Lightwave Symposium Dispersion Measurements of Presented by Johann L. Fernando, Product Manager 3-1
Topics Chromatic dispersion concepts Agilent 86037C Chromatic Dispersion Measurement System Polarization measurement concepts Agilent 8509C Lightwave polarization Analyzer 3-2
Chromatic Dispersion can limit the performance of optical components and signals. This is a problem at 10 Gb/s and a serious one at 40 Gb/s. When the chromatic dispersion is compensated, polarization mode dispersion (PMD) limits the rate at which pulses can be transmitted over optical fiber. There may be relatively minor effects at on a 2.5 Gb/s signal while 10 and 40 Gb/s signals suffer degraded bit error rates (BER). 3-3
Chromatic Dispersion Measurement Concepts 3-4
Chromatic Dispersion in Optical Fiber A high-speed pulse contains a spectrum of l components. Input pulse Single-mode optical fiber Frequency components of output pulse Red Blue Time Output pulse Time CD causes the shorter l to travel faster than the longer l. 3-5
Why Require Dispersion Measurements? Relationship of bit rate to dispersion 2.5 Gbit/s Fiber optic transmission system 10 Gbit/s Broader spectrum Narrower bit slot More spreading More sensitive to spreading CD becomes a serious problem at 10 Gb/s and beyond and leads to a a higher bit error rate (BER) Acceptable 1 amount of CD Bitrate 2 3-6
How Much Dispersion can a Link Tolerate? Assumptions: 1dB dispersion penalty Standard single-mode fiber (S-SMF) with D c = 17 ps/nm-km External modulation (negligible transmitter chirp) Disp. coeff. D ps Length, km nm-km L c = 10 B 5 2 Bit rate, Gb/s R.M.Jopson, Bell Labs, Lucent Technologies Measurement issues for dispersion compensation Symposium on Optical Fiber Measurements, SOFM 2000. Bit rate 2.5 b/s 10 Gb/s 40 Gb/s Max. dispersion (ps/nm) 16,000 1,000 63 Bit rate 2.5 b/s 10 Gb/s 40 Gb/s Max. link length (km) 941 59 4 The tolerable dispersion is inversely proportional to the square of the bit rate. Why? First, higher rates mean broader spectra and more pulse spreading. Second, the narrower bit slots are more susceptible to the spreading of neighboring pulses. The combination of these two effects produces the inverse square relationship. The upper table shows the tolerable dispersion for several bit rates. For 10Gb/s systems, the net dispersion is compensated to several hundred ps/nm. Note the much tighter demands of a 40Gb/s system. The lower table shows the allowable length of single mode fiber for the various bit rates. This is why chromatic dispersion compensation is needed. Note that at 10Gb/s barely one span of single-mode fiber can be tolerated without compensation! 3-7
Speed of Light Versus Speed of the Signal Phase velocity Group velocity Find this simulator at: http://www.phys.virginia.edu/classes/109n/more_stuff/applets/sines/groupvelocity.html Chromatic dispersion measurement systems of the type that Agilent provides characterize the relative group delay variation of the test device. What is group delay? In most high-speed digital systems, intensity is modulated in a way similar to this illustration. The intensity of the fundamental lightwave oscillations (red wave) is coded with information, in this case a simple sinusoid (very little information in this example!). The point of this slide is that two velocities are involved. The first and most basic is the phase velocity of the carrier, that is, the speed with which a crest of the carrier wave travels along the link. The second velocity is the group velocity, that is, the speed with which the crest of the modulation envelope travels along the link. In a dispersionless medium, the phase and group velocities are identical. In the typical fiber system, the group velocity is somewhat slower. Lightwave system designers care about signals, so dispersion measurement and management focus on the group delay characteristics of components and fibers. 3-8
Chromatic Dispersion Definitions CD measurements characterize how the velocity of propagation of a light pulse change with wavelength. Group delay Propagation time for a modulated lightwave Relative group delay t g (ps) CD Coeff. Is the slope of the relative GD curve Chromatic dispersion coefficient D (ps/nm-km) Zero-dispersion wavelength Dispersion slope D = 1 L t g l Chromatic dispersion is described by several internationally-recognized definitions. We review them now: Group delay is the propagation time for a modulated lightwave. If we were able to put a dot on the corner of a pulse (or any modulation envelope ), the group delay would be the time for that dot to travel through the test device. Relative group delay is the change in group delay from wavelength to wavelength. In narrowband device testing, this is the major concern. The chromatic dispersion coefficient is the slope of the relative group delay curve. The graph allows us to calculate the time spreading for a signal of a given spectral width. At the zero-dispersion wavelength, two closely-spaced frequency components travel at approximately the same speed. The dispersion slope tells us how the dispersion changes between channels of a DWDM system. 3-9
Dispersion Compensating Fiber (DCF) l oper Conventional single-mode fiber l oper Dispersion compensating fiber The most common way to compensate dispersion is to cascade the transmission fiber with a dispersion compensating fiber module. The dispersion compensating fiber compresses the wavelength components back together in time, restoring pulse shape. 3-10
Chirped Fiber Bragg Gratings Input pulse Frequency components reflect at different locations along the grating. c ir c u l a t o r Output pulse An alternative to DCF is the chirped fiber Bragg grating. A dispersed input pulse passes through the circulator to the grating, where different wavelength components reflect at different locations. Each wavelength component experiences a different delay, compensating the pulse. The pulse then passes through the circulator to the output. A fiber grating is formed by exposing a doped fiber to an intense, patterned UV light. The pattern of the mask is written into the fiber as a series of index variations. For a filter, the spacing of these variations is relatively uniform. Chirped gratings have a gradually changing pattern from end to end. The result is that different wavelengths travel different distances before reflection. 3-11
Compensating Chromatic Dispersion Compensate channels individually DWDM demux Dispersion compensator Dispersion compensator Dispersion compensator Dispersion compensator Compensate all channels simultaneously Typically NZDSF Dispersion slope compensation Trim S-SMF DWDM demux Fortunately, chromatic dispersion is a stable effect that can be compensated. There are two basic approaches to compensation in DWDM systems. First, we can individually compensate each DWDM channel (top figure). This method is expensive and complex. Secondly, we can compensate all channels together, before wavelength demultiplexing. This approach must include dispersion slope compensation. We will review several of the methods used for chromatic dispersion compensation 1. 1 Dispersion compensation techniques have many other uses, including preand post-compensation of dispersion, distributed dispersion management in linear and soliton systems, narrowing of pulses for RZ transmission and decorrelation of pulse streams in DWDM channels for transmission experiments. 3-12
CD Measurement using the 86037C Measurement system 3-13
Chromatic Dispersion Measurement Methods Modulation Phase Shift Method - Agilent More accurate and repeatable measurements Industry standard for measuring CD Differential Phase Shift Method Accuracy not as good due to dither of source Interferometric Method Accuracy not as good as modulation phase shift method 3-14
Modulation Phase Shift (MPS) CD Measurement* Used in all of Agilent s CD measurement solutions Fundamentally accurate (no adjustments ) Wavelength traces to the HeNe laser in the wavelength meter Time traces to the crystal time base in the network analyzer Direct measurement of relative group delay A polynomial curve fit improves D, l0, S 0 accuracy (fiber test) Measurement of absolute group delay (useful for device modeling) * Differs from the differential phase shift method in which the dispersion value is calculated directly from the phase shift produced by modulating the laser wavelength. 3-15
Agilent 86037C CD Measurement System Test Set Agilent 816XX Tunable laser Bias Modulator -10 db Device under test DAC out Computer Agilent 8753E RF Network Analyzer S R A -16 db Optical receiver GP-IB Agilent 86120C Wavelength meter 3-16
How Relative Group Delay is Determined Modulator drive signal Recovered modulation envelope Phase at l n-1 (adjusted to zero) Df Phase at ln Phase detector Compute relative group delay from: D Dt g = f 360 1 fmod Df 3-17
Test and Preview Modes New Phase samples Tunable laser wavelength Preview mode Very good accuracy Fastest measurement Test mode Best accuracy Time In Test mode, a number of phase measurements is performed at each l. In Preview mode, phase is measured periodically as the l is continuously swept. 3-18
Typical Measurement Result (DS fiber) 3-19
The Reference Wavelength Feature (stepped wavelength mode only) Phase measurement wavelength Phase samples l ref feature OFF l 6 l 5 l 4 l 3 l 2 l 1 Time Temperature-induced phase error Temperature Phase measurement wavelength l ref l 1 l ref feature ON l 2 l 3 l 4 Correction df l 5 l 6 Time 3-20
Example of the Impact of Modulation Frequency 400 Comparison of 2.5 GHz and 650 MHz modulation frequencies Group Delay (ps) 300 200 100 0-100 1549.7 1549.8 1549.9 1550 1550.1 1550.2 1550.3 Wavelength (nm) MPSM (2.5 GHz) MPSM (600 MHz) 3-21
Group Delay Effect of Modulation Frequency on Wavelength Resolution Example: With a modulation frequency of 2 GHz and small step sizes, this measurement is smeared. 4 pm step size The difference between the true group delay amplitude and the measured amplitude is the smearing error. RIPPLE lc 2 GHz = 32 pm sideband separation λ 3-22
Group Delay Effect of Reducing Modulation Frequency Reduction in modulation frequency increases wavelength resolution but 4 pm step size The difference between the true group delay amplitude and the measured amplitude is greatly reduced. RIPPLE lc λ 100 MHz = 1.6 pm sideband separation 3-23
Selecting the Best Wavelength Resolution for your DUT f c -f m +f m RGD Wavelength step nm GHz Modulation frequency MHz actual pm Optical fiber nm Narrowband devices pm Wavelength resolution Narrowband components often show rapid variations of relative group delay over their narrow channel wavelength range, as shown in the sketch at the far left in this slide. Such devices require small wavelength steps. Less widely understood is the fact that such devices may also require lower modulation frequency. If the modulation frequency is too high (sidebands too far out), the hardware is unable to resolve the fine variations in relative group delay. One benefit of the preview mode is that it allows fast measurement of variations of RGD. If necessary, the measurement can be repeated in step mode for best accuracy. 3-24
Thin Film Interference Filter 250-10 Group Delay (ps) 200 150 100 0.45 nm passband -15-20 -25 Loss (db) 50-30 0-35 1554.4 1554.5 1554.6 1554.7 1554.8 1554.9 1555 1555.1 1555.2 Wavelength (nm) This is a measurement of a thin-film interference filter of the type that is used in a DWDM multiplexer. Unlike fiber measurements, we do not fit a curve to component dispersion measurements. The reason for this is that the group delay data is complex and unique. It is of greater value to determine the variation of relative group delay, and express it in some form such as deviation from linear group delay. Many users export the relative group delay data and post-analyze it using other software tools. The 86037C provides a feature for measuring the absolute group delay, which is of use when the performance of certain types of gratings is to be compared with theoretical predictions. 3-25
Dispersion Compensating Fiber Bragg Grating Group Delay (ps) 168500 168400 168300 168200 168100 168000 167900 167800 167700 0.45 nm Operating range -10-15 -20-25 -30-35 -40-45 Loss (db) 167600-50 1550 1550.2 1550.4 1550.6 1550.8 1551 1551.2 1551.4 1551.6 1551.8 1552 Wavelength (nm) These days, fiber and components require relative group delay measurements. This example shows a measurement of a dispersioncompensating fiber-bragg grating. Wavelength step and modulation frequency are reduced to show the fine details of group delay variation. The modulation frequency in this case was 100 MHz. 3-26
Chirped FBG Measured in Step and Preview Modes Relative group delay 600 400 200 0-200 -400 Run 34, Step, 201 pts, 2 GHz Run 35, Step mode, 26 pts, 2 GHz Run 41, Preview, 10 pm, 2 GHz Data is offset vertically to coincide at 1554.1 nm -600 1553.7 1553.8 1553.9 1554 1554.1 1554.2 1554.3 1554.4 1554.5 Wavelength (nm) This graph compares measurements of a chirped fiber Bragg grating taken in the Step and Preview modes. The device is measured over its reflection passband, through an optical coupler. Agreement of the measurements is very good, the special value of the preview mode is that it can measure over a thousand wavelengths in less than a minute. 3-27
Polarization Mode Dispersion Measurement Concepts 3-28
Types of Dispersion in Optical Fiber Optical paths Optical frequencies Polarization modes ν 1 ν 2 (first order only) Modal dispersion (MMF) Chromatic dispersion Polarization-mode dispersion Signal components arrive at slightly different times Input pulse Output pulse (approx) Three dispersive phenomena degrade high-speed system performance by broadening the digital pulses and making the one/zero decision process less reliable. Modal dispersion occurs because light splits into many spatial paths, each having a different length and thus a different arrival time. This causes a pulse to spread. This affects multimode systems only. Chromatic dispersion arises from the waveguide and material properties of single-mode fiber. Pulse spreading is caused by the difference in propagation speed (and therefore arrival times) for different frequency components of the signal. Polarization-mode dispersion (PMD) becomes a performance limitation in high speed systems when chromatic dispersion is compensated by special fibers or devices. Pulse spreading is caused by the difference in propagation velocity between orthogonal polarization states. 3-29
Polarization Mode Dispersion (PMD) DUT Dt Dt - Differential Group Delay (DGD) * PMD spreads pulses by breaking them into two polarization modes that have slightly different speeds of propagation. * The relative delay between the two mode is known as DGD 3-30
Birefringence: The Root of Polarization Mode Dispersion (PMD) Example: Linearly birefringent crystal. Birefringence is the difference in refractive index between orthogonal polarization states. n slow Slow axis Fast axis n fast Birefringence = n slow - n fast Single-mode fiber 10-7 Polarization-maintaining fiber 10-4 3-31
Intrinsic PMD of Single-Mode Fibers Single-mode fiber (SMF) PMD is caused by polarization dependence of the index of refraction. (birefringence). Ideal core Oval core Slow axis This has two primary origins: Form birefringence characteristic of a non-circular waveguide. Stress birefringence due to forces set up by a non-circular core. The PMD of short fibers (meters in length) increases linearly with fiber length. 3-32
PMD Effects on Optical Digital Transmission Degraded Eye Diagram Birefringent Device D t 2.5Gb/s Bit Period Open Eye Diagram 10Gb/s Bit Period Note: Same Dt has greater impact on signals with higher bit rate 3-33
PMD Tolerance vs. Bit Rate 100 DGD (ps) 80 60 40 80ps 20% of a signal s Bit Period 10% of a signal s Bit Period 40ps 20ps 20 10ps 10ps 5ps 2ps 0 5ps 2.5 10 20 2.5p 40 100 s Bit Rate (Gb/s) 1ps 3-34
PMD Measurement Using The 8509C Lightwave Polarization Analyzer 3-35
The 8509C Lightwave Polarization Analyzer (main window) 3-36
8509C Measurement Menu and System Capabilities Propagation time difference between fast and slow polarization modes of test device Amount of variation in the insertion loss of a component over all polarization states Direct measurement of Jones matrix Measurement of the polarization crosstalk of PM fiber and components Measure angles between markers on Poincare sphere Polarization data as a function of time 3-37
Agilent 8509C LPA Jones Matrix Eigenanalysis PMD Measurement Configuration PMD measurement is done automatically. The 8509 tunes the TLS in wavelength steps via an HP-IB link. 8509B/C Lightwave Polarization Analyzer Agilent Tunable Laser Source HP 8509B Lightwave Polarization Analyzer HP-IB External Source Input Optical Output Optical Input Coupler Optical Cables DUT Agilent Multi-Wavelength Meter (Optional) Device Under Test 3-38
Agilent 8509C Instrument Block Diagram Internal Polarizer Unit Polarimeter Control circuitry 8509B Wavelength meter External optical input Tunable laser Polarizer & shutter Optical output DUT 32-state polarization controller Optical input Computer 3-39
8509C Internal Polarizer Unit External source input (from TLS) Internal polarizer unit Front panel knobs Polarization adjuster Polariz. ellipse Lens Linear Polarizers oriented at 60-degree increments (angles measured and stored) Polarizers A B C -3dB Shutter Lens Optical ouput (to DUT) Front panel knobs are adjusted to obtain circular polarization at the internal polarizer unit 3-40
The JME PMD Method (Jones Matrix Eigenanalysis) @ each λ, 3 linear polarizers are sequentially inserted in the signal path λ, λ... λ Linear polarizers: 1 2 N Tunable narrowband source Pol. adj. A B C 0 45 90 deg Fiber or component Test fiber B Polarimeter A C B * Output states corresponding to the different polarizer angles are measured * From this Data and from the known polarizer angles the system computes JME for each λ. JME measurement method determines the DGD DGD Wavelength 3-41
Fiber PMD Measurement (JME) PMD Delay 3-42
PMD of a DWDM multiplexer (AWG design) JME method DGD (ps) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 10 pm steps 25 pm steps 25 pm steps 50 pm steps 0.9 nm 0 1549.3 1549.5 1549.7 1549.9 1550.1 Wavelength (nm) 3-43
The Wavelength Scanning (Fixed Analyzer) method Example: spooled fiber * Not as accurate as JME method * Faster measurement 3-44
Reduce Test Time by Selecting Wavelength Step Test device is a DCF module overlay traces 1 nm steps 81 seconds 162 seconds 1/2 nm steps The JME measurement method produces valid results over a wide range of step sizes, bounded at small step size by electrical noise from the polarimeter and at large step size by undersampling of the DGD versus wavelength curve. Within this range, the results are relatively independent of step size. This slide shows a measurement of a dispersion compensating fiber module, measured at 1 nm and 0.5 nm step sizes. The average DGD is shown at the lower left of each screen. The traces are overlayed at the top of the slide. In such a case, it makes sense to use the larger step size and save test time. 3-45
Step size, Resolution and Repeatability Fiber spool, JME method Step: 0.25 nm Mean DGD: 0.099 ps Step: 1.0 nm Mean DGD: 0.098 ps Instrument noise has greater impact when steps are very small. Step: 5.0 nm Mean DGD: 0.079 ps Large steps miss the fine details of DGD(l) This slide shows how to recognize when the step size is too large or too small for measurement of optical fiber PMD. Note the shape of the center trace. This is typical for a good fiber DGD measurement. Note that the trace is relatively smooth and the peaks and valleys have a rounded shape. In this example, the wavelength resolution is as fine or finer than necessary to get a reliable average DGD value. The bottom trace shows a measurement of the same device with too large a wavelength step. The keys to recognizing this condition are that peaks and valleys are represented by one or two data points, and that the curve looks like a piece-wise linear approximation to the kind of smooth curve we expect from fiber PMD (middle trace). A qualifier to this advice is that fiber with extremely high PMD has very closely spaced peaks and valleys that appear to be sharply peaked when the data is viewed over a wide wavelength range. If this is the case, drop the step size and perform the measurement over a narrower wavelength range. The top trace is taken with a very small step size. As a result, the polarization shift per step at the output of the device is very small and the noise of the polarimeter can overwhelm the measurement. Note the white noise look of the data. The average DGD from such a trace may be quite useful, but the measurement takes much more time than necessary and some features may be obscured by noise. 3-46
Lightwave Symposium Dispersion Measurements of 3-47