High Resolution Optical Spectrum Testing System Based on LabVIEW

Transcription

1 High Resolution Optical Spectrum Testing System Based on LabVIEW Yichi Zhang 1,2, Changjian Ke 1,2,*, Deng an 1,2, Deming Liu 1,2 1,2 National Engineering Laboratory or Next Generation Internet Access System, 4374, Wuhan, China 1,2 College o Optical and Electronic Inormation, Huazhong University o Science and Technology, 4374, Wuhan, China * cjke@mail.hust.edu.cn Abstract A LabVIEW based spectrum testing scheme o optical passive device is proposed. Relying on LabVIEW programming control over tunable laser, programmable optical ilter, optical power meter and data acquisition () platorm, spectrum characteristics o various transer unctions can be obtained. Setting laser sweeping speed as 1nm/s, sweeping range as 1 nanometers, and sampling rate as 1MS/s, spectrum measurement or one device can be inished in 1 second, with spectrum resolution o 1 picometer. Compared with conventional spectrum testing method utilizing broadband light source and optical spectrum analyzer (), the testing perormance o proposed scheme approaches that o conventional one with even more delicate spectrum details. The experimental results show that our proposed scheme can be applied or spectrum measurement with the demand o high resolution, ast speed and high eiciency. Keywords Spectrum testing; Optical passive device; Tunable laser; LabVIEW; Data acquisition I. INTRODUCTION Optical spectrum analysis is one o the most undamental tools in optical communication systems. Optical spectrum analyzers () are widely deployed in the diagnostic o transmitted optical signals, in the characterization o transer unctions o various kinds o active and passive optical devices, and in the monitoring o optical communication networks [1, 6-7]. In recent years, with the dramatic improvement o transmission rates and development o novel modulation ormats, high resolution optical spectrum analysis is becoming increasingly necessary [2]. Currently, one way to acquire high resolution spectrum is relying on optical heterodyne technique, where the signal under test intererences with a local oscillator, subsequently radio requency domain spectral analysis is perormed [3-4]. Another promising method is based on stimulated Brillouin scattering between a swept tunable laser and the signal under test [2, ], which is also adopted in this paper. Since the acquired optical spectrum inormation needs to be urther processed in the electrical domain, high perormance, real-time hardware circuit is also an essential element o the integral high resolution optical spectrum analyzer [6-1]. We utilized the data acquisition card () o National Instrument (NI) to acquire the mass data inormation account or the ollowing reasons. On one hand, s are high speed circuits which satisy the need o real-time processing. On another, LabVIEW is a powerul sotware development environment that can easily control the hardware and interaces, process the acquired data low with predeined algorithms, and provide end users with easy-to-use graphical user interace (GUI). This article is mainly ocusing on the electrical-domain hardware and algorithm realization o the proposed high resolution optical spectrum analyzer. II. MEASUREMENT RINCILES & EXERIMENTAL SETU A. Measurement rinciples The signal under test (SUT) is injected to the high nonlinear iber (HNLF), while the pumping wave, namely, the tunable laser signal, is propagating along the iber in an opposite direction. The optical iber is utilized as the medium to generate the Brillouin scattering ater the interaction o the pump and SUT. The pump provides extremely narrow spectral window to pick up and ampliy the corresponding spectral component o SUT. While the pump is tuning in a certain spectral range, the spectrum o SUT can be retrieved rom the non-linear component o the Brillouin scattering eect [2, ]. Such a process can be simpliied as a tunable laser passing through a certain optical passive device with the spectrum eature we are interested in (equivalently SUT), which can be mathematically regarded as a convolution between the tunable laser and the ilter proile o the optical passive device [11-12]. And the convolution output is just the counterpart o the nonlinear component in the Brillouin scattering based process, which can be illustrated in Fig. 1. Hence, this paper investigates the property o the above simpliied process. A more detailed and integral structure considering photo-electrical conversion and urther signal processing is depicted in Fig. 2, where the whole process can be separated into optical and electrical domain. v SUT tunable laser source convoluted output * Fig. 1. Simpliied process o Brillouin scattering based optical spectrum analyzing method This work is supported by the National High Technology Research and Development rogram (863 rogram) o China (Grant No.213AA1343)

2 Deconvolution Smoothing In Fig. 2 above, e( ) is the emitting power o tunable laser, H( j) is the overall transer unction o ilter with certain spectrum eatures, photo detector and current-voltage converting circuit. Here we represent the transer unction o optical ilter, photo detector and current-voltage converting circuit by H ( j ), Hd ( j) and Hc( j ), respectively. Unit impulse responses o H( j) and H ( j ) can be represented as h( ) and r( ), respectively. v( ) is the output signal passed through optical ilter, photo detector and the converting circuit, which has the orm o V( j) in requency domain. Obviously, H( j) H ( j) H ( j) H ( j) d c RZ H ( j) where R and Z are constants, the ormer one means the responsivity o detector, and the latter one means the resistance value o converting current into voltage. Since in temporal domain (since the wavelength value o tunable laser is proportional to time), we have: where * represents convolution. (1) v( ) e( )* h( ) (2) Ater converting (2) into requency domain, and considering (1), we can obtain (3) as ollowing. where e( ) Optical Domain H( j) v( ) V ( j) E( j) H ( j) ( e( )) H( j) =RZ ( e( )) H ( j) represents Fourier Transorm. Ideally, laser linewidth is considered to be Hz, thereore: where represents the delta unction. Then equation (3) can be modiied as Electrical Domain v ( ) 1 v ( ) 2 Fig. 2. Detailed structure o the simpliied process (3) e( ) ( ) (4) V( j) RZ H p( j) () From (), we can know that under the condition o omitting the eect o laser linewidth, V( j ) is proportional to Hp( j ). Thereore output voltage generated ater passing through optical ilter, photo detector and converting circuit can describe the spectrum characteristics o optical ilters. As a consequence, in this practical testing system, i the laser linewidth is narrow enough compared with the bandwidth o the ilter, which is shown in (6), BW (6) where BW means the bandwidth o optical ilter, then () realizes. I the above assumption does not hold true, the output spectrum would be broadened severely [6]. Besides, the most signiicant source o internal noise in the optical analyzing system attributes to the photo detector [13]. The convolution output inevitably suers rom noise disturbance, which leads to the unpleasant luctuation along spectral axis. The above two actors contribute to the distortion o output spectrum, which is qualitatively shown in Fig. 1. Hence, in order to eliminate the eect o convolution and stochastic noise, the acquired optical spectrum needs to be urther processed in the electrical domain. In the rightmost part o Fig. 2, the retrieved optical spectrum is processed with deconvolution and smoothing algorithms to inhibit the side eect o convolution and random noise, respectively. Deconvolution is the inverse process o convolution, which aims at restore the broadened spectrum o the optical ilter. In (3), i the inverse transer unction o ( e( )) is obtained, then we can get the Fourier transorm o v ( ) as: 1 V j V j e RZ H j (7) ( ) ( ) 1 ( ( )) ( ) 1 where H ( j) is the approximation o the spectral eature o the optical ilter, however the data low ater deconvolution still suers rom noise disturbance. Subsequently, v ( ) 1 has to be divided into small portions where averaging is perormed among elements in each portion, then the output spectrum v 2 ( ) is much smoother compared with v 1 ( ).Such an operation is essential because it eliminates the uncertainty brought in by random noise, which improves the idelity o approximating the real spectrum. B. Experimental Setup The layout o high resolution optical spectrum measuring system proposed in this article is shown in Fig. 3. The tunable laser type is Santec TSL1, with maximum output power up to 2dBm, laser linewidth o about khz measured by Agilent E4447A spectrum analyzer. In order to veriy the testing perormance o the proposed method or dierent optical spectrum eatures, we choose a programmable optical ilter, Finisar WaveShaper 4S, to simulate spectrum characteristics o certain optical passive devices, which can be realized by predesigning optical transer unctions o WaveShaper S4. The photo-electrical converting section is realized by a Newport 1936-C power meter, equipped with a 818-IS-1detector, it can detect low level optical power precisely. Data acquisition system is the platorm provided by National Instruments, a windows embedded XIe-8133 controller and a XIe122 data acquisition card with sampling rate up to 2 MS/s are inserted into the XIe-182 metal classis, which can be seemed as a special purpose computer.

3 NI-SCOE.vi Split 1D Array Function.vi Mean.vi Waveorm Graph.vi Write wo Measurement File.vi 1_set_Trigee routputstep.vi 1_set_Trigee routput.vi 1_set_Sweep.vi 1_set_Sweep State.vi Optical In Optical Out Electrical In Electrical Out Santec TSL1 Newport 1936-C WaveShaper 4S 818-IS-1 Multiport Optical rocessor Finisar Detector Tunable Laser Optical assive Filter Detector ower Meter Electrical Trigger Command Line National Instruments NI IXe NI IXe NI IXe NI IXe NI IXe NI IXe-8133 Display Data Acquisition latorm VGA Fig. 3. Layout o spectrum measurement system The platorm is equipped with GIB, RS-232 and USB interaces, LabVIEW programs run on Windows X operation system installed on XI-8133, which coordinate working conditions between internal data acquisition card and external devices precisely via communication interaces. Meanwhile, analog electrical signal will be converted to digital domain. Algorithms such as smoothing and deconvolution are perormed in this unit. In this whole system, the tunable laser, optical passive devices, optical power meter and data acquisition platorm are interconnected via optical or electrical cables. To be more speciic, regarding Fig. 3, the working state o tunable laser TSL1 is set by receiving command rom XIe-8133 controller through a GIB interace. Wavelength sweeping range is rom a certain starting wavelength to an ending wavelength, with maximum sweeping range o nm starting rom 11nm to 163nm, which covers the whole C and L band. TSL1 can sweep in continuous mode and step mode. Here in our experiment, in order to achieve high speed testing perormance, we require the laser to work in continuous sweeping mode. In such a mode, sweeping speed can also be set, with maximum sweeping speed up to 1nm/s. Take the issue o synchronization with platorm into consideration, we set laser sweeping speed as 1nm/s. Once TSL1 starts to sweep, a pulse-shaped trigger signal is generated immediately rom the BNC interace on the rear panel o tunable laser, with maximum voltage o 3.11 Volts. The trigger signal is used as a messenger to activate the behavior o data acquisition o XIe The optical signal generated by TSL1 is then sent to WaveShaper 4S. Ater passing through a certain optical transer unction, the output signal is received by detector 818- IS-1 attached by power meter 1936-C, during which optical power is converted into current through detector, and then current is converted into voltage through power meter. And the analog voltage output can be retrieved rom BNC interace on the rear panel o 1936-C. Once received trigger impulse rom tunable laser, XIe122 starts to acquire data continuously with sampling rate o 1MS/s, ater that the obtained data are processed and then displayed. Then considering the sotware programming located in the platorm, LabVIEW is used to control the wavelength sweeping mode o tunable laser, the working condition o data acquisition card and the processing o obtained data low. Fig. 4 is the block diagram illustrating the sotware unctionality. Fig. 4. Block diagram o LabVIEW program The uppermost part is the realization o laser control via GIB interace. According to the user s command, working parameters o the tunable laser can be set such as what the trigger mode is, how the sweeping condition is, and when the sweep starts. Speciically, Users can modiy sweeping parameters such as starting wavelength, ending wavelength, sweeping speed and the timing o trigger generation. And then such parameters are read by VIs such as 1_set_Sweep.vi and TSL1_set_SweepState.vi in block diagram, altogether acting as hardware drivers o TSL1. Here we set the sweeping range rom 14nm to 1nm, and sweeping speed is set to be 1nm/s. The bottom part o this chart illustrates how the acquired data are processed through.vi modules. Ater the reception o trigger signal rom the tunable laser, data acquisition card immediately obtain 1,, points o data with sampling rate o 1MS/s, and then the obtained data low is processed and displayed. Note that the requency separation between any two adjacent sample points is.1pm, and the laser linewidth o tunable laser is around khz, namely 4m in the dimension o wavelength, considerably narrow compared with the bandwidth o our testing ilter spectrum o up to several nanometers, which means the linewidth has almost no eect on any two sample points. Hence, deconvolution is not a prerequisite or urther processing. Besides, the deconvolution algorithm is time-consuming, which hampers the real-time display o tested spectrum. So we can make a modiication o signal processing in Fig. 2 by deleting the section o deconvolution. And the subsequent smoothing algorithm can still work well. So here we irst utilize NI-SCOE Express.vi to obtain 1,, points o data by converting analog signals into discrete ones; Then the massive discrete points are divided into 1, proportions with length o 1 points by Split 1D Array Function.vi, which means the 1nm range is divided into 1, segments corresponding to 1, values o wavelength, hence the spectrum resolution o.1nm can be achieved; In every segment, the irst points are used or averaging, which is realized by Mean.vi, and the mean value is chosen as the received signal power o the current wavelength; Ater that, Waveorm Graph.vi is applied to display the

4 devices, and then send the output signal to traditional diraction grating based optical spectrum analyzer Agilent 8614B, the recorded spectrum characteristics are shown as red curves () rom Fig. 6(a) to Fig. 6(c), respectively, rom which we can see that, spectrum patterns obtained by method and method match reluctantly. That is because smoothing algorithm has not been perormed, which leads to the situation that spectral components with low power level are aected by noise strikingly. processed output data low; Finally, Write to Measurement File.vi to store the processed data into.txt or.lvm ormat or analysis. III. Fig. Front panel o the optical spectrum testing system SECTRUM MEASURING EXERIMENTAL RESULTS The ront panel provides user with an intuitive and operable graphical user interace (GUI), which is illustrated in Fig.. In order to veriy the spectrum measuring perormance o various types o optical ilters, we design rectangular, Gaussian and periodically sine-shaped ilters with the assistance o WaveShaper 4S. In Fig. 6, we set isolation=22db or dierent optical passive devices. Without perorming smoothing algorithm, spectral characteristics obtained by data acquisition system o dierent optical passive components are shown as black curves () rom Fig. 6(a) to Fig. 6(c). Meanwhile, or intuitive contrast between our proposed spectrum measuring method and conventional one, we utilize LIGHTCOMM C-band ASE light source to pass through WaveShaper 4S when simulating dierent optical passive Ater perorming smoothing algorithm, spectrum testing perormance o and methods can be shown in Fig. 6(d) to Fig. 6(). Considering the rectangular situation, our scheme has a satisied perormance acing spectrum mutations. And rom the comparison between Gaussian and periodically sinusoidal situations, we can make the conclusion that our method is able to identiy detailed inormation o spectral eatures. We believe that the proposed LabVIEW based testing scheme has the perormance approaching to the conventional one rely on broadband light source and diraction grating based optical spectrum analyzer with even higher resolution. Although there are luctuations around bottom part o the obtained spectrums, the overall trends o black curves can represent real spectrum characteristics under such experiment conditions. The bottom luctuation attributes to the restriction rom detecting lower optical power. Such phenomenon is caused by detector and data acquisition card introduced noise. Since the accuracy o XIe122 is 14 bits or A/D conversion, weak signals cannot be distinguished rom such thermal noise. As a result, our scheme has a limitation on detecting a certain low power level signal, so the luctuation o bottom part o optical spectrum by method is the representation o inaccurate detection o optical power (a) (b) (c) (d) (e) () Fig. 6. Spectrum measuring perormance beore and ater smoothing Beore smoothing: (a) Rectangular (b)guassian (c) Sinusoidal; Ater smoothing: (d) Rectangular (e)gaussian () Sinusoidal

5 Hence, we can make the conclusion that our proposed LabVIEW based testing scheme has the perormance approaching to the conventional one rely on broadband light source and optical spectrum analyzer. Besides, due to ast sweeping speed o tunable laser and data processing ability o data acquisition platorm, time duration o single sweep can be reduced dramatically compared with that using conventional method. As a consequence, the eiciency o spectrum testing is enhanced. IV. CONCLUSIONS Spectrum measurement system based on LabVIEW is realized in this article, which has high resolution and approaches the perormance o conventional spectrum measuring method utilizing broadband light source and optical spectrum analyzer. The LabVIEW based spectrum testing method is suitable or spectrum characteristic discretion o optical passive devices. REFERENCES [1] John G. Webster, Measurement, Instrument, and Sensors Handbook, Boca Raton: CRC ress LLC, 1999, pp [2] S. renssler, A. Zadok, A. Wiatrek, M. Tur, and T. Schneider, Enhancement o spectral resolution and optical rejection ratio o Brillouin optical spectral analysis using polarization pulling, Optics Express. vol.2, no. 13, pp , June 212. [3] B. Szaraniec, J. Y. Law, and D. M. Baney, Frequency resolution and amplitude accuracy o the coherent optical spectrum analyzer with a swept local oscillator, Optics Letters. vol. 27, no. 21, November, 22. [4] D. M. Baney, B. Szaraniec, and A. Motamedi, Coherent Optical Spectrum Analyzer, IEEE hotonics Technology Letters. vol. 14, no.3, pp [] J. M. Subias Domingo, J. elayo, F. Villuendas, C. D. Heras, and E. ellejer, Very high resolution optical spectrometry by stimulated Brillouin scattering, IEEE hotonics Technology Letters. vol. 17, no. 4, pp. 8-87, April 2. [6] W. Jiang, C. Ke, and D. Liu, Optical component auto-test system based on VC and GIB, Optics & Optoelecronic Technology. Wuhan: vol. 4, no. 3, pp. 43-4, June 26. [7] Q. Cao, D. Liu, W. Jiang, and C. Ke, Design o automatic measurement system or optical devices on GIB, Study on Optical Communications. Wuhan: no. 2, pp. 48, April 24. [8] T. Kawanishi, T. Sakamoto and M. Izutsu, Fast optical requency sweep or ultra-ine real-time spectral domain measurement, Electronics Letters. Stevenage: vol. 42, no. 17, pp , August 26. [9] X. Yi, Z. Li, Y. Bao, and K. Qiu, Characterization o assive Optical Components by DS-Based Optical Channel Estimation, IEEE hotonics Technology Letters. New Jersey: vol. 24, no. 6, pp , March 212. [1] X. Yi, Z. Li, Y. Bao, and K. Qiu, Ultra-ast measurement o passive optical components by optical channel estimation, 1 th International Conerence on Optical Communication and Networks (ICOCN 211). Chengdu: pp. 1-2, november 211. [11] M. Azadeh, Fiber Optics Engineering, New York: Springer, 29, pp [12] A. D. McAulay, Military Laser Technology or Deense: Techonology or Revolutionizating 21 st Century Warare, New Jersey: John Wiley & Sons, 211, pp [13] Optical Spectrum Analysis: Application Note 1-4, Agilent Technologies, pp. 2-21, printed in 2.