975 SSB oise Figure Measuremets of Frequecy Traslatig Devices. Otegi,. Garmedia, J.M. Collates, M. Sayed Electricity ad Electroics Departmet, Uiversity of the Basque Coutry, Apdo. 644, 48080 Bilbao, Spai Microwave & MillimeterWave Solutios, Sata Rosa, Califoria, USA Abstract A complete procedure ad formulatio for Sigle- Side-Bad (SSB) oise figure characterizatio of frequecy coverters are proposed i this paper. The procedure takes ito accout systematic errors due to mismatch ad icludes receiver oise calibratio. Measuremets of three devices with differet coversio gai, oise ad match characteristics are give. Comparisos of the proposed methodology with several other approximatios of the SSB oise figure calculatio are also provided. Idex Terms oise figure, oise measuremets, mixer characterizatio. I. ITRODUCTIO oise figure characterizatio of frequecy coverters is a complex task that ca be affected by differet parameters (port matchig, coversio at image frequecy ad idler frequecies...). I additio, there is ofte cofusio ad misuderstadig regardig the stadard defiitio of oise figure i frequecy traslatig devices []. However, IEEE provides a clear defiitio for the oise figure at a specific iput frequecy []: F () kt G 0 where, is the total oise power per uit badwidth at the put frequecy available at the put port whe the oise temperature of its iput termiatio is T 0 90 K at all frequecies. kt 0 G is the portio of that is egedered at the iput frequecy by the iput termiatio at temperature T 0. I the case of a frequecy coverter, G is the available coversio gai relatig the iput frequecy to the put frequecy (cosiderig that i ormal operatio of the heterodye system, sigal is oly preset at the iput frequecy). Thus, the deomiator i () represets the oise from the iput termiatio which appears i the put via the pricipal frequecy trasformatio of the device G. This deomiator does ot iclude cotributios from imagefrequecy coversio or other idler frequecy coversio. It should be oted that this defiitio does ot exclude i ay oise geerated at image or idler frequecies, but, o the cotrary, excludes i the deomiator the gai that would correspod to a iexistet sigal. I this way, () represets the Sigle Side Bad (SSB) oise figure of a frequecy coverter. It is importat to remark that () is completely cosistet with the fact of cosiderig the oise figure as a figure of merit that characterizes the degradatio of the sigalto-oise ratio from the iput to the put of the device. Typical istrumets for oise figure characterizatio (oise figure aalyzers or, more recetly, spectrum aalyzers) make use of the well-kow Y-factor techique to perform the measuremets. I the Y-factor techique, the oise figure of the DUT is computed from the ratio Y of two oise powers measured with a oise source at two temperatures (T h, T c, hot ad cold temperatures respectively). The oise source is a broadbad device that provides extra oise i a wide frequecy rage whe biased i its hot state. Thus, i the case of frequecy coverters, the obtaied result is equivalet to a All-Side-Bad (ASB) oise figure i which all the oise cotributios of the coversio at image ad idler frequecies are preset i the deomiator (G,,G : coversio gais at image ad idler frequecies): ( Th T0 ) FASB () Y kt B G + G +... + G ( ) 0 Ofte, i order to obtai the SSB oise figure from a Y- factor techique the followig assumptio is made []: image coversio is equal to pricipal coversio (G G ) ad all idler coversios are egligible (G G 0). The, SSB oise figure is cosidered to be simply db higher tha the measured oe. Obviously, this assumptio will ot be satisfied uder ay coditios ad for ay frequecy traslatig device. Obtaiig a true SSB oise figure measuremet through a Y-factor techique requires the iclusio of a filter at the iput of the device that filters image ad idler frequecies. However, impedace termiatios of the mixer iput port at image ad idler frequecies ca have a o egligible ifluece o the device oise performaces []. Therefore, if the filter is ot required for the device regular operatio, some amout of error should be expected i the characterizatio. A additioal source of error i the stadard Y-factor oise figure characterizatio is related with the mismatch i DUT ad measuremet path. Sice Y-factor oly makes use of scalar power measuremets, o vector correctios are applied to accout for mismatch. However, systematic errors associated to device put mismatch are magified whe the device has loss istead of gai [4], which is the case of most mixers. I additio, poor iput match maximizes the error due to chages i the oise source reflectio coefficiet betwee its cold ad hot temperatures. 0-780-954-5/06/$0.00 006 IEEE
976 I this work a differet procedure is proposed to characterize the oise figure of frequecy traslatig devices. Istead of Y-factor, a vector-corrected cold-source techique is used, sice this oe allows a direct determiatio of the SSB oise figure. A Spectrum Aalyzer (SA) with oise measuremet capabilities is used to characterize the oise geerated by the DUT with a 50 Ω termiatio, at ambiet temperature, coected at its iput port. The spectrum aalyzer ad a sigal geerator serve to measure the pricipal coversio gai G. A vector corrected formulatio is give to compesate for all the mismatch errors. For that, two additioal steps are required. First, reflectio coefficiets at iput ad put frequecies have to be measured with a Vector etwork Aalyzer (VA). Secod, a oise calibratio of the oise receiver of the SA is also give i order to obtai its gai-badwidth product ad its four oise parameters. Three diode-based mixers with differet characteristics are measured followig the proposed procedure. I additio, comparisos with other simplified approaches are aalyzed. II. SSB OISE FIGURE CALCULATIO I order to obtai the SSB oise figure defied by (), the available oise power at the put of the device ad the available coversio gai are required. To compute the available gai, iput ad put match of the device must be characterized. I fact, the frequecy traslatig device (FTD) ca be cosidered as a cascade of two blocks, oe at iput frequecy ad oe at put frequecy, with a certai c coversio gai relatig the two blocks. I this sese, the FTD ca be see as a ordiary twoport, whose behavior is ruled by the Local Oscillator (LO) sigal ad where the coversio from iput to put implies also frequecy offset (Fig. ). IPUT block C OUTPUT block S S IPUT frequecy OUTPUT frequecy LO Fig.. Simplified scheme of a frequecy traslatig device Therefore, the available coversio gai from iput frequecy to put frequecy ca be expressed as i (): Γ G Γ () Γ ( Γ ) ( s ) ( s ) c ss where Γ s is the source reflectio coefficiet ad Γ is the put reflectio coefficiet of the FTD. It is importat to ote that s ad Γ must be obtaied uder operatig coditios, i.e. with the LO power at its operatig level, to correctly describe the device behavior. Besides, it should be remided that Γ s ad s are measured at iput frequecy, while Γ must be characterized at put frequecy. c ca be obtaied by meas of a SA ad a sigal geerator. For that, two oise power measuremets are performed: P, with the geerator directly coected to the SA ad P, which is the power delivered by the device to the SA. From these measuremets c ca be calculated as, ( Γ ) Γ s Γ Γ P SA _ Fi ge SA_ F P ( ΓSA _ F ) ΓgeΓSA_ Fi c where Γ ge is the reflectio coefficiet of the geerator ad Γ SA_Fi ad Γ SA_F are the iput reflectio coefficiets of the SA at iput ad put frequecies, respectively. With typical sigal geerator ad SA match characteristics, (5) ca be a fair approximatio of the available coversio gai. G ( Γs ) P Γss P Γ ( ) Oce device gai is characterized, the oise figure is computed from a uique oise power measuremet of the FTD, i the stadard maer of cold-source techiques. The oise power is measured, i our case, with the oise capabilities provided by the SA. For that, a 50 Ω load at ambiet temperature is coected at the iput of the device. The oise cotributio of the oise receiver of the SA must be elimiated applyig the secod stage correctio. Usig straight forward algebra, the SSB oise figure of the device defied by () ca be obtaied from (6). F SSB meas kt BG MM ( Γ ) G 0 rec ( )( ) Tc T0 G+ G +... + G Frec ( Γ T G G 0 where meas is the measured oise power, kbg rec the gaibadwidth product of the SA oise receiver ad F rec its oise figure. MM(Γ ) is a mismatch factor betwee device ad SA: _ rec Γ MM ( Γ ) s Γ with s _rec the iput reflectio coefficiet of the SA oise receiver at put frequecy. Equatio (6) is cumbersome sice it icludes coversio gais at image ad idler frequecies (G,,G ). However, i most cases, ambiet temperature T c ca be approximated to referece temperature T 0. I this case, the SSB oise figure ca be obtaied from: meas Frec ( Γ FSSB (8) kt BG MM ( Γ ) G G 0 rec The kbg rec term is computed i a previous calibratio stage, i which a stadard oise source is coected to the SA ad two power measuremets (cold ad hot states) are performed, c_rec ad h_rec. (4) (5) (6) (7)
977 kbg rec T T h _ rec c _ rec ( s h) ( rec( s h) ( h )) ( s c) ( rec( s c) ( c )) T MM Γ F Γ T + T T 0 0 T MM Γ F Γ T + T T 0 0 beig Γ s_c, Γ s_h the reflectio coefficiets of the oise source i its cold ad hot states, respectively. Fially, the depedece of the oise figure of the SA oise receiver o its source reflectio coefficiet Γ through four oise parameters has to be determied: R Γ Γ Frec ( Γ ) Fmi + 4 (0) Z +Γ ( Γ ) 0 This is a crucial step that caot be igored. eglectig this depedece i devices with low gai ad mediocre match (as could be the case of some mixers) leads to sigificat errors i the oise figure calculatio, as show i [4]. Here, a oise calibratio procedure origially developed for vector etwork aalyzers [5] is applied to obtai the four oise parameters F mi, R ad Γ of the SA oise receiver. Proceedig as i [5], the depedece of the secod stage correctio with source reflectio coefficiet is cocetrated i the followig term: I C + C Γ + C Γ cos Γ C () ( ) 4 where C, C, C ad C 4 are four equivalet oise parameters that ca be aalytically calculated from the measuremet of four kow passive loads. The the classical oise parameters ca be deduced from: C CZo Fmi C + ; R ( + Γ ) Γ 8 Γ () ( C+ C) Γ + Γ + 0 ; Γ C4 C where Z 0 is the characteristic impedace. This oise calibratio procedure avoids the use of typical tuers ad imizatio processes, which may be excessively cumbersome ad time-cosumig. III. MEASUREMET RESULTS Measuremet results of three diode-based mixers, with differet gai ad match characteristics, are give i this sectio. The three mixers have the same IF frequecy. The measuremet setup makes use of a SA, a VA, a oise source ad two sigal geerators i order to compute ecessary gai, mismatch ad oise powers as required by the procedure described i sectio II. The characterizatio is performed versus LO power sice the oise geerated by a diode-based mixer ca be affected by this oe [6]. Measuremet temperature T c ca be realistically approximated to T 0. (9) I the followig we will label the vector-corrected cold-source procedure defied by (8). This approach has bee applied to obtai the SSB oise figure of the three mixers uder study. I additio, results have bee compared to three differet approximatios of the SSB oise figure: Oe Y- Factor based ( ), ad two cold-source based approximatios with lower level of correctios ( ad ). - : This approximatio estimates the SSB oise figure by addig db to the result obtaied from a stadard Y-Factor measuremet, (). I this approach oly scalar measuremets are performed, therefore, either mismatch correctios or oise parameter calibratio are icluded. Parameters of the secod stage correctio (8) are approximated by F rec (Γ s ) ad all-side-bad isertio gai G is. These are calculated from the hot ad cold measuremets ( h_rec, c_rec ) performed with the oise source directly coected to the receiver. F ( T T ) F ( Γ ( db) h 0 rec s YF + + h_ meas c_ meas G is G is h_ meas c_ meas h_ rec c_ rec () () h_meas ad c_meas are the hot ad cold oise powers measured durig the DUT characterizatio. - : this approximatio, as show i (5), represets a scalar versio of (8). Available coversio gai is substituted by the isertio gai of the pricipal coversio, G is P /P. MM(Γ ) MM(Γ s_h ) MM(Γ s_c ) ad Γ s_c Γ s_h Γ s are cosidered i (8) ad (9). The oise figure of the receiver is agai approximated by F rec (Γ s ) sice o oise parameter calibratio is take ito accout. meas Frec ( Γs FCS _ SCALAR (5) kt BG G G 0 rec is is - : this approximatio adds all the correctios to except for the calibratio to get the four oise parameters of the SA oise receiver. Thus, the oly differece betwee this approach ad the proposed is the use of F rec (Γ s ) istead of F rec (Γ ) i (8). Fig. shows the results obtaied with the first mixer, Mix (RF GHz, LO GHz, IF GHz). Mix ca be see as a good test device i our case: it presets equal coversio loss for pricipal ad image frequecies ad coversios from idler-frequecies are egligible (G G, G G 0). I additio, put retur loss is better tha 0 db i all the aalyzed rage. As it could be expected, ad all the approximatios,, lead to comparable results. It ca be observed that, i this case, provides a good approximatio of the SSB oise figure, as expected from the coversio properties of the device. Besides, it is show that either vector correctios or oise calibratio are ecessary, due to the good match of Mix.
978 The secod mixer Mix (RF. GHz, LO 0. GHz, IF GHz) is ow cosidered. I this case, (G + G + + G ) is larger tha G i all of the measuremet rage, while put match is better tha 0 db. Results plotted i Fig. show that uderestimates the SSB oise figure. This is agai cosistet with the coversio loss characteristics of Mix. O the cotrary, the two cold-source approximatios, ad, provide idetical results to, due to the good put match of Mix. Therefore, ca be a judicious approach for obtaiig the SSB oise figure of mixers with fair match characteristics. SSB F (db) 0 4 6 8 0 Fig.. SSB oise figure characterizatio of Mix vs. LO power. SSB F (db) 0 4 6 8 0 Fig.. SSB oise figure characterizatio of Mix vs. LO power. SSB F (db) 0 4 6 8 0 Fig. 4. SSB oise figure characterizatio of Mix vs. LO power. Fially, Fig. 4 shows the measuremet results for the third case uder aalysis, Mix (RF GHz, LO GHz, IF GHz). This device, i additio to havig differet pricipal ad image coversio losses, is poorly matched at the put port, as show i Fig. 5. As it ca be observed i Fig. 4, four differet resposes have bee obtaied. The error associated to has icreased from Mix, due to the poor put match. For the same reaso, caot provide a accurate result. Fially, it should be oted the highest error give by, cofirmig that mismatch correctios require a receiver oise calibratio to be rigorous ad efficiet [4]. Γ (db) -.6 -.8-4 -4. -4.4-4.6 4 6 8 0 Fig. 5. Output reflectio coefficiet of Mix vs. LO power. IV. COCLUSIO A vector-corrected cold-source methodology ad formulatio are give to characterize the SSB oise figure of frequecy traslatig devices. Mismatch correctios ad receiver oise calibratio are icluded i the approach. Three diode-based mixers with differet characteristics have bee measured followig the proposed procedure. Measuremet results have bee compared with those from other approximatios of the SSB oise figure calculatio. Compariso results served to evidece which approaches ad approximatios are judicious depedig o device gai ad match characteristics, or, o the cotrary, whe the complete SSB formulatio is essetially required. ACKOWLEDGEMET The authors wish to ackowledge Agilet Techologies for their support. This work has bee fuded by Spaish Commissio of Sciece ad Techology (TIC00-0445). REFERECES [] S. A. Maas, oise i Liear ad oliear Circuits, Artech House, orwood, MA, 005 [] IRE Subcommittee 7.9 o oise, Descriptio of the oise Performace of Amplifiers ad Receivig Systems, Proc. IEEE, vol. 5, pp. 46-44, 96. [] S. A. Maas, Microwave Mixers, Secod Editio, Artech House, orwood, MA, 99 [4] J. M. Collates, R.D. Pollard, M. Sayed, Effects of DUT Mismatch o the oise Figure Characterizatio: A Comparative Aalysis of Two Y-Factor Techiques, IEEE Tras. Ist& Meas, vol. 5, o. 6, December 00, pp. 50-56. [5]. Otegi, J. M. Collates, M. Sayed, Calibrated oise Figure Measuremets i Vector etwork Aalyser, Electroic Letters, vol. 4, Issue, September 005, pp. 999-000 [6] G. M. Hegazi, A. Jeleski, K.S. Ygvesso, Limitatios of Microwave ad Millimeter-Wave Mixers Due to Excess oise, IEEE Tras. o MTT, Vol.,., Dec. 985, pp. 04-09
979 Trasiet Aalysis of Pulse Propagatio o Plaar Ultra-widebad Atea by Usig Trasiet Electroic ear-field Mappig System Kyoug-Hwa Oh ad Jog-I Sog Departmet of Iformatio ad Commuicatios, GIST Oryog-dog, Buk-gu, Gwagju, Korea 500-7, jisog@gist.ac.kr Abstract Trasiet aalysis of pulse propagatio o a coplaar waveguide (CPW)-fed rectagular slot atea for ultra-widebad (UWB) commuicatio system applicatio was performed usig a trasiet electroic ear-field mappig system. Details of short electrical pulse propagatio ad reflectio i the UWB atea were characterized. The system was very effective i characterizig trasiet behavior of short electrical pulses i the UWB atea. Idex Terms Electroic measuremets, ear-field, pulse propagatio, trasiet aalysis, UWB atea. I. ITRODUCTIO Ultra-widebad (UWB) commuicatios have received worldwide attetio sice the allocatio of the. 0.6 GHz frequecy rage by the Federal Commuicatios Commissio (FCC) i Feb., 00 [], []. Atea is oe of key compoets for UWB commuicatio systems ad various types of atea for UWB applicatio have bee proposed [], [4]. I a arrow-bad system, frequecy-domai parameters such as retur loss, gai, ad far-field radiatio patters at the specific operatig frequecy, are eough to evaluate a atea performace. However, i a UWB commuicatio system, above-metioed parameters do ot provide eough iformatio for atea performace over the UWB bad. Eve with the parameters coverig the etire UWB bad, it is still difficult to idetify whether a atea is suitable for UWB commuicatios. UWB commuicatio systems deal with geeratio ad trasmissio of short electric pulses i the order of pico-secods. These pulses have a broad badwidth coverig the FCC frequecy rage (. 0.6 GHz). Therefore, istead of usig the frequecy-domai parameters, UWB ateas should be ivestigated from a differet perspective. ew characterizatio approaches of UWB ateas that ca provide trasiet aalysis i the time-domai have bee proposed [5]-[8]. However, those approaches reported oly showed temporal shape of electrical pulses by usig a oscilloscope. Recetly, a trasiet electroic ear-field mappig system, which ca provide actual ear-field distributios of a plaar atea structure, was successfully demostrated for characterizig trasiet ear-field patters of plaar passive microwave circuits [9]. This trasiet electroic ear-field mappig system ca provide iformatio of propagatio ad iteral reflectio of short electric pulses i UWB ateas i detail. I this paper, trasiet aalysis of short pulse propagatio i a CPW-fed rectagular slot atea is demostrated by usig trasiet electroic ear-field mappig system. The system could visually show the trasiet behavior of short electrical pulse i the UWB atea. II. ATEA STRUCTURE Plaar-type ateas have the advatages icludig compact size ad ease of itegratio with trasceiver circuits. A CPW-fed rectagular slot atea proposed by R. Chair, et al. [0] for UWB atea was fabricated. Figure shows the fabricated UWB atea. The slot is etched at the ceter of a 60 mm 60 mm groud plae. The substrate (RO400) has the dielectric costat ad thickess of.8 ad 0.5 mm, respectively. The rectagular slot has the width ad legth of 6 mm ad 0.6 mm, respectively. A 50 Ω coplaar waveguide (slot width of 0.5-mm ad ceter coductor width of.8 mm) was used for feedig the atea. The U-shaped tuig stub embedded withi the slot termiates the CPW feed. Dimesio of the stub was mm alog x-directio ad 0 mm i z-directio. Fig. shows simulated ad measured scatterig parameters of the fabricated UWB atea which covers the FCC frequecy rage (. 0.6 GHz). Fig.. Fabricated UWB atea. The dotted area (0 mm mm) d 0-780-954-5/06/$0.00 006 IEEE