meetcertain requirements:

Zoom: Insertion Loss of a Transmission Window, AWG-type filter (average of 4 pol. States, 10 measurements)

0.00002

0.000021

0.000022

0.000023

0.000024

0.000025

0.000026

0.000027

1.53

4E-0

6

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

41E-

06

1.53

42E-

06

1.53

42E-

06

1.53

42E-

06

1.53

42E-

06

1.53

42E-

06

1.53

42E-

06

Wavelength[m]

pow

er [W

]

SeriesSeriesSeriesSeriesSeriesSeriesSeriesSeriesSeriesSeries

Power Stability

Wavelength Accuracy

Figure 10: Evaluating the measurement setup includeswavelength repeatability and power stability.

• The tunable laser source must have a stable poweroutput. Any variation in the output power of the lasersource is not recognized in a PDL measurement, andmay be misconstrued as polarization sensitivity ofthe DUT.

• Wavelength accuracy and wavelength repeatabilityplay important roles in the quality of a measurement.Wavelength accuracy determines the absolutelocation of the filter curve along the wavelength axis.Wavelength repeatability is especially important forthe Mueller Method, where the filter curve ismeasured four times at different input polarizationstates. Any deviation in wavelength between the fourmeasurement results can severely affect the finalPDL result.

Power stability and wavelength repeatability can easily bequalified by repeated measurement of a filter transmissioncurve at a fixed input polarization, as demonstrated inFigure 10. Power stability is best evaluated at the peak ofthe filter transmission curve. The overlap at the slope ofmultiple filter curves is a valuable measure of wavelengthrepeatability. A sample measurement series is shown inFigure 11.

Wavelength Accuracy: Slope of a Transmission Window, AWG-type filter (average of 4 pol. States, 10 measurements)

0.000020.00002020.00002040.00002060.0000208

0.0000210.00002120.00002140.00002160.0000218

1.53

4180

E-06

1.53

4181

E-06

1.53

4181

E-06

1.53

4182

E-06

1.53

4182

E-06

1.53

4183

E-06

1.53

4183

E-06

1.53

4184

E-06

1.53

4184

E-06

1.53

4185

E-06

1.53

4185

E-06

1.53

4186

E-06

1.53

4186

E-06

1.53

4187

E-06

1.53

4187

E-06

1.53

4188

E-06

1.53

4188

E-06

1.53

4189

E-06

1.53

4189

E-06

1.53

4190

E-06

1.53

4190

E-06

Wavelength[m]

pow

er [W

]

0.5pm

~ 0.5pm

Figure 11: Demonstration of the wavelength repeatabilityof 10 wavelength sweeps.

The detector in the test setup also plays an important role.As stated in the introduction, photodetectors are amongthe components that exhibit polarization dependence.Thus, it is essential to use detectors with low PDL. Asmentioned earlier, the PDL of different componentscombines in an uncontrolled manner, so the PDL of thedetector can significantly affect the PDL measurement.Moreover, spectral ripple of the power detectors candegrade the measurement quality.

The latest Agilent power meter modules provide theflexibility needed to meet the requirements of differenttest environments. The choice of power sensor module isdriven by the measurement priorities. For the highestaccuracy, the single-channel optical power sensors(81633A, 81634A), or optical power heads (8162xA), arepreferred. The dual-channel power sensors (81635A)provide an economical solution with slightly lowerperformance. The intrinsic PDL of the Agilent 81635Adual-channel power sensor module is specified as typ.±0.015dB. However, the two channels of each powersensor save space in the test environment.

Where the requirement is for the highest possibleaccuracy, optical heads, with their low intrinsic PDL ( typ.± 0.002 dB), provide the best solution. When used with adual-channel interface module, optical heads meetdemands for the highest accuracy while providing aneconomic solution in terms of the mainframe's modulecapacity.

7

Using single -channel power sensors, with their low PDL,can meet a requirement for low uncertainty. However, fora given number of channels, twice as many mainframeslots are required than for a dual-channel solution. Extramainframes may be required to host all the power sensormodules.

Not only the detector, but also every other passivecomponent in the setup can influence the PDLmeasurement. Therefore, to reduce the measurementuncertainty, it is essential to minimize the number ofoptical interfaces and components. Open angledconnectors, for example, have an intrinsic PDL dependenton the angle between the front-end surface and the planenormal to the direction of light propagation. An open 8°angled connector exhibits a PDL of 0.019dB. Used as thefinal connector between the DUT and power metermodule, the intrinsic PDL of the connector influences themeasurement result. This effect cannot be calibrated out,so it is essential to use a straight connector to the powermeter module.

PDL over Wavelength

Most often, the PDL of a DUT at different wavelengthsmust be measured.

Generally, the Polarization Scanning technique can beshown to be best suited for PDL measurement at singlewavelengths, and the Mueller Method for PDLmeasurement over a wavelength range, as shown in Figure12.

PDL Measurement at three points in channel with polarization scanning.

PDL measured over transmission and attenuation bands of one or more channels with Mueller Method and wavelength sweep

Figure 12: Measuring PDL over wavelength withpolarization scanning and Mueller method.

The Polarization Scanning technique exposes the DUT tomany states of polarization, so the PDL can be measuredonly at one wavelength at a time. It is clear that capturingthe PDL of a DUT at many wavelengths can quicklybecome very time-consuming. However, if the PDL is onlyrequired at certain points, such as the center wavelength

or the 3dB bandwidth wavelengths of a passband, thePolarization Scanning technique is sufficiently fast.

Compared to the Mueller Method, the PolarizationScanning technique is relatively easy-to-implement anddoes not involve extensive mathematical calculations,excepting Equation 1. The Polarization Scanning techniqueis the preferred solution for this case.

The 11896A Polarization Controller is specified foroperation in a broad wavelength range (1250nm �1600nm). The fiber-based design of the polarizationcontroller means that wavelength effects can be presumedto be negligible.

The laser source employed depends on the wavelengthaccuracy required and the range of wavelengths ofinterest.

For example, the Agilent 81689A compact tunable lasersource covers a 50nm wide range (1520nm � 1570nm).The lack of continuous sweep capability does not play arole, because the Polarization Scanning technique onlyallows the wavelength range to be covered in steps.Furthermore, the transmission properties of the DUT aremeasured only at specific wavelengths, which need not beequally spaced.

In contrast, the Mueller Method, in conjunction with acontinuous wavelength scan, should be used where anentire channel, or even a number of channels, must becharacterized for PDL. In other words, where there are alarge number of wavelength points with fixed spacing.

The Agilent 81680A tunable laser source, designed forpassive component test in the C-band, is capable ofcontinuous wavelength scan, which decreasesmeasurement time when many wavelength points must bemeasured. In addition, this laser source has high dynamicrange, low power fluctuations over time, as well asoutstanding wavelength accuracy and repeatability.

Despite the advantage in measurement time, using theMueller Method with a continuous wavelength scan istricky.

As mentioned previously, the 8169A PolarizationController synthesizes polarization states using a polarizerand two wave plates, a λ/4 and a λ/2 retarder. Thedesign wavelength of the retarders is 1540nm. Only atthis wavelength do the wave plates act strictly as λ/4 andλ/2 retarders. To produce the four defined polarizationstates for other wavelengths, the settings of the retardersmust be corrected. However, this correction is impossibleduring a continuous wavelength scan. One way to accountfor the wavelength dependence of the generated

8

polarization states is to generalize the equation systemdescribed in Figure 8 in terms of the applied Stokesvectors. However, solving the equation for the Muellercoefficients turns out to be more difficult.

Theoretical investigations have shown that thewavelength dependent retardation of the λ/4 and a λ/2retarder plates is less critical than source output powervariations.

A second complication is that the transmission of the8169A Polarization Controller's polarizer depends on thepolarization of the incident light. During wavelengthscanning, there is a periodic change in the inputpolarization state caused by the retarding property of thefiber. This means that the transmission through thepolarizer is periodic over wavelength, as shown in Figure13.

Reference Measurement

5.00E-05

5.50E-05

6.00E-05

6.50E-05

7.00E-05

7.50E-05

8.00E-05

1.46E-06

1.48E-06

1.50E-06

1.52E-06

1.54E-06

1.56E-06

1.58E-06

1.60E-06Wavelength [m]

Tran

smis

sion

LHP

LVP

L45

RHC

Figure 13: Müller method: Periodic variation in outputpower of the polarization controller.

In principle, using polarization-maintaining (PM) fibershould reduce this periodic variation in powertransmission. The signal generated by the laser is linearlypolarized. This linear polarization state is maintainedbecause the 81640A and 81680A tunable laser sourceshave PM fiber pigtailed optical outputs. If the linearlypolarized light were coupled exactly into a principle axis ofthe PM fiber, the state of polarization would remainconstant. However, the 8169A polarization controllercontains a single mode fiber connection from the opticalinput to the polarizer. Even if a PM fiber were usedbetween the TLS and the polarization controller, avariation of polarization state over wavelength occurswithin the polarization controller itself. Most critical,however, is the coupling into the PM fiber. Unless the PMfibers' principal axes are very well aligned, an evenstronger variation of the polarization state overwavelength arises from the high birefringence of PM fiber.The consequence is a power variation with a shorterperiod and a higher peak-to-peak amplitude change overwavelength. A comparison of the output power over

wavelength between SM fiber and PM fiber are shown inFigure 14. This illustrates a worst-case example of theimpact PM fiber can have on the reference measurement.

Polarizer Output Power with SMF, PMF

-15.5-15

-14.5-14

-13.5-13

-12.5-12

-11.5-11

-10.5-10

1.5E-06 1.5E-06 1.5E-06 1.5E-06 1.5E-06 1.5E-06 1.5E-06Wavelength [m]

Pow

er[ d

Bm

] SMF

PMF

Figure 14: Polarizer output power measured with SM fiberand PM fiber

PM fiber should only be used if the linear polarization oflight from the laser source can be maintained along the PMfiber between the source and the polarization controller.This depends primarily on the alignment of the connectorsto the PM fiber's principal axis. In worst-case scenarios,PM fiber introduces more uncertainty to themeasurements and decreases the quality of the PDLresults.

Power transmission variation does not affect the final PDLmeasurement if it is equal for both the referencemeasurement and the DUT measurement. Provided thefiber from the source to the polarization controller is keptfixed, the evolution of the polarization state overwavelength is constant. However, any movement of thefiber between the two steps of the PDL measurementchanges the polarization transformation characteristics ofthe fiber. This produces a wavelength and amplitude shiftof the periodic power variation that is not reflected by theinitial reference measurement.

The measurement is even more affected by a PM fiber ifthe linear polarization state of light from the laser sourceis not fully coupled into one of the fiber's principal axes.The high birefringence of the fiber results in an increasedsensitivity to environmental changes. A small change intemperature can, for example, translate into significantvariation of power transmission over wavelength.

To reiterate, a small change to the fiber's propertiesbetween reference and device measurement can havenoticeable impact simply because these effects on powertransmission in PM fiber have relatively high amplitude andshort periodicity.

In conclusion, when making PDL measurement using theMueller Method, the reference measurement is not only

9

relevant to recording power variation over polarizationstate, but also over wavelength.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1533 1533.2 1533.4 1533.6 1533.8 1534 1534.2 1534.4 1534.6 1534.8 1535

Wavelength

PDL

Scanning 10msScanning 1ms

Mueller Method

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

1533.5 1533.6 1533.7 1533.8 1533.9 1534 1534.1 1534.2 1534.3 1534.4 1534.5 1534.6 1534.7

Figure 15: PDL measurement of a grating based WDMusing the Polarization Scanning technique and the MuellerMethod.

A sample measurement of a grating-based WDM filterusing the Polarization Scanning technique and the MuellerMethod is shown in Figure 15. The results of the variousmeasurements compare well.

PDL and Insertion Loss

A PDL measurement performed as described here alwaysyields the insertion loss of the DUT. Thus, two propertiesof the DUT can be obtained without additional effort.Furthermore, isolation, or cross talk, can be derived fromthe insertion loss determination. However, the accuratemeasurement of device transmission characteristicsrequires a high dynamic range during the insertion lossmeasurement.

Agilent 81680A and 81640A tunable laser sources providea low SSE output. The source spontaneous emission isattenuated by around 60dB compared to the signal level.Even if the laser signal is attenuated by the filter, and thegreater part of the SSE passes through, the setup is stillable to render a true measurement of the filter�s rejectiondepth.

PDL and Insertion Loss of multiplechannels

TLS

PM

DWDM-MUX (DUT)

PM

PM

PM

Tunable laser

Power meters

PC with plug&play

drivers

Polarization Controller

...

Figure 16: Scheme of a multi-channel setup for PDL andinsertion loss measurements.

Many passive components transform one input signalcontaining many channels into the corresponding numberof output signals containing only one channel, or vice-versa. It is desirable to characterize all output channels inparallel, in other words in one measurement. This reducesboth device measurement and device handling time.

A typical setup for a multi-channel test that includes PDLmeasurement is shown in Figure 16. A tunable laser servesas the optical source. Either the 11896A or the 8169Apolarization controller is employed for signal conditioning.The input line of the DUT is connected to the output portof the polarization controller. The output lines from theDUT are connected to power meter modules.

Automated remote control of the Agilent 816x mainframeseries and related modules ,such as tunable lasers orpower modules, is simplified by employing the VXI plug andplay driver. This driver contains an extensive library whosefunctions can substitute for one or more GPIB commandsand provide parameter check and error handling. Acontinuous wavelength scan can be programmed usingonly three functions. These functions configure andexecute the wavelength scan and yield the measurementresults. All the necessary operations, such as wavelengthlogging or power level stitching, are performed internallyby the driver.

Wavelength logging captures the real wavelength at whicha power measurement is triggered. Measurements are nottaken at exactly the step size that has been set becausethe wavelength scan is continuous. However, the plug andplay driver ensures that power measurements are obtainedat the equally spaced wavelength points required for PDLcalculations.

10

Power level stitching allows measurements to be taken ina maximum of three different power ranges. The plug andplay driver combines the data obtained from the threepower ranges to yield a full characterization of the DUTover a wide dynamic range.

The plug and play driver can easily be integrated intosoftware development environments.

A more detailed discussion of the state-of-the-artcharacterization of optical components can be found in [4].

Interpretation of Measurement Results

The interpretation of PDL measurement results requires adeep understanding of the polarization characteristics ofthe DUT. The properties of integrated optical devices alsodepend on the polarization of the incident light wavesignal. A shift in wavelength of the filter curve is one sucheffect. An arrayed waveguide grating (AWG) is a typicalexample of a device that exhibits a strong wavelengthdependence in its filter curves.

An example of an insertion loss measurement at fourpolarization states is shown in Figure 17. Figure 18uncovers a characteristic of the DUT (an early AWG-typeWDM): the polarization dependency of the filtertransmission curves. The filter transmission curves shift inwavelength for different states of polarization. Moreover,the shape of the transmission curves slightly changes forthe different polarization states. The maximum shift, alsoknown as the TE-TM1 shift, of the filter curves does notcorrelate to the maximum and minimum transmissioncurves. Assume two filter curves (TE and TM), that exhibitmaximum wavelength shift. At a particular wavelength,the curves cross each other; in other words they have thesame power level at a particular wavelength. For otherpolarization states of the incident light wave, the filtercurves are located between the TE and TM curves. As aresult, these filter curves have a different power levelcompared to the TE-TM curves at their crossing point. TheTmax and Tmin curves of the device over wavelength donot correspond to the TE-TM filter curves.

1 TE and TM represent here the two polarization eigenmodes of thewaveguide.

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

1543.00 1543.50 1544.00 1544.50 1545.00 1545.50 1546.00Wavelength [nm]

Loss

[dB

]

LHPLVPL+45RHC

Figure 17: Polarization dependent filter trans-missioncurves of an AWG type filter.

At two extremely small wavelength intervals, the fourcurves cross each other. Translated to the PDL, thiscorresponds to two regions with very low PDL. In turn, atthe slopes of the filter curves, significant power variationsoccur due to the wavelength shift over polarization. Thiseffect results in an increasing PDL. Both effects areclearly visible in Figure 19, where the PDL of the DUT isshown.

-35

-31

-27

-23

-19

-15

-11

-7

1544.25 1544.3 1544.35 1544.4 1544.45 1544.5 1544.55 1544.6Wavelength [nm]

Loss

[dB

]

Crossing regions

High PDL points

LHPLVPL+45RHC

Figure 18: Zoom of transmission curves.

The points with lowest PDL correspond in wavelength tothe crossing intervals of the four filter curves. However,for various reasons, the measured PDL is not zero. Firstly,the filter curves do not cross at exactly the samewavelength. Even if they did, it is questionable whetherthis can be captured within the resolution of the source2.

Also, power noise in the system limits the performance ofthe setup when extremely small PDL values are measured.The finite accuracy and resolution of the setup preventsthe measurement of the ideal zero PDL, if it exists.

In contrast, the steep slope in PDL is caused by powertransmission differences in the wavelength-shifted filtercurves. The PDL spectrum varies from 0.027dB, to more

2 Maximum resolution is 0.1pm using the Vxi plug and play driver Rev.2.51 and higher with the tunable laser sources 81640A and 81680A.

11

than 14dB in the transmission band of the filter. As can beseen in Figure 19, the maximum PDL peak is followed byanother minimum PDL. This PDL point corresponds to thecrossing point of the filter curves at the low end of theirslopes.

0

2

4

6

8

10

12

14

1544.00 1544.10 1544.20 1544.30 1544.40 1544.50 1544.60 1544.70 1544.80 1544.90 1545.00

Wavelength [nm]

PDL

[dB

] 00.025

0.050.075

0.10.125

0.150.175

0.20.225

0.250.275

0.3

1544.40 1544.45 1544.50 1544.55 1544.60

Points of lowest PDL

Steep PDL Slope

Figure 19: PDL of an AWG-type filter channel.

Integrated optical devices can also exhibit strongtemperature dependence. A temperature change of 1 Kelvin may lead to a wavelength shift in the filter curves

of around 10pm. Therefore, it is important to keep theDUT at a constant temperature.

Nowadays, athermic AWG-type filters circumvent anytemperature drifts in the filter curves.

SummaryThe Polarization Scanning technique and the MuellerMethod are suitable methods for measuring thepolarization dependent loss of passive optical components.Both methods can be extended to obtain the evolution ofPDL over wavelength. While the Polarization Scanningtechnique is preferable for determining PDL at a specificwavelength, the Mueller Method has clear advantageswhen PDL must be characterized at numerous wavelengthpoints with equal spacing. This method allows the use of atunable laser source capable of continuously sweeping thewavelength range.

Both techniques yield the insertion loss characteristic ofthe DUT with the PDL measurement, and support parallelmulti-channel testing, so are preferred methods formanufacturing tests.

Agilent Technologies�Test and MeasurementSupport, Services, and AssistanceAgilent Technologies aims to maximize the value you receive, while minimizing your risk and problems. We strive to ensure that you get the test and measurement capabilities youpaid for and obtain the support you need. Our extensive support resources and services can help you choose the right Agilent products for your applications and apply themsuccessfully. Every instrument and system we sell has a global warranty. Support is available for at least five years beyond the production life of the product. Two concepts underlieAgilent's overall support policy: "Our Promise" and "Your Advantage."

Our PromiseOur Promise means your Agilent test and measurement equipment will meet its advertised performance and functionality. When you are choosing new equipment, we will help youwith product information, including realistic performance specifications and practical recommendations from experienced test engineers. When you use Agilent equipment, we canverify that it works properly, help with product operation, and provide basic measurement assistance for the use of specified capabilities, at no extra cost upon request. Many self-help tools are available.

Your AdvantageYour Advantage means that Agilent offers a wide range of additional expert test and measurement services, which you can purchase according to your unique technical andbusiness needs. Solve problems efficiently and gain a competitive edge by contracting with us for calibration, extra-cost upgrades, out-of-warranty repairs, and on-site educationand training, as well as design, system integration, project management, and other professional engineering services. Experienced Agilent engineers and technicians worldwide canhelp you maximize your productivity, optimize the return on investment of your Agilent instruments and systems, and obtain dependable measurement accuracy for the life of thoseproducts.

By internet, phone, or fax, get assistance with all your test & measurement needs

Online assistance: Related Agilent Literature:www.agilent.com/comms/lightwave

Phone or FaxUnited States: (tel) 1 800 452 4844

Canada:(tel) 1 877 894 4414 (fax) (905) 206 4120

Europe: (tel) (31 20) 547 2323(fax) (31 20) 547 2390 Japan: (tel) (81) 426 56 7832(fax) (81) 426 56 7840 Latin America:(tel) (305) 269 7500(fax) (305) 269 7599

Australia: (fax) (61 3) 9210 5947

New Zealand:(tel) 0 800 738 378(fax) 64 4 495 8950

Asia Pacific:(tel) (852) 3197 7777(fax) (852) 2506 9284

Product specifications and descriptions in this document subject to change without notice.

Copyright © 2002 Agilent TechnologiesMarch, 14 2002

5988-1232EN

[1]: �Polarization dependent loss measurementsusing modular test system configurations�, PN5965-5720E, Agilent Technologies[2]: Nyman, B.; Optical Fiber Conference, OFC1994, Technical Digest, page 230, ThK6[3]: �PDL Measurements using the HP8169Apolarization controller�, PN 5964-9937E, AgilentTechnologies[4]: �State-of-the-Art characterization of opticalcomponents for DWDM applications�, PN 5980-1454E, Agilent Technologies