2-2 Power Meter Calibration Power Meter Calibration 1 (1 mw, 50 ohm)

Transcription

1 esearch and evelopment of Calibration Technology - ower eter Calibration --1 ower eter Calibration 1 (1 mw, 50 ohm) Tsutomu UGIYAA, ojiro AAI, ouichi EBATA, Iwao NIHIYAA, and atsumi FUJII NICT performs calibration services for power meters with coaxial sensors in the frequency range from 100 khz to 50 GHz, and for those with waveguide sensors from 50 GHz up to 170 GHz. In order to ensure the accuracy, the simultaneous comparison and substitution method is adopted as the principle of calibration. The calibration system has the expanded of 0.66 % (coverage factor k=) at 100 Hz. In this report, we describe the calibration methods, systems, and uncertainties in the frequency range 100 khz to 110 GHz. 1 Introduction The output levels and modulation methods are established for TV and radio broadcast waves, and various radio waves produced from wireless devices such as mobile phones and radar. For measurement of their output levels, high-frequency power meters (hereinafter, ower eters ) are mainly used. ower eters are also used when calibrating measurement instruments, such as signal generators and spectrum analyzers. ower eters are important measuring instruments used in evaluating characteristics of wireless devices, and in calibration of various measuring instruments, so their calibration is required in order to measure accurately. As of November 016, NICT is calibrating ower eters in frequencies from 100 khz to 170 GHz. For sensors (Type-N 50 Ω, 3.5 mm,.4 mm) with various coaxial connectors as input terminals (hereinafter abbreviated as Coaxial ensors ), NICT is calibrating from 100 khz to 50 GHz. For sensors with waveguides as input terminals (V-Band (50 75 GHz), W-Band ( GHz), -Band ( GHz)) (hereinafter abbreviated as Waveguide ensors), NICT is calibrating up to 170 GHz. This report describes calibration methods for ower eters from 100 khz to 110 GHz frequencies (1 mw or less, Coaxial ensors and Waveguide ensors), and calibration systems and calculation methods. efinitions General ower eters have a structure divided into the indicator part and sensor. Both are connected by a cable (Fig. 1). Equation (1) shows the relationship between the instruction value and incident power in into the sensor. (1) in Here, is the calibration factor, and calibration is seeking the value of this. General ower eters contain a reference signal source (50 Hz, 1 mw), and when using a ower eter, a sensor is first attached to the reference signal source, and the 1 mw value is aligned. Therefore, the calibration factor also includes the precision of this reference signal source, and at NICT, the indicator part and sensor are calibrated as one unit. 3 Calibration methods 1 Basic components of high-frequency ower eters 3.1 imultaneous comparison and substitution method There are various ower eter calibration methods: comparison method, simultaneous comparison method, simultaneous comparison and substitution method, etc. The simultaneous comparison and substitution method has 13

2 esearch and evelopment of Calibration Technology ower plitter EF, ignal ource #1 # #3 T UT substitute imultaneous comparison and substitution method UT ort # T ort #3 advantages: it is not affected by reflection of signal source, is strong against output variation of the signal source, etc. NICT uses the simultaneous comparison and substitution method for calibration of ower eters (1 mw or less, Coaxial ensors and Waveguide ensors). Figure shows a conceptual diagram of the simultaneous comparison and substitution method. The signal from the signal source is connected to ort #1 of the power splitter, and power that goes through the power splitter is distributed into Test ort # and eference ort #3. After that, calibration is done by the steps below. 1) Connect the reference device (EF) to eference ort #3 (do not remove until calibration ends) ) Connect the standard device (T) to Test ort # 3) Adjust the signal source connected to ort #1 so the T measurement value of Test ort # is 1 mw. 4) eek the ratio ) of T measurement ( value vs. EF measurement value at that time 5) Change T connected to Test ort # to evice Under Test (UT) 6) ake the signal source the same output as 3) ( 7) eek the ratio ) of UT measurement value vs. EF measurement value at that time 8) From 4) and 7), use Equation () to determine the calibration factor of UT. [1] () where, is the calibration factor of T used by an upperlevel calibration organization. (3) W ignal Generator (4) Here, mismatch is expressed in the following equation from the equivalent signal source reflection coefficient Γg, the reflection coefficient Γ of T, and the reflection coefficient Γ of UT. 1 g (5) 1 g Γg can be obtained from parameters of the power splitter. 1 g 3 (6) plitter 3 Calibration system 3. Calibration of power meter using coaxial sensor For a Coaxial ensor, it is possible to calibrate ower eters with 1 mw power, 100 khz to 50 GHz frequency [1]. Figure 3 shows a calibration system photo, with a block diagram in Fig. 4. This calibration system uses the simultaneous comparison and substitution method. Output from the signal generator is input via a switch into the power splitter (ort #1), and the power splitter splits it into eference ort #3 and Test ort #. The calibration steps are the steps explained in the steps of the simultaneous comparison and 14 Journal of the National Institute of Information and Communications Technology Vol. 63 No. 1 (016)

3 --1 ower weter CalierCelon 1 (1 mw, 50 ohm) ignal Generator W Control witch #1 ower ensor (EF) #3 ower plitter ower eter (EF) ort # # ort #3 ower plitter ower ensor (T) ower ensor (UT) substitute 4 Calibration system block diagram ower eter (T) ower eter (UT) ignal Generator ource odule irectional Coupler Isolator substitution method. One advantage of this calibration system is it is designed with the calibration system s eference ort #3 and Test ort # facing upward, so each port plane and sensor s connector plane contact equally. oreover, by switching the switches, one can select the power splitter corresponding to the frequency range and the output port (test port or reference port) corresponding to the sensor s connector, and one can thereby calibrate various connectors (Type-N 50 Ω, 3.5 mm,.4 mm) from 100 Hz to 50 GHz. When connecting each connector, to ensure reproducibility, we use a torque wrench, and always tighten to the same torque. 3.3 Calibration of power meter using waveguide sensor In the case of a Waveguide ensor, calibration of the ower eter is possible with 1 mw V-Band, and 0.1 mw W-Band[]. There are different calibration systems for V-Band and for W-Band. Figure 5 shows a photo for the W-Band system, and Fig. 6 shows its block diagram. This system also uses the simultaneous comparison and substitution method for its calibration method. However, the Coaxial ensor s calibration system used a power splitter, but this system uses a directional coupler (degree of coupling: 6 db) instead of a power splitter. Output from the signal generator is multiplied via a multiplier, multiplied 4 times in the case of V-Band, and multiplied 6 times in the case of W-Band, then input into the directional coupler (ort #1). In the directional coupler, the signal is split into each, and in the direction of the traveling waves, two isolators are also connected, and that output terminal is Test ort #. Also, eference ort #3 is a terminal in the direction of the reflected wave of the directional coupler. Like the Coaxial ensor calibration system, an advantage of this calibration system is that it is designed with the calibration system s eference ort #3 and Test ort # ignal Generator 5 Calibration system (W-band) ource odule 4 (V-Band) 6 (W-Band) EF #3 irectional Coupler (6dB) 6 Calibration system (W-band) block diagram T or UT # Isolator 0 db Isolator 0 db facing upward, so each port and Waveguide ensors waveguides contact equally. oreover, the calibration system s waveguides are very delicate, so it has fixtures to secure the Waveguide ensor to prevent shaking. 3.4 Calibration value Calibration is done by the steps described above. and are each measured 100 times, that average is obtained, and Equation () is used to calculate the calibration value. The calibration factor is calculated with as 1, and the effect due to considering as 1 is evaluated as. However, at high frequencies, we can no longer ignore s effect on calibration value, so Γg, Γ and Γ (all of them complex quantities) are used to calculate the value of defined in Equation (5), that result is applied in Equation (), and the corrected calibration factor is calculated. Figure 7 is a graph of values of when UT (.4 mm connector, 1 to 50 GHz) is calibrated in a Coaxial ensor calibration system. As shown in the figure, especially when more than #1 15

4 esearch and evelopment of Calibration Technology TTable 30 GHz, is far different from 1, and we cannot ignore the effect of on calibration value (in Fig. 7, at 50 GHz, it has about 1% effect on calibration factor). 4 Traceability 7 graph 1 JC certification range Calibration and measurement Frequency(GHz) capability (%) (Level of confidence approximately 95%) , 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.1, 0.15, 0., 0.5, , 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, , 1.5, 1.6, , 3.0, , All ower eter calibrations are traceable at the National etrology Institute of Japan (NIJ), which sets Japan s national standards (secondary standard device). Among those, ower eters (1 mw, Type-N 50 Ω sensor) with frequencies from 10 Hz to 18 GHz can be calibrated by JC (international A compatible) based on IO/IEC1705[3] [5]. JC is the easurement Act traceability system based on Japan s easurement Act, with the National Institute of Technology and Evaluation managing the registration system for calibration laboratories. These registration criteria are whether there is compliance with the easurement Act related regulations and items required by IO/IEC1705. Table 1 shows the frequencies and calibration and measurement capabilities of JC certification. For calibration and measurement capability, it shows the smallest value when UT is calibrated, for when reflection Γe of UT sensors is 0; this is written in the registration certificate during JC registration. 5 Uncertainty 5.1 Coaxial sensor Equation (7) shows the propagation equation for of simultaneous comparison and substitution method[6]. u( s ) is the of calibration used by an upper-level calibration organization, and u(x) shows the standard for x. u C u u u u s where, the T calibration factor, and due to the power splitter, 1, 1 (7) oreover, if the calibration factor is calculated with equal to 1 (corrections are not made using ), then the estimates of,, and in Equation (7) are all 1, so the sensitivity coefficient for each factor of is 1 or -1, which are 1 when squared, so the relative standard is obtained using the following equation. u( ) u( ) u( ) u( ) u( ) s( ) (8) In the case where corrections are not made using, the factors of are (1) Uncertainty of T, () ifference of ambient temperature during UT calibration vs. during T calibration conducted by an upper-level organization, (3) Change over time of T, (4) Number of digits displayed of UT, (5) Number of digits displayed of T, (6) isalignment between UT and power splitter and between T and power splitter, (7) easurement 16 Journal of the National Institute of Information and Communications Technology Vol. 63 No. 1 (016)

5 --1 ower weter CalierCelon 1 (1 mw, 50 ohm) TTable Factor of Uncertainty budget example (100 Hz frequency) (No correction) Uncertainty istribution ivisor tandard ensitivity coefficient Contribution to u(x) c(x) c(x) u(x) Upper-level calibration % Normal Temperature change 0.4% 0.4% Normal Change over time 0 0 Normal s UT resolution Uniform T resolution Uniform ismatch U eproducibility easure 5 times Normal Combined tandard Uncertainty Expanded relative (k= ) Frequency 0.1 GHz, = % variability. (1) uses the value (differs by frequency) shown in the calibration certification of T (normal distribution). () is obtained by actual measurements of the effect on calibration value due to the difference between the temperature shown in the calibration certification of T (3 ± 1ºC), and the ambient temperature during calibration (3 ± ºC) (normal distribution). (3) is obtained from fluctuations in calibration value measured by an upper-level calibration organization over a one-year period (normal distribution). (4) and (5) are determined from the digits read of UT and T during calibration (round to the nearest 4th decimal) (uniform distribution). (6) is calculated from actual measurements of the reflection coefficients of T and UT and the parameter of Test ort # (U distribution). In (7), calibration is repeated n times, and the variability is calculated. However, the surface joining T and Test ort # is taken as a fixed size, while the position of circumference direction of the surface joining UT and Test ort # divided into n pieces (Type-N 50 Ω connector s female core conductor is divided an even number (4 or 6) of times[7], so it is desirable that n is 3 or a higher odd number) is measured repeatedly over the entire circumference (360/n degrees x n times), and the variability is measured (normal distribution). Generally, sensors are designed so the reflection coefficients Γ and Γ are sufficiently small. And from Equation (6), we know that Γg is also sufficiently small, so g 1 and g 1 (9) are true. Therefore, Equation (5) becomes 1 g 1 g g 1 g (10) Therefore, generated when is approximately 1 is expressed by the following equation. u g g (11) Table shows an example (100 Hz frequency) of an budget of calibration, from the factors listed above, in the case where correction due to is not done Also, Equation (7) is used to obtain the in the case where correction due to is done[8]. In this case, of differs from Equation (11); it adds of the measurements Γg, Γ, and Γ: u u 4 u u 4 (1) g Also, from Equation (6), we get: g 1 3 u u u u 3 1 (13) u g Here, u(xx) is the standard when xx is measured. The value of is calculated using measurements of parameters, by a calculation formula of the company that manufactures the Vector Network Analyzer (VNA). As shown in Fig. 7, especially when exceeding 30 GHz, is far different from 1, so in the case of ower eters using Coaxial ensors, NICT currently does correction using at 30 GHz or greater. Table 3 and 4 show examples 17

6 esearch and evelopment of Calibration Technology TTable Factor of 3 Uncertainty budget example (50 GHz frequency) (With correction) Uncertainty istribution ivisor tandard ensitivity coefficient Contribution to u(x) c(x) c(x) u(x) Upper-level calibration % Normal Temperature change 0.4% 0.4% Normal Change over time 0 0 Normal s UT resolution Uniform T resolution Uniform ismatch U eproducibility easure 5 times Normal Combined tandard Uncertainty Expanded relative (k= ) Frequency 50 GHz, = % Factor of TTable 4 Uncertainty budget example (100 GHz frequency) tandard Uncertainty istribution ivisor ensitivity coefficient Contribution to u(x) c(x) c(x) u(x) Upper-level calibration % Normal Temperature change 0.4% 0.4% Normal Change over time 0 0 Normal s UT resolution Uniform T resolution Uniform ismatch U eproducibility easure 4 times Normal Combined tandard Uncertainty Expanded relative (k= ) Frequency 100 GHz, = % of the budget of calibration when is corrected. 5. Waveguide sensor Also in the case of a Waveguide ensor, like in the case of Coaxial ensors, Equation (7) is used to obtain. Also, for Equation (6), in cases where 1 3 (14) We focus on ability to approximate (15) g As shown in Fig. 6, in a calibration system for Waveguide ensors, signals output from the multiplier are input into the directional coupler. The directional coupler s degree of coupling is 6 db (uniform value), so the parameter () between ort #1 and eference ort #3 becomes approximately 6 db. On the other hand, two isolators are inserted between ort #1 and Test ort #, and due to the loss when passing through these, and the loss when passing through the directional coupler, the parameter (1) between ort #1 and Test ort # becomes approximately 6 db 18 Journal of the National Institute of Information and Communications Technology Vol. 63 No. 1 (016)

7 --1 ower weter CalierCelon 1 (1 mw, 50 ohm) Also, considering the parameter between Test ort # and eference ort #3, there is 40 db isolation due to two isolators, and the directionality of the directional coupler used ( 0log 1 ) is 0 db or greater. Therefore, 3 becomes approximately 60 db smaller than the reflection of Test ort # (). Therefore, in the calibration system shown in Fig. 6, the relationship of Equation (14) holds true, so we can use the Equation (15) approximation formula. Thus it is not necessary to use VNA to measure all parameters of the calibration system. o we can obtain Γg by only measuring (possible by -ort VNA) the reflection coefficient of Test ort # (). The factors of are similar to those for calibration of Coaxial ensors, but variability of measurements are evaluated by measuring n times (n is an even number) with the direction faced (waveguide) changed 180 degrees for Test ort # and UT. Table 4 shows an example of an budget of calibration (100 GHz frequency). 8 ame UT calibration results (10 Hz to 18 GHz) 6 Changes over years In order to check the calibration system and validity of calibration results, the same calibration of UT is done each year, and changes in calibration results over the years are evaluated. Figure 8 shows ower eter calibration results from 011 to 014 (Type-N 50 Ω sensors, 10 Hz to 18 GHz frequency). Figure 9 shows their calibration results (W-Band) from 01 to 016. However, frequencies that were not calibrated are not marked. It was decided to evaluate changes over the years by the number of En [9]. The number of En is used in evaluations such as round robin tests. En is expressed by the following equation. E n LAB EF (16) U U LAB EF where, LAB : easurement value from a participating calibration organization EF : easurement value from a reference calibration organization ULAB : Expanded from a participating calibration organization (k = ) UEF : Expanded from a reference calibration organization (k = ) and the evaluation is satisfactory if n, not satisfactory if n. 9 ame UT calibration results (75 to 110 GHz) Evaluations were performed using Equation (16) with the calibration value of the reference year (final year) as EF ( UEF), and each year s calibration value as LAB ( ULAB). The number of Eon is obtained for each frequency, and the highest value over the past four years is obtained. However, the calibration value of T by an upper-level calibration organization differs slightly each year, but only the calibration results are evaluated, so the ratio vs. the reference year (final year s T calibration value/each year s T calibration value) is multiplied by each calibration value, and the T change portion is removed. The evaluation results are that the maximum number of Eon is 0.4 for ower eters using Type-N 50 Ω sensors (10 Hz to 18 GHz frequency), 0.15 Eon for.4 mm sensors (1 to 50 GHz frequency), 0. Eon for V-Band sensors, and 0.11 Eon for W-Band sensors. All are evaluated as satisfactory. 19

8 esearch and evelopment of Calibration Technology As described above, calibration results over the past four years by this system using the simultaneous comparison and substitution method show that the maximum number of Eon within the evaluation range is less than 1, and stable calibration results are obtained. 7 Conclusion This described a method for calibration of ower eters until 110 GHz frequency, a calibration system, and a method for calculating, that enable very precise calibration using the simultaneous comparison and substitution method. This also showed that stable results were obtained for calibration of ower eters by this calibration system. epresentative expanded uncertainties (coverage factor k = ) were 0.66% for 100 Hz, 3.% for 50 GHz, and 3.7% for 100 GHz. The method of ower eter calibration using the simultaneous comparison and substitution method can also be applied to absolute values of calibration of spectrum analyzers, and is actually being used in calibrations. Appendix. erivation of Equation () In the calibration system with simultaneous comparison and substitution method shown in Fig., when a standard device is connected to Test ort #, and it is expressed using parameters, the following equation is obtained. b1 b b G Gb a1 a a 3 (A.1) a a (A.) a a (A.3) b (A.4) 3 b 3 Here, the matrix expresses characteristics of a three port circuit shown by the dashed lines containing the power splitter. ag is the source power of signal source, ΓG is the reflection coefficient of signal source, Γ is the reflection coefficient of reference device, and Γ is the reflection coefficient of the standard device. From these equations, the power input into the standard device, and the power input into the reference device, are each obtained by in ()(11) b3 ag (A.6) where: 1 11G 1 13 det 1G 1 3 G (A.7) 1 3 det (A.8) ( 1)(11) det 3 3 (A.9) ( )(11) 1 Here, det[a] expresses the matrix formula of matrix A. Now, if we simultaneously measure two incident powers and obtain their ratio, then by (A.5) and (A.6), we obtain: 1 T 33 3 in (1)(11) 1 1 in ()(11) (A.10) T Here, the relationship between in and in was used. and are each the calibration factors of the standard device and reference device. Next, if instead of the standard device, a device under test (UT) is connected to Test ort #, then measuring the input power obtains: 1 UT 33 3 in 1 1 (A.11) in Here, Γ is the reflection coefficient of the device under test (UT), where the relationships of and in were used. is the calibration factor of the device under test. Now, if we calculate the ratio of Equation (A.10) and Equation (A.11), we get (A.1) If we transform the equation, we obtain the following Equation (). T in (1)(11) b ag (A.5) 0 Journal of the National Institute of Information and Communications Technology Vol. 63 No. 1 (016)

9 --1 ower weter CalierCelon 1 (1 mw, 50 ohm) (A.13) ouichi EBATA enior esearcher, Electromagnetic Compatibillity Laboratory, Applied Electromagnetic esearch Institute Calibration of easuring Instruments and Antennas for adio Equipment, geodesy In the equation s derivation process, the power splitter s 1 and, the calibration factor and reflection coefficient Γ of the reference device, and matrix formula shown in Equation (A.7), are all deleted, so there is no need to actually obtain them. Iwao NIHIYAA Electromagnetic Compatibillity Laboratory, Applied Electromagnetic esearch Institute Calibration of easuring Instruments and Antennas for adio Equipment eference 1. FUJII, T. UGIYAA, A. UZUI, T. HINOZUA, and Y. YAANAA, evelopment of a ower eter Calibration ystem for illimeter Wave Frequencies, IEICE Technical eport, ECJ006-57, W , pp.37 41, Oct FUJII, T. UGIYAA, A. UZUI, T. HINOZUA, and Y. YAANAA, evelopment of a ower eter Calibration ystem for the use of V and W bands, IEICE Technical eport, ECJ007-, pp.7 1, April IO/IEC 1705:005, General requirements for the competence of testing and calibration laboratories, IWAA,. FUJII, H. AUZAWA,. OIE,. AAAI, A. UZUI, Y. IYAZAWA, Y. YAANAA, and T. HINOZUA, evelopments of IO/ IEC1705 Calibration ystems in Wireless Communications epartment, Journal of the NICT, vol.53, no.1, pp.43 57, National Institute of Technology and Evaluation, JC Technical Guide for Estimation Examples of Uncertainty: ower ensor, JCG111-06, June 010. (in Japanese) 7 eysight Technologies, icrowave and illimeter-wave coaxial connector, ec (in Japanese) 8. INOHITA,. HIAOA, and. OIYAA, Comparison method for the calibration of microwave power meters and its analysis, NIJ Technical eport, vol.6, no.3, pp , ept JIQ 17043: 011 (IO/IEC 17043: 010) Conformity Assessment-General requirements for proficiency testing. atsumi FUJII, r. Eng. esearch anager, Electromagnetic Compatibility Laboratory, Applied Electromagnetic esearch Institute Calibration of easuring Instruments and Antennas for adio Equipment, Electromagnetic Compatibility Tsutomu UGIYAA enior esearcher, Electromagnetic Compatibillity Laboratory, Applied Electromagnetic esearch Institute Calibration of easuring Instruments and Antennas for adio Equipment ojiro AAI Technical Expert, Electromagnetic Compatibility Laboratory, Applied Electromagnetic esearch Institute Calibration of easuring Instruments and Antennas for adio Equipment 1