Optical and Quantum Electronics (2005) 37: Ó Springer 2005 DOI /s

Transcription

1 Optical and Quantum Electronics (2005) 37: Ó Springer 2005 DOI /s Cryogenic radiometer based absolute spectral power responsivity calibration of integrating sphere radiometer to be used in power measurements at optical fiber communication wavelengths OG UZ C ELIKEL*, ÖZCAN BAZKIR, MEHMET KU C U KOG LU AND FARHAD SAMEDOV National Metrology Institute of Turkey (*author for correspondence: oguz.celikel@ume.tubitak.gov.tr) Received 15 December 2004; accepted 14 March 2005 Abstract. This paper covers the absolute spectral power responsivity calibration of spectralon-coated Integrating Sphere Radiometer (ISR) equipped with 3 mm diameter InGaAs photodiode to be used as a transfer standard in fiber optic power measurements against Electrical Substitution Cryogenic Radiometer (ESCR) in Optics Laboratory of National Metrology Institute (TUBITAK UME) of Turkey. The initial uncertainty arising from the use of the Electrically Calibrated Pyroelectric Radiometer (ECPR) as a transfer standard in radiometric scale is 0.5% (k=2), which particularly comes from irregularity in the surface homogeneity of ECPR. In order to eliminate the ECPR step as well as its initial uncertainty contribution in fiber optic power measurements, the calibration application herein was carried out. Moreover power stabilization measurements of DFB laser sources at both nm and nm, the beam size determinations, and spectral analyses of these laser sources as well as spatial and angular dependence of spectral responsivities of the ISR were presented in this paper. The total expanded uncertainties were calculated as 0.283% and 0.315% in the determination of absolute spectral power responsivities of the ISR for nm and nm wavelengths respectively (k=2). Key words: cryogenic radiometer, fiber optic power measurement, integrating sphere radiometer, spectralon 1. Introduction The calibration of fiber optic power meters used in optical communication is another important side of the optical fiber metrology, considering loss, distance and reflectance calibrations of OTDR instruments. The aim of the application described herein is to establish a convenient and traceable transfer standard to our primary radiometric standard, so called Electrical Substitution Cryogenic Radiometer (ESCR), at communication wavelengths with the uncertainty less than that of our transfer standard, Electrically Calibrated Pyroelectric Radiometer (ECPR). By means of the application described in the paper, ECPR step in radiometric transfer process was eliminated associated with its initial uncertainty contribution extending up to

2 530 O. C ELIKEL ET AL. 0.5%, in particular, owing to irregularities of surface homogeneity of ECPR (Bazkir and Samedov 2004). In UME Optics Laboratory, radiometric scale was realized by means of ESCR, using Ar +, He Ne, Nd:YAG laser sources radiating at nm, nm, nm, nm, nm and nm with an uncertainty of a few part of 10 )4 (Bazkir et al.). The realized radiometric scale was extended from 250 nm to 2500 nm by using ECPR (Electrically Calibrated Pyroelectric Radiometer), having flat spectral response by scanning the reflectance of the active surface LiTaO 3 of the ECPR over the mentioned spectral range with double monochromator (Bazkir and Samedov 2004). The uncertainty of the power measurement to be obtained using the ECPR in the realization of the radiometric scale can attain up to 1% or higher. In the spectral power responsivity calibration of ISR, it is obvious to see that the uncertainty will increase, if ECPR was engaged into calibration process as a transfer standard. To eliminate the step of ECPR in the traceability chain of our radiometric responsivity scale in fiber optic power measurements, and therefore to reduce the uncertainty level in fiber optic power measurements below the uncertainty level obtained from the configurations having ECPR, an ISR with 3 mm InGaAs detector was used as transfer standard and its calibration against the primary radiometric standard was performed as described herein. In comparison of the characteristic properties of ISR with those of single InGaAs photodiode; When the spectral responsivity of InGaAs photodiodes varies from 0.9 to 1.0 A/W for spectral range extending from the 1300 nm to 1600 nm, the spectral responsivity of ISR with InGaAs is relatively much lower than that of single InGaAs. This characteristic of ISR enables optical fiber power measurements to expand the higher power levels (up to a few Watts) without saturating InGaAs photodiode. However the relatively lower spectral power responsivity of ISR limits its operation in the lower power levels. The linearity of the ISR depends on the linearity of the assembled detector. Therefore it is shown that the ISR have perfect linearity for the incident power levels up to 300 lw for the wavelengths 1000 nm and greater (Boivin 2000). The spatial uniformity of ISR is better than 0.1%. This uniformity is much better than that of individual InGaAs detector (Boivin 2000; Envall et al. 2004). For an angular span of ±5, while the angular dependence of spectral power responsivity of ISR changes within 0.03%, the angular dependence of spectral responsivity of individual InGaAs was observed as slightly higher than 0.06% for unpolarized light. The measurements were performed from 900 nm to 1600 nm, over full spectral range (Boivin 2000).

3 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 531 The beam pattern emerging from the end of an optical fiber in a solid angle suits Gaussian beam profile, except for that of dispersion-modified fiber. Fiber optic power meters should measure these polarized, divergent and approximately Gaussian beams. This divergence of optical flux exiting from end of an optical fiber results in problems due to spatial and angular non-uniformities of the fiber optic power meters (Envall et al. 2004). The responsivity of the transfer standard to be used to calibrate the fiber optic power meters should therefore be independent of the angular variation of the divergent incident beam (Corredera et al. 1998). To overcome such angular problem arising from the divergence angle of the beam exiting from both single mode and multimode optical fiber, an integrating sphere, angular insensitiveness of spectral responsivity of which is inherent in its geometrical and physical structure, is a good solution. 2. Primary radiometric standard 2.1. PRIMARY RADIOMETRIC STANDARD, ELECTRICAL SUBSTITUTION CRYOGENIC RADIOMETER To give short description of the ESCR facility will be beneficial so that the alignments of the Near IR (NIR) optical beams at nm and nm into the absorbing cavity of ESCR facility, measurements of the absolute optical powers and their calculations can be explained. The ESCR used in the absolute spectral responsivity calibration is manufactured by Oxford Instruments Ltd. The principle element of the ESCR is the absorbing cavity which has been designed and manufactured in such a way that the optical radiation trapped its inside is completely absorbed. The absorbing cavity, on which the electrical heater resistor wire is wrapped, is linked to a section, which is called as reference block in liquid helium bath (4.2 K), via weak thermal link. A window inclined at Brewester angle is settled in front of the ESCR. The absorbing cavity is approximately 50 cm behind the window. There is a Silicon (Si) quadrant photodiode on straight optical path between the window and the absorbing cavity and the quadrant photodiode is located near the absorbing cavity. The quadrant photodiode have 45 mm active area and 9 mm diameter central hole. The optical flux is dropped into the absorbing cavity by means of the 9 mm diameter hole of the Si quadrant photodiode without any loss photodiode without any loss (Bazkir et al.) MEASUREMENT AND CALCULATION OF LASER POWER APPLIED TO ESCR Absolute optical power measurement in ESCR was performed using static substitution method. This method is based on sandwiching optical temperature

4 532 O. C ELIKEL ET AL. between two electrical temperature points. In the static substitution method a measurement cycle consists of three optical points (which should be at identical temperatures) and two electrical points calculated to be slightly above and slightly below the optical power value (Bazkir et al.) Since the optical radiation can be reduced by scattering S(k), the window transmittance s (k), the imperfect cavity absorbance a (k) and nonequivalence N, the measured optical power P opt, was corrected for these parameters by using the following equation (Stock and Hofer 1993; Gentile et al. 1996) P corr:opt ðþ¼ k 1 s(k) NP opt a(k) þ SðÞ k The optical power used in any absolute responsivity calibration is the corrected optical power P corr.opt, and the corrected optical power is then calculated by ESCR software. The transmittance values for both wavelengths and other parameters were entered into software for making this correction. ð1þ 3. Transfer standard For the visible spectral range while the Si trap detectors (together with ECPR to expand the radiometric responsivity scale) is used as transfer standards widely, the InGaAs/Ge single detectors or InGaAs trap configurations are engaged as transfer standards in near infrared range. Our transfer standard in fiber optic power measurements is an integrating sphere having 5.08 cm inner diameter (Fig. 1). The material of the integrating sphere radiometer (ISR) is Spectralon and manufactured by Labsphere Inc. Spectration block Auxillary port The diameter of the hole is 5 mm Calibration cap for auxillary port Auxillary port cap without hole Triax, low noise connector 1.8 mm 3 mm active area InGaAs Detector Assembly The diameter of the hole is 5 mm FC/PC connector adapter Ferrule doesn't disturb the curvature of the inner surface of the sphere Calibration Cap To calibrate the sphere radiometer against to ESCR with IR Laser beams in free space Fig. 1. Geometrical structure of integrating sphere radiometer to be used as transfer standard in fiber optic power measurements.

5 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 533 The Spectralon operates as a Labertian diffuser in the infrared range. Its reflectance for the mentioned spectral range is equal to or higher than 99% for the spectral range extending from 400 nm to 1550 nm (Labsphere INC). The ISR has three ports, apart from 90 from one another. While two ports are used for optical inputs, the third port is used as detector port. ISR has four port caps the interior surfaces of which are coated with BaSO 4. Two of them, manufactured UME and so called calibration caps, have 5 mm diameter holes. One of the port caps, so called original auxiliary port cap, has no hole and its own original port cap of the ISR is used to close the auxiliary port in calibration processes. The fourth cap, manufactured in UME, and to be mounted to the port in fiber optic power meter calibrations, is modified with standard FC/PC connector adapter. The purpose of the calibration caps is to enable calibrating the sphere radiometer against to ESCR with collimated NIR laser beam at communication wavelengths. During the absolute spectral power responsivity calibration of ISR when the port with the first calibration cap having 5 mm diameter hole was used to allow the collimated laser beams to get into the sphere radiometer, auxiliary port is closed by its original port cap. The second calibration cap with 5 mm diameter hole (to be mounted on the auxiliary port) was used to observe the effects of the calibration cap with 5 mm diameter hole on the detector current and to compensate it after the completion of the spectral power responsivity calibration. The possible losses on the current owing to the use of the calibration cap with 5 mm hole was measured by repetitively removing and inserting the calibration caps with/without 5 mm diameter hole and this loss on the detectorcurrent was added to the detector current values as a correction factor. After applying this correction to the detector current, the original auxiliary port cap without hole, and port cap with standard FC/PC adapter are closed for the service of the calibration of the fiber optic power meters. The hole diameter of the standard FC/PC connector adapter to be used in calibration of working standards is approximately 1.8 mm, which is fully suitable for the diameter of the standard FC/PC connector s ferrule. InGaAs with 3 mm active area was used to detect the NIR optical power sent into ISR. Regarding signal to noise ratio (SNR) values, temperature dependences of spectral responsivities and shunt resistance of InGaAs and Ge photodetectors manufactured by Hamamatsu; when InGaAs has a noise equivalent power (NEP) of 6 10 )14 at 25 C, Ge has NEP values of changing from 10 )12 to 10 )13 at 25 C. The shunt resistance of 3 mm InGaAs photodiode at 25 C is an order of 10 MX, but the shunt resistance of Ge doesn t exceed a few hundreds of kx. In addition, the temperature dependence of spectral responsivity of InGaAs photodiode is smaller than that of Ge. There is no baffle interior of the sphere to prevent the diode from seeing the portion of the sphere on which the radiation first impinges. The use of this type of baffles was observed as unnecessary (Boivin 2000).

6 534 O. C ELIKEL ET AL. 4. Experimental In this section, before calibrating the sphere radiometer, all necessary experiments to be considered in the determination of responsivities of the ISR and the calculation of the uncertainties, except for linearity measurement, were presented in this section SPECTRAL ANALYSES OF DFB LASER SOURCES Because the aim of the implementation described herein is to determine the absolute spectral power responsivity of the ISR, and the spectral power responsivity is a function of the wavelength, the precise determination of the source wavelengths used in the calibration is very important nm and 1550 nm DFB laser source modules have 20 mw and 40 mw continuous output power levels respectively. The spectral distributions of the laser sources were obtained via a calibrated OSA. The spectral power distributions of laser diodes were given in Fig. 2. Fig. 2. Spectral power distributions of and nm laser sources.

7 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 535 The correct wavelengths of 1310 nm and 1550 nm HP 81663A DFB laser source modules were obtained at nm and nm from the calibrated OSA with 1 nm resolution. In this case the determined spectral responsivities of the ISR to be determined is valid for the stated wavelengths OPTICAL EMISSION STABILITY MEASUREMENTS In order to compensate fluctuations in the optical power and generate a geometrically well-defined laser beam, Laser Intensity Stabilizer (LIS) systems as shown in Fig. 7 were used for both NIR beams and He Ne beam to be engaged in alignment of NIR beams into the absorbing cavity of the ESCR. In principle, both visible and NIR laser intensity stabilizers have the same function except for the monitor photodiodes. A LIS consists of an Electro-Optic Modulator (EOM), temperature controlled monitor photodiode (Si in visible range and InGaAs in NIR range), and control electronics. Laser beam incident on the LIS s EOM first passes through a liquid crystal, which variably dims the beam, then passes through polarizing beam splitter. The optical beam after passing through the EOM is partially reflected from the wedged beam splitter. Precision monitor photodiode receives the reflected beam from the wedges window and sends it back to the EOM. This feedback loop controls the optical beam to deliver a constant signal power output. Fig. 3. Stabilized intensities graphics of and nm DFB laser sources versus time.

8 536 O. C ELIKEL ET AL. Before starting the calibration of the ISR, the stability measurements were performed. In order to reduce the multiple Fabry Perot reflections between two flat surfaces, which corrupts the stability of the laser (FC/PC style ferrule tip and the front panel optics of DFB laser modules), laser module outputs were connected to patchcords with FC/APC style connectors. Photocurrent generated by 5 mm diameter InGaAs, which is settled inside the cryostat, is converted into the voltage by means of a 110 db transimpedance amplifier and the voltage data related to optical emission stability was sampled via the 81/2 digit DVM. Before starting measurements both DFB laser sources and NIR LIS system were operated about 2 h. Results were introduced in Fig. 3. Uncertainties arising from the instabilities of laser sources were calculated as 2.2 parts in 10 )4 and2.9partsin10 )4 for nm and nm, respectively SPATIAL UNIFORMITY MEASUREMENT FOR TWO ORTHOGONAL AXES OF ISR To find out spectral power responsivity change depending on the spatial uniformity of the entrance of ISR is one of the key parameters necessary to determine the diameters of NIR beams to be used in absolute spectral power responsivity calibration of ISR against ESCR with collimated beam. Spatial uniformity measurement was carried out with the calibration cap with 5 mm diameter hole in two orthogonal axes via step motor facility controlled by a PC. The beam size used in scanning is approximately 1 mm and the measurement was carried out at nm. The change in uniformity was presented Fig Voltage change (mv) 4 x (mm) y (mm) 4 6 Fig. 4. Spatial uniformity map of the ISR having the calibration port with a 5 mm diameter hole.

9 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 537 According to the spatial uniformity result, the change in spatial uniformity of ISR with 5 mm diameter calibration cap is restricted within 10 parts in 10 4 for a region of 2 mm diameter and 12 parts in 10 4 for a region of 2.5 mm diameter around the center of the input port BEAM SIZE DETERMINATION OF NIR LASER BEAMS In the calibration of ISR against the ESCR a well-collimated beam in free space is required. Sending the NIR beam into the absorbing cavity of ESCR is somewhat difficult and time consuming due to invisibility of NIR radiation by human eye. To overcome this problem, we coincided the NIR beams with the visible He Ne beam on a beam splitter depicted in Fig. 7. Considering that the spectral response of Si quadrant photodiode is up to 1100 nm and its central hole diameter is 9 mm, the beam size and divergence of both the nm and nm laser beams should precisely be determined via InGaAs CCD camera and the resultant beam profiles were given in Fig. 5. These beam profiles were obtained by settling IR CCD camera on the same axis as that of the absorbing cavity of ESCR after passing through the Glan polarizer depicted in Fig. 7. When He Ne spot size was adjusted to 6 mm, considering the spatial uniformity of ISR the spot sizes of both nm and nm laser beams were adjusted to 2 mm. However with CCD camera inspection the beam size of nm was found as little greater than 2 mm. This results in an increase of 2 parts in 10 4 in uncertainty contribution coming from spatial uniformity at nm. One of the critical points in this study is to settle the NIR beams on the center of the He Ne laser beam profile. This was achieved by using a beam splitter and an FC/PC patchcord with two sides collimator. First the NIR beams and He Ne beam were superimposed on the beam splitter and the superimposed beams were then launched into the two sides collimator patchcord via the collimator. The mixed beams were obtained the output of the FC/PC patchcord. This conjunction on the center of He Ne beam was checked by means of IR viewer card together with IR CCD camera. Fig. 5. Beam profiles of and nm DFB laser sources after collimating optics.

10 538 O. C ELIKEL ET AL ANGULAR DEPENDENCE Angular dependence of the spectral power responsivity is very important parameter for ISR because the ISR will be calibrated with a collimated beam and then divergent beams will be applied to the calibrated ISR in calibration process of working standards or fiber optic power meters. The angular dependence measurement of ISR was carried out using nm laser source with NIR laser intensity stabilization unit and step motor controlled precision rotational stage. The ISR was located on the center of the rotational stage, perpendicular to the direction of the incident beam. The perpendicularity of the entrance port of the ISR was provided by a mirror, and IR viewer card. The beam size is adjusted to 1 mm. The variation of the photocurrent depending on the angle of incident beam was given in Fig. 6 The uncertainty arising from angular dependence of the spectral power responsivity of the ISR was calculated as 2.6 parts in 10 4 for an angle range of ± 5. This angle range is the worst case for our calibrations and applications TRANSMITTANCE MEASUREMENTS In the measurements of optical power performed by means of ESCR, one of the major uncertainty contributions comes from the window due to contaminations on the window, and deviations from the Brewester angle (Bazkir et al.). Before initializing the transmittance measurements, the window was cleaned with drop and drag method by ethanol and lens paper. The cleaned window was located on a gimbal holder and its angle was adjusted to Brewester angle. The transmittance measurements were carried out by taking the ratio of the signals obtained from the cryostat with 5 mm active area nitrogen-cooled InGaAs detector with and without the window in the path of the beam at both wavelengths with the beam sizes to be used in responsivity calibrations. The NIR beams were vertically polarized by means of a Glan polarizer. The Fig. 6. Variation in photocurrent of ISR depending on the angle of the incident beam.

11 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 539 Fig. 7. The calibration setup of ISR to be used in fiber optic power measurements at both and nm against the ESCR. transmittance values were obtained as and 96.49% at nm and nm respectively. These data were used as correction factors in the calculation of true optical power by ESCR (Equation 1). The uncertainty coming from window transmittance was included in the uncertainty of ESCR facility DESCRIPTION OF ABSOLUTE SPECTRAL POWER RESPONSIVITY CALIBRATION SETUP OF ISR Absolute spectral power responsivity calibration setup for the calibration of ISR against ESCR is depicted in Fig. 7. The calibration of the transfer standard against the primary radiometric standard means that with well-collimated beam in free space is necessary. In order to achieve this and to keep the polarization states of the light propagating the optical fiber to be steady we used a polarization maintaining (PM) (hi-bi) fiber patchcord on the output of polarization controller, one side of which was connectorized with a standard FC/PC connector and another side of which was mounted to a 5 mm collimator. To improve the optical power stability we added an FC/APC patchcord between laser source and polarization controller. In order to prevent the instant variation in the polarization state of the light propagating in the FC/APC patchcord, this patchcord was fixed onto the surface of the optical bench tightly. The beam emerging the collimator is single mode and polarized. This beam was stabilized via NIR LIS system. The purpose of use of the Polarization Controller module is to polarize the beam propagating in the PM fiber and to provide the polarized beam for the polarizing beam splitter of the NIR LIS system.

12 540 O. C ELIKEL ET AL. The intensity stabilized and vertically polarized NIR and He Ne beams were superimposed on the beams splitter. These superimposed beams were then launched into a patchcord with 5 mm diameter collimator, another side of which has a 2 mm diameter collimator. In order to prevent the divergences of the both NIR and He Ne beams, that occurs at relatively long distance (more than 1.5 m for our optical bench) between the beam splitter and the absorbing cavity of the ESCR, this patchcord was used. Diameters of the both NIR and He Ne beams were adjusted to 2 mm and 6 mm, respectively and collimation and therefore the divergence of these beams were checked at various distances via CCD camera. NIR beam was aligned into the absorbing cavity through the alignment of the He Ne beam. Prealignment of He Ne beam into the absorbing cavity was done using the method described in literature (Ko hler et al. 1996). The scattered portion of the beam was detected by the Si quadrant photodiode with 9 mm central hole. The precise adjustments of ESCR were done until the 6 mm diameter He Ne beam passed through the center of 9 mm diameter aperture of the Si quadrant photodiode. Since NIR and He Ne beams were superimposed and the size of the NIR beams approximately 1/3 of the He Ne beam, then this NIR beam also drops into the absorbing cavity in this configuration. Most optical materials and optical fibers exhibit the refractive index asymmetry in their structures. The asymmetry in refractive index is called as birefringence. Therefore birefringence in core structure of optical fibers causes that the polarized incident light is resolved into two orthogonal polarization states (Dericson 1998). In this case the NIR beam launched into the FC/PC patchcord via 5 mm diameter collimator in the setup was vertically re-polarized by using an additional Glan polarizer in front of ESCR window so as to perform true optical power measurement. 5. Uncertainty budged Optical temperature, electrical temperature and electrical power and nonequivalence of the absorbing cavity of the ESCR, window transmittance, scattered optical power, cavity absorbance, and repeatability of the optical power measurements constitute the total uncertainty of the cryogenic radiometer. Even though the relative standard uncertainty of ESCR is relatively low in visible spectral range (approximately 1.3 parts in 10 4 for k=2) (Bazkir et al.), this uncertainty will increase in IR range due to the difficulty of alignment of invisible light and the stability etc. Especially it is difficult to detect the NIR photons scattered and reflected from the window of ESCR located at Brewester angle by using the IR viewer card, InGaAs photodiode and CCD camera because vertically polarized light passes through the window at Brewester angle with a very small quantity of scattering and reflection (without loss in ideal

13 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 541 situation). However after completion of transmittance measurements, the angle of the window removed from the settling position on the Gimbal holder so as to mount to its original place in front of ESCR is so adjusted that the NIR beam reflected and scattered from the window at Brewester angle is minimized using InGaAs photodiode. This effect increases the uncertainty values coming from the cryogenic radiometer and measurement repeatability in IR calibration. Other uncertainty contributions are angular dependence, spatial uniformity and linearity of the ISR and laser source stabilities. Regarding the uncertainty arising from non-linearity of the ISR with InGaAs, for a dynamic range varying from 10 )9 to 10 )4 A, it is observed that the plain InGaAs photodiodes are linear from 5 10 )8 Ato1 10 )4 Ain photocurrent with an uncertainty of 0.08% (k=2) (Yoon et al. 2003). At wavelengths greater than 1000 nm, the ISR is perfectly linear up to an incident power level of approximately 300 lw. For this spectral region the non-linearity uncertainty of ISR with InGaAs was calculated as 1 part in At wavelengths below 1000 nm the error arising from non-linearity increases with decreasing wavelength. For the spectral range below 1000 nm, the uncertainty of non-linearity of the ISR can be reduced to 5 parts in 10 4 or less by approximately matching the power levels at which the primary calibration (using the cryogenic radiometer) and transfer calibration to other detectors are carried out (Boivin 2000]. Considering the spectral power responsivity calibration described herein, the optical power changed from 80 lw to 310 lw at the read out of ESCR and the photocurrent generated by InGaAs of the ISR varied from 6 10 )8 to )8 A for both nm and nm. By taking these into account, the uncertainty due to non-linearity of InGaAs of the ISR was estimated as 3 parts in 10 4 for both wavelengths. In addition there is a small difference between the uncertainties coming from spatial uniformity of ISR because the beam size of nm is a little greater than that of nm. Table 1. Uncertainty budged of the absolute spectral power responsivity calibration the ISR against the ESCR Uncertainty sources Relative standard uncertainty (10 )4 ) nm nm Cryogenic radiometer Angular dependence Spatial uniformity Linearity Stability of laser sources Repeatability Combined uncertainty Total expanded uncertainty (k=2)

14 542 O. C ELIKEL ET AL. The uncertainty budged relevant to absolute spectral power responsivity calibration of the ISR against primary radiometric standard, ESCR, was presented in Table Conclusion The experimental setup depicted in Fig. 7 was constructed in Optics Laboratory of National Metrology Institute (TUBITAK UME) of TURKEY to calibrate the NIR integrating sphere radiometer against our primary radiometric standard, Electrical Substitution Cryogenic Radiometer, ESCR. ISR, the diameter of which is 5.08 cm, is mainly composed of Spectralon inner coating, 3 mm active area InGaAs detector and two input ports one of which is reserved as calibration port with 5 mm hole and another is reserved as an auxiliary port to evaluate the effects of the calibration port with 5 mm hole and an angular variation of 90 of input port on the absolute spectral power responsivity calibration. For alignment of NIR lasers into the cavity of ESCR, He Ne laser (632.8 nm) was coincided with nm and nm laser beams on a beam splitter. The mixed beams ( nm nm and nm nm) were launched into a fiber patchcord and the output of the fiber patchcord was collimated via a 2 mm diameter collimator. Collimated NIR beams were sent into the absorbing cavity of ESCR by means of He Ne beam and then the He Ne was turned off during the absolute responsivity calibration of the ISR with NIR beams against ESCR. During the absolute spectral power responsivity calibration, the calibration cap with 5 mm diameter hole was removed and inserted again and again to assess the removing/inserting repeatability. The variation in photocurrent of the ISR did not exceed 0.5 parts in 10 )4. The responsivities of ISR at nm and nm were calculated as )4 (A/W) and )4 (A/W) with the expanded uncertainties of and 0.315%, respectively. Theoretical power range for ISR extends from a few hundreds of nw to a few watt. Considering these responsivity values NIR optical power measurements at the mentioned wavelengths can be carried out up to several watts without saturating the InGaAs. References Bazkir O., and F. Samedov. Electrical substitution cryogenic radiometer based spectral responsivity scale between nm wavelengths. Opt. Appl. Vol. XXXIV, No. 3, 427, Bazkir O., S. Ugur, F. Samedov and A. Esendemir. High-accuracy optical power measurements by using electrical substitution cryogenic radiometer, Opt. Eng. 44(1) 1, Boivin, L.P. Properties of sphere radiometers suitable for high-accuracy cryogenic-radiometer-based calibrations in the near infrared, Metrologia , 2000.

15 CRYOGENIC RADIOMETER BASED ABSOLUTE SPECTRAL POWER 543 Corredera, P., J. Campos, M.L. Hernanz, J.L. Fontecha, A. Pons and A. Corrons. Calibration of nearinfrared transfer standards at optical-fibre communication wavelengths by direct comparison with a cryogenic radiometer, Metrologia , Dericson, D. Fiber Optic Test and Measurement, Prentice Hall PTR, New Jersey, Envall, J., P. Karha and E. Ikonen. Measurements of fibre optic power using phodiodes with and without an integrating sphere, Metrologia , Gentile, T.R., J.M. Houston, J.E. Hardis, C.L. Cromer and A.C. Parr. National institute of standards and technology high-accuracy cryogenic radiometer, Appl. Opt , Ko hler, R., R. Goebel and R. Pello. Experimental procedures for the comparison of cryogenic radiometers at the highest accuracy, Metrologia , Labsphere INC. Spectralon Diffuse Reflectance Material Catologue. Stock, K.D. and H. Hofer. Present state of the PTB primary standard for the radiant power based on cryogenic radiometry, Metrologia , Yoon, H.W., J.J. Butler, T.C. Larason and G.P. Eppeldauer. Linearity of InGaAs photodiodes, Metrologia 40 S154, 2003.