CCEM KEY COMPARISON CCEM.RF-K1d.W (GT-RF/97-3) Measurement Techniques and Results of an Intercomparison. for RF Power in R 900

Transcription

1 CCEM KEY COMPARISON CCEM.RF-K1d.W (GT-RF/97-3) Measurement Technques and Results of an Intercomparson for RF Power n R 900 at frequences of 75 and 94 GHz. Fnal Report of the Plot Laboratory Joanna Beaumont and Geoff Orford Natonal Physcal Laboratory (NPL) Teddngton, Mddlesex, TW11 0LW July of 17

2 CCEM KEY COMPARISON CCEM.RF-K1d.W (GT-RF/97-3) Measurement Technques and Results of an Intercomparson for RF Power n R 900 at frequences of 75 and 94 GHz. PARTICIPANTS Joanna Beaumont / Geoff Orford Natonal Physcal Laboratory (NPL) Teddngton TW11 0LW. Unted Kngdom John R Juroshek Natonal Instute of Standards and Technology (NIST) Boulder, Colorado. USA Bernard Mellouet/ Joseph Achkar Laboratore Central des Industres Electrques (BNM-LCIE) Fontenay-aux-Roses Cedex. France eter Jank Physkalsch-Technsche Bundesanstalt (PTB) Braunschweg Germany ABSTRACT A measurement comparson of effectve effcency has been carred out between four natonal metrology laboratores n wavegude R 900 at 75 and 94 GHz. The orgnal ndentfcaton of the ntercomparson was GT-RF/97/3 but was subsequently renumbered as CCEM.RF-K1d.W. One wavegude thermstor mount has been measured. The followng four natonal laboratores of metrology partcpated NPL (Unted Kngdom), BNM-LCIE (France), NIST (Unted States of Amerca), PTB (Germany). The Natonal Physcal Laboratory (Unted Kngdom) acted as the plot laboratory for the comparson. Good agreement between the values provded has been observed. It can be noted from tables 2 to 5 of Appendx A that the degrees of equvalence for all the values reported, wth respect to the reference value and between pars of laboratores, are less than ther assocated uncertanty, at a level of confdence of 95%. 2 of 17

3 1. INTROUCTION urng the BIPM meetng n France on 23 June 1998 an nterest was shown n undertakng an nternatonal comparson n Wavegude Sze R900. The comparson, detaled n Table IV of the Workng Group Report [1], suggested a measurement of the Effectve Effcency of a Bolometer Mount at 94 GHz wth 75 GHz and 110 GHz as optonal frequences. The ntercomparson was gven the dentfcaton GT-RF/97/3 but was subsequently renumbered as CCEM.RF-K1d.W. All partcpants reported results for effectve effcency at 94 GHz and three partcpants reported results at 75 GHz. Three of the four partcpants also report measurements of the Voltage Reflecton Coeffcent (Magntude) of the Bolometer Mount. All the reported results are ncluded n ths report. 2. TRAVELLING STANAR AN EFFECTIVE EFFICIENCY The travellng standard provded by the plot laboratory was a Hughes Thermstor Mount Model 45776H-1100 ftted wth a short secton of wavegude havng a standard R900 flange. Bolometer mounts are used n natonal standards laboratores for the absolute determnaton of RF power. The thermstor element of the mount s connected as one of four resstors n a self-balancng brdge. When the resstance of the thermstor element changes wth temperature the brdge s suppled wth a C bas current to keep the resstance of the thermstor element constant. The RF performance of the bolometer mount s descrbed by ts effectve effcency, η and s gven by: η = P PC RF (1) where P C s the C substtuted power and P RF s the mcrowave power dsspated n the mount. Partcpatng laboratores were asked to provde a measure of effectve effcency of the travellng standard and f possble to measure the voltage reflecton coeffcent. 3. THE COMPARISON PROTOCOL AN SCHEULE The travellng standard was crculated to the partcpatng laboratores and they were asked to provde a measure of the effectve effcency of the travellng standard at frequences of 75 GHz and 94 GHz. Measurement of the voltage reflecton coeffcent of the travellng standard was optonal. In three of the four laboratores the measurements were made wth the secton of wavegude connected to the Thermstor Mount but n the case of BNM-LCIE the measurements were made wth the secton of wavegude removed and then the results 3 of 17

4 were corrected for the loss n ths secton, ths was necessary for the travellng standard to ft n the BNM-LCIE mcrocalormeter. The table below gves the date of measurement at each of the partcpatng laboratores and the measurements that were made on the travellng standard, the dates of measurement are consstent wth the proposed schedule. Measurements made Measurements made Laboratory ate of at 75 GHz at 94 GHz Measurement Effectve Reflecton Effectve Reflecton Effcency Coeffcent Effcency Coeffcent BNM-LCIE (France) September 1998 PTB (Germany) January 1999 NPL (UK) March 1999 NIST (USA) Aprl 1999 NPL (UK) ecember METHOS OF MEASUREMENT NPL (1) & NPL (2) The measurement system used was a WG27 (R 900) multstate reflectometer (MSR) n combnaton wth WG27 (R 900) standards, whch have been establshed by prevous measurements n a waterbath mcrocalormeter [2]. In operaton the MSR s frst of all calbrated wth a seres of spacers, short crcuts and matched loads to establsh the correcton factors for the reflectometer. The standard thermstor mount s placed at the measurement plane and s used to calbrate the MSR as a power transfer system. The devce under test s attached to the MSR port and the effectve effcency and voltage reflecton coeffcent are measured usng the prevously obtaned calbraton data. The travellng standard was measured four tmes wth dsconnecton and 180 rotaton of the wavegude flange between each measurement. Measurements were performed at (23 ± 1) C. The nomnal power dsspated n the travellng standard mount was between 3 mw and 6 mw. BNM-LCIE The measurements of the effectve effcency of the devce under test were performed, wthout the wavegude secton, on the BNM-LCIE waterbath symmetrcal mcrocalormeter 4 of 17

5 [3]. A dode source wth a source lockng mcrowave counter was used. The thermstor mount was measured 5 tmes wth 14 cycles of wth RF/wthout RF. The RF power dsspated n the thermstor mount was 5mW. The effectve effcency of the travellng standard, ncludng the wavegude secton, has been obtaned usng the followng correcton: η 2 ( 1 Γ ) ( 1 Γ ) 2 1 S where η 1 s the effectve effcency of the mount wthout the wavegude, Γ 1 and Γ are respectvely the magntude of the reflecton coeffcent of the mount wthout and wth the wavegude. S 21 s the attenuaton of the wavegude. The magntudes of the reflecton coeffcents were measured wth 3 connectng/dsconnectng owng to the tuned reflectometer method. The measurements were performed under ambent envronmental condtons of (23 ± 1) C and less than 60% relatve humdty. η (2) NIST The effcency of the devce under test was measured on the NIST WR-10 dual sx port measurement system [4,5]. The dual sx-port was calbrated for reflecton coeffcent measurements usng precson wavegude transmsson lnes and the thru-reflect-lne (TRL) calbraton technque. The WR-10 dual sx port was calbrated for effectve effcency measurements usng a WR-10 power standard whose effectve effcency had been determned from measurements n the NIST mcrocalormeter [6]. The devce under test was measured on each port of the dual sx port 4 tmes wth dsconnect and 180 rotaton of the devce between each connect. The dual sx port measurement system was calbrated for reflecton coeffcent and power measurement twce. PTB The devce under test was connected to one of the R 900 feedng lnes of the symmetrcal twn-type PTB R 900 mcrocalormeter, whch s smlar to the wavegude mcrocalormeter descrbed n [7]. The reference mount used for the symmetrcal PTB mcrocalormeter was a smlar R 900 bolometer mount. By measurng the heatng of the mount n the mcrocalormeter durng C substtuton measurements the effectve effcency of the devce under test was determned [7]. The effectve effcency was measured 6 tmes for each frequency always dsconnectng and reconnectng the mount agan after each measurement. As a bolometer brdge an NBS Type II bolometer brdge was used. 5 of 17

6 5. MEASUREMENT RESULTS AN REFERENCE VALUES The measurement results were presented to the plot laboratory n the format of the mean of the effectve effcency and the mean of the magntude of the reflecton coeffcent. Partcpants were asked to provde estmates of Type A and Type B uncertantes and the combned uncertanty for both measured quanttes, at one standard devaton. The measurement results and assocated uncertantes together wth the reference values and assocated uncertantes are shown n Fgures 1a and 1b for 75 GHz and 2a and 2b for 94 GHz. The complete set of measurements for each partcpant s presented n Appendx A, together wth the assocated uncertantes and degrees of equvalence wth respect to the reference value and between laboratores. Subsequent to the measurements, and n keepng wth current gudelnes on the presentaton of results, the partcpants were asked for more detals concernng the contrbutons to measurement uncertanty, for the prmary measured quantty only (effectve effcency). The uncertanty budgets for each laboratory are gven n Appendx B. The uncertantes quoted by each of the partcpants were at the one standard uncertanty level. The expanded uncertanty was obtaned by multplyng the standard uncertanty by a coverage factor of 2.0, whch was suffcent to provde a level of confdence for the expanded uncertanty of approxmately 95% for all the partcpants. The reference value for each parameter and at each frequency was obtaned from the unweghted mean of all the reported results, where the NPL result was obtaned from the mean of the two NPL measurements. The unweghted mean was selected for the reference value because all four laboratores gave smlar uncertantes and there were no results that could be consdered as outlers. The uncertanty n the reference value was obtaned from the root sum of squares of the ndvdual uncertantes dvded by the number of results, however, the NPL uncertanty was only counted once, see Appendx A page 11. Ths method s vald for uncorrelated measurements. The results show good agreement between all the partcpants. Ths s demonstrated by the crtera that none of the degrees of equvalence,, gven n tables 2 to 5 of Appendx A exceed the expanded uncertanty, U(). It can be noted that the measurements of all the partcpants are based on the use of mcrocalormeters, ether drectly or va transfer standards. It would be of great nterest to compare results from measurements based on some dfferent technque f and when one was developed at natonal standards level. 6 of 17

7 Measurement Results 0.76 Results for Effectve Effcency at 75 GHz Effectve Effcency PTB NPL (1) NIST NPL (2) The Reference Value s shown as the sold lne wth ts uncertanty ( k=2.0) as the dashed lne Fgure 1a: The effectve effcency at 75 GHz and ts assocated expanded uncertanty (k =2.0) gven by each of the partcpatng laboratores together wth the reference value and ts assocated uncertanty (k = 2.0) Results for Voltage Reflecton Coeffcent at 75 GHz Voltage Reflecton Coeffcent (Mag) NPL(1) NIST The Reference Value s shown as the sold lne wth ts uncertanty (k=2.0) as the dashed lnes NPL (2) Fgure 1b: The voltage reflecton coeffcent magntude 75 GHz and ts assocated expanded uncertanty (k =2.0) gven by each of the partcpatng laboratores together wth the reference value and ts assocated uncertanty (k = 2.0). 7 of 17

8 0.71 Results for Effectve Effcency at 94 GHz Effectve Effcency BNM-LCIE PTB NPL (1) NIST NPL (2) T he R eference V alue s shown as the sold lne wth ts uncertanty (k=2.0) as the dashed lnes Fgure 2a: The effectve effcency at 94 GHz and ts assocated expanded uncertanty (k =2.0) gven by each of the partcpatng laboratores together wth the reference value and ts assocated uncertanty (k = 2.0). Voltage Reflecton Coeffcent (Mag) Results for Voltage Reflecton Coeffcent at 94 GHz BNM-LCIE NPL (1) The Reference Value s shown as the sold lne wth ts uncertanty (k=2.0) as the dashed lnes NIST NPL (2) Fgure 2b: The voltage reflecton coeffcent magntude at 94 GHz and ts assocated expanded uncertanty (k =2.0) gven by each of the partcpatng laboratores together wth the reference value and ts assocated uncertanty (k = 2.0). 8 of 17

9 6. REFERENCES [1] Report of the Workng Group on Radofrequency Quanttes of 23 rd June 1997 at BIPM Sèvres, France. Table IV New comparsons [2] A Fantom, Rado frequency and Mcrowave Power Measurements. Peter Peregrnus Ltd. Publshers, IEE Electrcal Measurement Seres 7, [3] B. Torralba, J Achkar, evelopment of new mcrocalormeters n the frequency range GHz, n Proc. CPEM 96, 1996, pp [4] C. A. Hoer, Performance of a dual sx-port automatc network analyzer, IEEE Trans. on MTT, Vol MTT-27, ec 1979, pp [5] M. P. Wedman., A semautomated sx-port for measurng mllmeter-wave power and complex reflecton coeffcent, IEEE Trans. on MTT., vol MTT-12, ec 1977, pp [6] J. W. Allen., F. R. Clague, N. T. Larsen and M. P. Wedman., NIST mcrowave power standards n wavegude, NIST Tech. Note 1511, U. S. ept. of Commerce, February, [7] N. S. Chung, J. Shn, H. Bayer and R. Hongbaum, Coaxal and wavegude mcrocalormeters for RF and mcrowave power standards, IEEE Trans. Instrumentaton Meas., Vol. 38, No.2 Apr 1989, pp of 17

10 APPENIX A Key comparson CCEM.RF-K1d.W MEASUREANS: Effectve Effcency and Voltage Reflecton Coeffcent (Magntude) Plot laboratory: NPL(UK) η U(η ) Γ U(Γ ) result of measurement of effectve effcency carred out by laboratory expanded uncertanty (k = 2.0) of η, reported by laboratory result of measurement of voltage reflecton coeffcent (magntude) carred out by laboratory expanded uncertanty (k = 2.0) of Γ, reported by laboratory Lab Results and Expanded Uncertanty (k = 2.0) at 75 GHz Results and Expanded Uncertanty (k = 2.0) at 94 GHz ate of η U(η ) Γ U(Γ ) η U(η ) Γ U(Γ ) Measurements BNM-LCIE Sept 1998 PTB Jan 1999 NPL(1) March 1999 NIST Aprl 1999 NPL(2) ecember 1999 Mean (Ref) of 17

11 Key comparson CCEM.RF-K1d.W MEASUREANS: Effectve Effcency and Voltage Reflecton Coeffcent (Magntude) The key comparson reference values of ths comparson are obtaned from the mean of the results of the partcpants, as follows: η R η = BNM LCIE + η NPL 4 + η PTB + η NIST where: η NPL η = + η NPL( 1) NPL(2) 2 and Γ R Γ = BNM LCIE + Γ 3 NPL + Γ NIST where: Γ NPL Γ = + Γ NPL( 1) NPL(2) 2 The degrees of equvalence of each laboratory wth respect to the reference value s gven by a par of terms: η 2 2 = ( η η ) and ts expanded uncertanty U ( ) = 2.0 u ( η ) + u ( η ) 2cov( η η ) R η R, R where: u( η ) R = u ( η ) + u ( η ) + u ( η ) + u ( η ) BNM LCIE equvalent expressons were used to obtan PTB 4 NPL u Γ and Γ, R U Γ R NIST 11 of 17

12 Table 2 egrees of Equvalence for Effectve Effcency at 75 GHz Lab j Lab Reference BNM-LCIE PTB NPL NIST η U η η U η η U η η U η η BNM-LCIE PTB NPL NIST U η Table 3 egrees of Equvalence for Effectve Effcency at 94 GHz Lab j Lab Reference BNM-LCIE PTB NPL NIST η U η η U η η U η η U η η BNM-LCIE PTB NPL NIST U η 12 of 17

13 Table 4 egrees of Equvalence for Voltage Reflecton Coeffcent at 75 GHz Lab j Lab Reference BNM-LCIE PTB NPL NIST Γ U Γ Γ U Γ Γ U Γ Γ U Γ Γ U BNM-LCIE PTB NPL NIST Γ Table 5 egrees of Equvalence for Voltage Reflecton coeffcent at 94 GHz Lab j Lab Reference BNM-LCIE PTB NPL NIST Γ U Γ Γ U Γ Γ U Γ Γ U Γ Γ U BNM-LCIE PTB NPL NIST Γ 13 of 17

14 CCEM.RF-K1d.W Effectve Effcency n R900 egree of equvalence, η, and ts expanded uncertanty (k =2.0), U( η ) BNM-LCIE 94 NPL 75 NPL 94 η PTB 75 PTB 94 NIST 75 NIST The uncertanty n the reference value (k = 2.0) s shown by the dashed lnes for 75 GHz and the sold lne for 94 GHz Fgure 3a: egrees of Equvalence wth respect to the reference value for Effectve Effcency and the assocated expanded uncertantes (k = 2.0) CCEM.RF-K1d.W Reflecton Coeffcent n R900 egree of equvalence, Γ, and ts expanded uncertanty (k =2.0), U( Γ ) NPL 75 NPL 94 NIST 94 Γ BNM-LCIE 94 NIST The uncertanty n the reference value (k = 2.0) s shown by the dashed lnes and s the same value for both 75 GHz and 94 GHz Fgure 3b: egrees of Equvalence wth respect to the reference value for Voltage Reflecton Coeffcent (Magntude) and the assocated expanded uncertantes (k = 2.0). 14 of 17

15 APPENIX B The tables below gve detals of the uncertanty contrbutons approprate to the measurement of Effectve Effcency for each of the partcpants. Table 7 BNM-LCIE at 94 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Standard devaton of measurement Type A Gaussan Measurement of C voltages Type B Rectangular Reproducblty, msmatch Type B Gaussan Thn wall wavegude correcton Type B Gaussan Adapter correcton Type B Gaussan Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan Table 8 NPL(1) & (2) at 75 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Calbraton of Standard Thermstor Mount n Mcrcalormeter Type B Gaussan Measurement of C ratos Type B Gaussan Msmatch Correcton Type B Gaussan Flange connecton repeatablty Type A Rectangular Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan of 17

16 Table 9 NPL (1) & (2) at 94 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Calbraton of Standard Thermstor Mount n Mcrcalormeter Type B Gaussan Measurement of C ratos Type B Gaussan Msmatch Correcton Type B Gaussan Flange connecton repeatablty Type A Rectangular Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan Table 10 PTB at 75 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Bolometer C Voltage rato Type A Gaussan Correcton for C rato Type B Rectangular Mcrocalormeter thermal voltage rato Type A Gaussan Correcton of thermal voltage rato Type B Rectangular Mcrocalormeter equvalence correcton factor Type B Rectangular Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan Table 11 PTB at 94 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Bolometer C Voltage rato Type A Gaussan Correcton for C rato Type B Rectangular Mcrocalormeter thermal voltage rato Type A Gaussan Correcton of thermal voltage rato Type B Rectangular Mcrocalormeter equvalence correcton factor Type B Rectangular Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan of 17

17 Table 12 NIST at 75 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Calbraton of Standard n Mcrocalormeter Type B Gaussan Sx-port Transfer System Type A Gaussan Flange Connecton Repeatablty Type A Gaussan Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan Table 13 NIST at 94 GHz Source of uncertanty effectve method of probablty contrbuton to degrees of evaluaton dstrbuton standard uncertanty freedom Calbraton of Standard n Mcrocalormeter Type B Gaussan Sx-port Transfer System Type A Gaussan Flange Connecton Repeatablty Type A Gaussan Standard uncertanty Gaussan Expanded Uncertanty (k = 2.0) Gaussan Note: The uncertantes gven n the table n Appendx A page 10 are those reported to the plot laboratory by the partcpants and have, n some cases, been rounded up to two sgnfcant places compared to the values gven n the tables n Appendx B. The calculatons for egrees of Equvalence gven n Appendx B are based on the orgnally reported uncertantes. 17 of 17

18 GT-RF/ A Key comparson CCEM.RF-K1d.W MEASURAN : Effectve Effcency η : u η : result of measurement of effectve effcency carred out by laboratory standard uncertanty of η Lab 75 GHz 94 GHz ate of η u (η ) U ( η ) η u (η ) U ( η ) measurement BNM-LCIE PTB NPL(1) NIST NPL(2) Reference MEASURAN : Γ : u Γ : Reflecton Coeffcent result of measurement of reflecton coeffcent carred out by laboratory standard uncertanty of Γ Lab 75 GHz 94 GHz ate of Γ u(γ ) U ( Γ ) Γ u (Γ ) U ( Γ ) measurement BNM-LCIE PTB NPL(1) NIST NPL(2) Reference /2

19 GT-RF/ A Key comparson CCEM.RF-K1d.W MEASURANS : Effectve Effcency and Voltage Reflecton Coeffcent (Magntude) at 75 GHz and 94 GHz Note: Both Effectve Effcency and Reflecton Coeffcent are dmensonless quanttes Key comparson reference values: The arthmetc mean of the results has been used as the reference value for both effectve effcency and reflecton coeffcent For Effectve Effcency : η R = (η BNM-LCIE + η PTB + η NPL + η NIST )/4 where: η NPL = (η NPL(1) + η NPL(2) )/2 For Reflecton Coeffcent : Γ R = (Γ BNM-LCIE + Γ PTB + Γ NPL + Γ NIST )/4 where: Γ NPL = (Γ NPL(1 ) + Γ NPL(2) )/2 The degree of equvalence of each laboratory wth respect to the reference value s gven by a par of numbers: η = (η -η R ) and U ( η ), ts expanded uncertanty (k = 2.0) at the 95% Confdence Level U ( η ) = 2(u 2 (η ) + u 2 (η R )-2cov( η, η R )) 0.5 u (η R ) = (u 2 (η BNM-LCIE ) +u 2 (η PTB )+u 2 (η NPL )+u 2 (η NIST ) ) 0.5 /N N=Number of results equvalent expressons were used to obtan Γ, u (Γ R ) and U ( Γ ) The degree of equvalence between two laboratores s gven by a par of numbers: η = η - ηj and U (η ), ts expanded uncertanty (k = 2.0) at the 95% Confdence Level U (η ) = 2(u 2 (η ) + u 2 (η j )) 0.5 equvalent expressons were used to obtan Γ and U (Γ ) Effectve Effcency at 75 GHz Lab j Lab BNM-LCIE PTB NPL NIST η U ( η ) η U (η ) η U (η ) η U (η ) η U (η ) BNM-LCIE PTB NPL NIST Effectve Effcency at 94 GHz Lab j Lab BNM-LCIE PTB NPL NIST η U ( η ) η U (η ) η U (η ) η U (η ) η U (η ) BNM-LCIE PTB NPL NIST Reflecton Coeffcent at 75 GHz Lab j Lab BNM-LCIE PTB NPL NIST Γ U ( Γ ) Γ U (Γ ) Γ U (Γ ) Γ U (Γ ) Γ U (Γ ) BNM-LCIE PTB NPL NIST Reflecton Coeffcent at 94 GHz Lab j Lab BNM-LCIE PTB NPL NIST Γ U ( Γ ) Γ U (Γ ) Γ U (Γ ) Γ U (Γ ) Γ U (Γ ) BNM-LCIE PTB NPL NIST /2