Network Analysis is NOT. Router Bridge Repeater Hub Your IEEE 802.3 X.25 ISDN switched-packet data stream is running at 147 MBPS with -9 a BER of 1.523 X 10...
W hat Types of Devices are Tested? Integration High Low D uplexers Diplexers Filters Couplers Bridges S plitters, dividers Combiners Isolators Circulators Attenuators Adapters O pens, shorts,loads Delay lines Cables Trans mission lines Waveguide Resonators Dielectrics R, L, C's Antennas S witches Multiplexers Mixers Sa mplers Multipliers Diodes RFICs MMICs T/R modules Transceivers Receivers Tuners Converters VCAs Amplifiers VCOs VTFs Oscillators M odulators VCAtten s Transistors Passive Device type Active
Device Test Measurement Model Complex Response tool Si mpl e 84000 Ded. Testers VSA SA VNA TG/SA SNA NF Mtr. Imped. An. Param. An. Power Mtr. Det/Scope Harm. Dist. LO stability Image Rej. I-V LCR/Z Gain/Flat. Compr'n Phase/GD AM-PM Isolation Rtn Ls/VS W R Impedance S-parameters Absol. Power Gain/Flatness NF NF RFIC test Intermodulation Distortion Measurement plane DC CW Swept Swept Noise 2-tone M ulti- Co mplex Pulsed- Protocol freq power tone modulation RF Si mple Complex Stimulus type BER EVM ACP Regrowth Constell. Eye Full call sequence Pulsed S-parm. Pulse profiling
Lightwave Analogy to RF Energy Incident Trans mitted Reflected Lightwave DUT RF
Why Do We Need to Test Components? Verify specifications of building blocks for more complex RF systems Ensure distortionless transmission of com munications signals linear: constant amplitude,linear phase /constant group delay nonlinear: harm onics,intermodulation, compression, AMto-PM conversion Ensure good match when absorbing power (e.g., an antenna) KPWR FM 97
The Need for Both Magnitude and Phase 1. Complete characterization of linear networks 2. Complex impedance needed to design matching circuits S 21 S 11 S 22 S 12 4. Time-do main characterization Mag 3. Co m plex values needed for device modeling High-frequency transistor model Time 5. Vector-error correction Base Emitter Collector Measured Error Actual
Agenda W hat measurem ents do we make? Transmission-line basics Reflection and transmission parameters S-parameter definition Network analyzer hardware Signal separation devices Detection types Dynamic range T/R versus S-parameter test sets Error models and calibration Types of measurement error One- and two-port models Error-correction choices Basic uncertainty calculations Exa mple measurements Appendix
Transmission Line Basics Low frequencies wavelengths >> wire length + - I current(i)travels down wires easilyforefficient power transmission m easured voltage and current not dependent on position along wire High frequencies wavelength or << length oftransmission medium need trans mission lines for efficient power trans mission matching to characteristicimpedance (Zo) is very importantforlow reflection and maximum power transfer measured envelope voltage dependenton
Transmission line Zo Zo determines relationship between voltage and current waves Zo is a function of physical dimensions and ε r Zo is usuallya realimpedance (e.g. 50 or 75 ohms) Twisted-pair 1.5 Wave guide 1.4 1.3 attenuation is lowest at 77 ohms a b Coaxial εr h h normalized values 1.2 1.1 1.0 0.9 0.8 50 ohm standard w2 w1 w 0.7 0.6 power handling capacity peaks at 30 ohms 0.5 10 20 30 40 50 60 70 80 90100 Coplanar Microstrip characteristicimpedance for coaxial airlines (ohms)
Power TransferEfficiency RS Load Power (nor malized) RL 1.2 1 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 7 8 9 10 RL / RS For complex im pedances, maximu m power transfer occurs when ZL = ZS* (conjugate match) Rs +jx -jx RL Maximum power is transferred when RL = RS
Transmission Line Terminated with Zo Zs = Zo Zo = characteristic impedance of transmission line Zo V inc Vrefl = 0! (allthe incident power is absorbed in the load) For reflectio n, a transmission line terminated in Zo behaves like an infinitely long transmission line
Transmission Line Terminated with Short, Open Zs = Zo V inc Vrefl In-phase (0 o ) for open, out-of-phase (180 o ) for short For reflection, a trans mission line ter minated in a short or open reflects all power back to source
Transmission Line Terminated with 25 Ω Zs = Zo ZL = 25 Ω V inc Vrefl Standing wave pattern does not go to zero as with shortor open
High-Frequency Device Characterization Incident R Reflected A REFLECTION Trans mitted B TRANSMISSION Reflected Incident = A R Trans mitted Incident = B R SWR S-Parameters S 11, S 22 Reflection Coefficient Γ, ρ Return Loss Impedance, Admittance R+jX, G+jB Gain / Loss S-Parameters S 21, S 12 Trans mission Coefficient Τ,τ Insertion Phase Group Delay
Reflection Parameters Reflection Coefficient Γ = V reflected V incident = ρ Return loss = -20 log(ρ), ρ = Φ Γ = Z L Z O Z L + ZO E max E min Voltage Standing Wave Ratio E max VS WR = E min = 1 + ρ 1 - ρ No reflection (ZL = Zo) Fullreflection (ZL = open, short) 0 ρ 1 db RL 0 db 1 VS WR
Smith ChartReview. +jx Polar plane 90 o 1.0 0 +R.8.6 + 180 o.4.2-0 o -jx Rectilinear impedance plane Smith Chart maps rectilinear impedance plane onto polar plane 0 Z = 0 L Γ Z = Zo L Γ = 0-90 o (short) Z L = = 1 ±180 O Smith chart Constant X Γ Constant R = 1 0 O (open)
Transmission Parameters V Incident DUT V Trans mitted Transmission Coefficient = Τ = V Trans mitted VIncident = τ φ V Trans Insertion Loss (db) = - 20 Log V Inc = -20 log τ V Trans Gain (db) = 20 Log V Inc = 20 log τ
Linear Versus Nonlinear Behavior Sin 360 o * f * t A t o A A * Sin 360 o * f (t - t o ) Time phase shift = to * 360 o * f Linear behavior: input and output frequencies are the same (no additional frequencies created) outputfrequency only undergoes magnitude and phase change Time f 1 Frequency Input DUT Output f 1 Frequency f 1 Time Frequency Nonlinear behavior: outputfrequency may undergo frequency shift (e.g. with mixers) additionalfrequencies created (harmonics, intermodulation)
Criteria for Distortionless Transmission Linear Networks Constant amplitude over band width ofinterest Linear phase over band width of interest Magnitude Phase Frequency Frequency
Magnitude Variation with Frequency F(t) = sin wt + 1/3 sin 3wt + 1/5 sin 5wt Time Time Linear Network Magnitude Frequency Frequency Frequency
Phase Variation with Frequency F(t) = sin wt + 1 /3 sin 3wt + 1/5 sin 5wt Linear Network Time Time Magnitude Frequency 0-180 Frequency Frequency -360
Deviation from Linear Phase Use electrical delay to remove linear portion of phase response o Phase 45 /Div RF filter response Linear electricallength added (Electrical delay function) + yields Deviation from linear phase o Phase 1 /Div Frequency Frequency Frequency Low resolution High resolution
Group Delay Phaseφ φ Frequency ω ω t g t o Group delay ripple Average delay Group Delay (t ) g = d φ d ω = 1 360 o d φ * d f φ ω φ in radians in radians/sec in degrees f in Hertz (ω = 2 π f) Frequency group-delay ripple indicates phase distortion average delay indicates electricallength of DUT aperture of measurement is very important
W hy Measure Group Delay? Phase Phase d φ d ω f d φ d ω f Group Delay Group Delay f Sa me p-p phase ripple can result in different group delay f
C haracterizing Unknown Devices Using param eters (H,Y, Z, S) to characterize devices: gives linear behavioral model of our device measure parameters (e.g.voltage and current) versus frequency under various source and load conditions (e.g. short and open circuits) compute device parameters from measured data predict H-para circuitperformance meters Y-para under meters any source Z-para and meters load conditions V1 = h11i1 + h12v2 I1 = y11v1 + y12v2 V1 = z11i1 + z12i2 I2 = h21i1 + h22v2 I2 = y21v1 + y22v2 V2 = z21i1 + z22i2 h11 = V1 I1 V2=0 h12 = V1 V2 I1=0 (requires short circuit) (requires open circuit)
W hy Use S-Parameters? relatively easy to obtain at high frequencies measure voltage traveling waves with a vector network analyzer don't need shorts/opens which can cause active devices to oscillate or self-destruct relate to familiar measurements (gain,loss,reflection coefficient...) can cascade S-parameters of multiple devices to predict system performance can co mpute H, Y, or Z parameters from S-parameters if desired can easilyimport and Incident S21 use S-parameterfiles Trans mitted in our electronicsimulation tools S11 a 1 b2 Reflected DUT S22 Port 1 Port 2 b Reflected 1 a2 Trans mitted S12 Incident b1 =S11a 1 + S 12 a2 b2 = S21 a 1 + S22 a2
Measuring S-Parameters Incident S 21 Transmitted Forward b 1 a 1 b 2 Z 0 S 11 Reflected DUT a 2 = 0 Load S 11 = Reflected Incident S 21 = Transmitted Incident = b 1 a 1 a 2 = 0 = b 2 a 1 a 2 = 0 S 22 = Reflected Incident S 12 = Transmitted Incident = b 2 a 2 a 1 = 0 = b 1 a 2 a 1 = 0 Z 0 Load a 1 = 0 b 1 DUT Trans mitted S 12 S 22 Reflected Incident b 2 a 2 Reverse
Equating S-Para meters with Com m on Measurement Terms S11 = forward reflection coefficient (input match) S22 = reverse reflection coefficient (output m atch) S21 = forward trans mission coefficient (gain or loss) S12 = reverse transmission coefficient (isolation) Re me mber, S-parameters are inherently co m plex, linear quantities -- however, we often express the m in a log- magnitude format
Criteria for Distortionless Transmission Nonlinear Networks Saturation, crossover, intermodulation, and other nonlinear effects can cause signal distortion Effect on system depends on amount and type of distortion and system architecture Time Time Frequency Frequency
SPE CT R U M AN A LYZ E R 9 k Hz - 26.5 G H z Measuring Nonlinear Behavior Most com m on measurements: using a network analyzer and power sweeps gain compression A M to PM conversion using a spectru m analyzer + source(s) harmonics, particularly second and third intermodulation products resulting from two or more RF carriers LPF DUT 8563A RL 0 dbm ATTEN 10 db 10 db / DIV LPF CENTER 20.00000 MHz SPAN 10.00 khz RB 30 Hz VB 30 Hz ST 20 sec
. What is the Difference Between Network and Spectrum Analyzers? 8563A SPECTRU M ANALYZER 9 k Hz - 26.5 GHz Amplitude Ratio Measures known signal Amplitude Measures unknown signals Frequency Network analyzers: measure components, devices, circuits, sub-asse mblies contain source and receiver display ratioed am plitude and phase (frequency or power sweeps) offer advanced error correction Frequency Spectru m analyzers: measure signal am plitude characteristics carrierlevel,sidebands, harmonics...) can de modulate (& measure) complex signals are receivers only (single channel) can be used forscalar component test (no h ) ith t ki t
Agenda What measurements do we make? N etwork analyzer hardware Error models and calibration Example measurements Appendix
Generalized Network Analyzer Block Diagram Incident DUT Trans mitted SOURCE Reflected SIGN AL SEPARATION INCIDENT (R) REFLECTED (A) TRANSMITTED (B) RECEIVER / DETECTOR PROCESSOR / DISPLAY
Source Supplies stimulus for system Swept frequency or power Traditionally NAs used separate source Most Agilent analyzers sold today have integrated, synthesized sources
Signal Separation SOURCE Incident Reflected DUT Transmitted SIG N AL SEPARATION INCIDENT (R) REFLECTED (A) TRANSMITTED (B) RECEIVER / DETECTOR measure incident signalfor reference separate incident and reflected signals PROCESSOR / DISPLAY splitter bridge directional coupler Detector Test Port
Directivity Directivity is a measure ofhow well a coupler can separate signals moving in opposite directions (undesired leakage signal) (desired reflected signal) Test port Directional Coupler
Interaction of Directivity with the DUT (WithoutError Correction) 0 D UT RL = 40 db Data Max Return Loss 30 Device Directivity Add in-phase 60 Frequency Directivity Device Data Min A dd out-of-phase (cancellation) Directivity Device Data = Vector Sum
D etector Types Diode Scalar broadband (no phase information) SOURCE INCIDENT (R) Incident Reflected REFLECTED (A) DUT SIG N AL SEPARATION TRANSMITTED (B) RECEIVER / DETECTOR Transmitted RF DC AC PROCESSOR / DISPLAY Tuned Receiver RF IF = F LO ± F RF ADC / DSP Vector (magnitude and phase) IF Filter LO
Broadband Diode Detection Easy to make broadband Inexpensive compared totuned receiver Good for measuring frequency-translating devices Improve dynamic range by increasing power Medium sensitivity / dynamic range 10 MHz 26.5 GHz
Narrowband Detection -Tuned Receiver ADC / DSP Best sensitivity / dynamic range Provides harmonic /spurious signal rejection Improve dyna mic range by increasing power, decreasing IF band width, or averaging Trade off noise floor and measurementspeed 10 MHz 26.5 GHz
Comparison of Receiver Techniques 0 db Broadband (diode) detection 0 db Narrowband (tuned-receiver) detection -50 db -50 db -100 db -100 db -60 dbm Sensitivity < -100 dbm Sensitivity higher noise floor false responses high dynamic range harmonic immunity Dyna mic range = maximum receiver power - receiver noise floor
Dynamic Range and Accuracy 100 Error Due to Interfering Signal 10 - Error (db, deg) 1 0.1 + magn error phase error Dyna mic range is very important for measurem e nt accuracy! 0.01 0.001 0-5 -10-15 -20-25 -30-35 -40-45 -50-55 -60-65 -70 Interfering signal (db)
T/R Versus S-Parameter Test Sets Trans mission/reflection Test Set S-Para meter Test Set Source Source Transfer switch R R A B A B Port 1 Port2 Port 1 Port 2 Fwd DUT Fwd DUT Rev RF always comes out port 1 port 2 is always receiver response, one-port cal available RF comes out port 1 or port 2 forward and reverse measurements two-port calibration possible
HP-IB STAT US Processor / Display Incident DUT Transmitted 50 MH-20GHz NETWORK ANYZER SOURCE Reflected CH1 S 21 log MAG 10 db/ REF 0 db 1_ -1.9248 db CH2 S 12 log MAG 10 db/ REF 0 db 1_ -1.2468 db 839.470 000 MHz PRm Duplexer Test - Tx-Ant and Ant-Rx Cor 1 1 ACTIVE CHANNEL RESPONSE ENTRY SIG N AL SEPARATION Hld 1 PRm 880.435 000 MHz PASS STIMULUS INSTRUMENT STATE R CHANNEL Cor 2 INCIDENT (R) REFLECTED (A) TRANSMITTED (B) Hld CH1 START 775.000 000 MHz CH2 START 775.000 000 MHz PASS STOP 925.000 000 MHz STOP 925.000 000 MHz T R L S RECEIVER / DETECTOR PORT 1 PORT 2 PROCESSOR / DISPLAY CH1 S 21 log MAG 10 db/ REF 0 db CH2 S 12 log MAG 10 db/ REF 0 db PRm Duplexer Test - Tx-Ant and Ant-Rx Cor 1 1_ -1.9248 db 1_ -1.2468 db 839.470 000 MHz 1 markers limitlines pass/failindicators linear/log formats grid/polar/smith charts Hld 1 PRm Cor Hld CH1 START 775.000 000 MHz CH2 START 775.000 000 MHz PASS 880.435 000 MHz 2 PASS STOP 925.000 000 MHz STOP 925.000 000 MHz
InternalMeasurement Automation Simple:recallstates More powerful: Test sequencing available on 8753/ 8720 families keystroke recording some advanced functions IBASIC available on 8712 family sophisticated programs custom user interfaces ABC DEF G HIJKLM N O P Q RSTUV W X YZ0123456789 + -/* = < > ( ) & "" ",./? ;:'[] 1 ASSIGN @ Hp8714 TO 800 2 OUTPUT @Hp8714;"SYST:PRES; *WAI" 3 OUTPUT @Hp8714;"ABOR;:INIT1:CONT OFF;*WAI" 4 OUTPUT @ Hp8714;"DISP:ANN:FREQ1:MODE SSTOP" 5 OUTPUT @ Hp8714;"DISP:ANN:FREQ1:MODE CSPAN" 6 OUTPUT @ Hp8714;"SENS1:FRE Q:CENT 175000000 HZ;*WAI" 7 OUTPUT @Hp8714;"ABOR;:INIT1:CONT OFF;:INIT1;*W AI" 8 OUTPUT @ Hp8714;"DISP:WIND1:TRAC:Y:AUTO ONCE" 9 OUTPUT @Hp8714;"CALC1:MARK1 ON" 10 OUTPUT @Hp8714;"CALC1:MARK:FUNC BWID" 11 OUTPUT @Hp8714;"SENS2:STAT ON; *WAI" 12 OUTPUT @Hp8714;"SENS2:FUNC 'XFR:PO W:RAT 1,0';DET NBAN; *WAI" 13 OUTPUT @Hp8714;"ABOR;:INIT1:CONT OFF;:INIT1;*W AI" 14 OUTPUT @ Hp8714;"DISP:WIND2:TRAC:Y:AUTO ONCE" 15 OUTPUT @Hp8714;"ABOR;:INIT1:CONT ON;*WAI" 16 END
Agilent s Series of HF Vector Analyzers Microwave 8720ET/ES series 13.5, 20, 40 GHz economical fast, small,integrated test mixers, high-power amps RF 8712ET/ES series 1.3, 3 G Hz low cost narrowband and broadband detection IBASIC / LAN 8510C series 110 GHz in coax highest accuracy modular, flexible pulse systems Tx/Rx module test 8753ET/ES series 3, 6 G Hz highest RF accuracy flexible hardware more features Offset and harmonic RF sweeps
Agilent slf/rf Vector Analyzers Co mbination NA / SA 4395A/4396B 500 MHz (4395A), 1.8 GHz (4396B) impedance-measuring option fast, FFT-based spectrum analysis time-gated spectru m-analyzer option IBASIC standard testfixtures LF E5100A/B 180, 300 MHz economical fast, small target markets: crystals,resonators,filters equivalent-circuit models evaporation-monitor-function option
SPE CT R U M AN A LYZ E R 9 k Hz - 26.5 G H z Spectrum Analyzer / Tracking Generator RF in DUT LO IF 8563A Spectrum analyzer TG out f= IF Tracking generator DUT Key differences from network analyzer: one channel--no ratioed or phase measurem ents More expensive than scalar NA (butbetter dynamic range) Only errorcorrection available is normalization (and possibly open-short averaging) Poorer accuracy Smallincremental costif SA isrequired forother measurements
Agenda W hat measurements do we make? N etwork analyzer hardware Error models and calibration Example measurements Appendix W hy do we even need error-correction and calibration? Itisimpossibleto make perfect hardware It would be extremely expensive to make hardware good enough to eliminate the need forerror correction
Calibration Topics W hat measurements do we make? Network analyzer hardware Error models and calibration measurementerrors what is vector error correction? calibration types accuracy examples calibration considerations Example measurements Appendix
Measurement Error Modeling CAL RE-CAL Syste matic errors due to imperfections in the analyzer and testsetup assumed to be time invariant (predictable) Rando m errors vary with time in random fashion (unpredictable) main contributors: instru ment noise, switch and connector repeatability Drift errors due to system performance changing after a calibration has been done primarily caused by temperature variation Errors: Measured Data SYSTE MATIC RANDO M DRIF T Unknown Device
Systematic Measurement Errors R Directivity A Crosstalk B DUT Frequency response reflection tracking (A/R) transmission tracking (B/R) Source Mismatch Load Mismatch Six forward and six reverse error terms yields 12 error terms for twoport devices
Types of Error Correction response (normalization) simple to perform only corrects fortracking errors stores reference trace in me mory, thru then does data divided by m e mory vector requires more standards requires an analyzerthat can measure phase accounts forall major sources of systematic error SHORT S 11a OPEN thru S 11m LOAD
What is Vector-Error Correction? Process of characterizing systematic error terms measure known standards remove effects from subsequent measurements 1-port calibration (reflection measurements) only 3 systematic error term s measured directivity, source match, and reflection tracking Full 2-port calibration (reflection and transmission measurements) 12 systematic errorterms measured usually requires 12 measurem ents on four known standards (SOLT) Standards defined in cal kit definition file network analyzer contains standard calkit definitions CAL KIT DEFINITION MUST MATCH ACTUAL CAL KIT USED! User-builtstandards mustbe characterized and entered into user al kit
Reflection: One-Port Model RF in Ideal RF in Error Adapter 1 E D = Directivity S11 A E D E S S11A E RT E S = Reflection tracking = Source Match S11 M S11 M E RT S11 M S11 A = Measured = Actual To solve for error terms, w e measure 3 standards to generate 3 equations and 3 unknowns S11A S11M = E D + ERT 1 - ES S11A Assumes good termination at porttwo iftesting two-portdevices Ifusing port 2 of NA and DUT reverse isolation islow (e.g.,filter passband): assu mption of good termination is not valid two-port error correction yields betterresults
Before and After O ne-port Calibration 0 2.0 Return Loss (db) 20 40 data before 1-port calibration 1.1 1.01 VS WR 60 data after 1-port calibration 1.001 6000 1 MHz
a 1 b 1 Two-Port Error Correction Forward model Port 1 EX Port 2 S 21 E b A TT 2 E S E D S 11 S 22 a A A 2 E L E RT S 12 A Reverse model Port 1 Port 2 E RT' S 21 a A 1 E L' S 11 S 22 E E D' A A S' b 1 E TT' S 12 A E X' b 2 a 2 E D E S E RT E D' E S' E RT' = fwd directivity = fwd source match = fwd reflection tracking = rev directivity = rev source match = rev reflection tracking Each actual S-parameteris a function of allfour measured S-parameters Analyzer mustmake forward and reverse sweep to update any one S-parameter Luckily,you don'tneed to know these equations to use network analyzers!!! E L E TT E X E L' ETT' E X' = fwd load match = fwd transmission tracking = fwd isolation = rev load match = rev transmission tracking = rev isolation 11 S m E D S E E RT E S E m E X S 1 22 ' 21 12m E X ' ( )( + ') L ( )( ) RT ' E TT E TT ' S m E D' S E m E D S E S E RT E S E L E m E X S 1 11 + 1 22 ' + 21 12m E X ' ( )( ') ' L ( )( ) RT ' E TT E TT ' S m E D S11 a = 21 S22m E D ' ( )( 1 + ( E E TT E S ' E L )) RT ' S m E D S 1 11 E m E D E S 1 22 ' S ' ( + )( + E RT E S ') E L ' E ( 21m E X S )( 12m E X ) RT ' L E TT E TT ' S m E X S21 a = S12a = S22a = S E ' S E ( 12m X )( 1 11m D + ( E ' )) E TT ' E S E L RT S ' ( m E D S ' E )( m E D S ') ' ( )( ) E S E RT E RT ' S E L E m E X S 1 11 1 22 21 12m E X + + L E TT E TT ' S22m ' ( E D S )( 11 m E D S ' E ) '( )( ) E RT ' E S E 21 m E X S 12 m E X 1 + L RT E TT E TT ' S11 m E D S ( 1 + 22m E E D ' S21m E S E RT E S E L E E X S12m E X ' )( 1 + ') ' L ( )( ) RT ' E TT E TT '
Crosstalk: Signal Leakage Between Test Ports During Trans mission Can be a problem with: high-isolation devices (e.g., switch inopen position) high-dynamic range devices (some filter stopbands) Isolation calibration adds noise to error model(measuring near noise floor of system) only perform ifreally needed (use averaging if necessary) ifcrosstalkis independent of DUT match, use two terminations ifdependent on DUT match, use DUT with termination on output LOAD DUT DUT LOAD DUT
Convenient Generally not accurate No errors removed Errors and Calibration Standards UNCORRECTED RESPONSE FULL 2-PORT DUT thru DUT Easy to perform Use when highest accuracy is not required Removes frequency response error ENHANCED-RESPONSE Co mbines response and 1-port Corrects source match fortransmission measurements SHORT OPEN LOAD DUT For reflection measurements Need good termination for high accuracy with twoport devices Removes these errors: Directivity Source match Reflection tracking 1-PORT SHORT OPEN LOAD DUT Highest accuracy Removes these errors: Directivity Source,load match Reflection tracking Transmission tracking Crosstalk thru SHORT OPEN LOAD
Calibration Summary Reflection Reflection tracking Directivity Source match Load match Test Set (cal type) T/R (one-port) S-parameter (two-port) SHORT OPEN LOAD * error can be corrected error cannot be corrected enhanced response cal corrects for source match during transmission measurements Trans mission Trans mission Tracking Crosstalk Source match Load match * ( ) Test Set (caltype) T/R S-parameter (response, isolation) (two-port)
Reflection Example Using a One-Port Cal Directivity: 40 db (.010) Load m atch: 18 db (.126) DUT 16 db RL (.158) 1 db loss (.891) Remember:convert all db values to linear for uncertainty calculations! -db ( ) 20 ρ orloss (linear) = 10.158 (.891)(.126)(.891) =.100 Measure ment uncertainty: -20 *log (.158 +.100 +.010) = 11.4 d B (-4.6d B) -20 *log (.158 -.100 -.010) = 26.4 d B (+10.4 db)
Using a O ne-port Cal + Attenuator Directivity: 40 db (.010).158 Load m atch: 18 db (.126) 10 db attenuator (.316) SWR = 1.05 (.024) DUT 16 db RL (.158) 1 db loss (.891) Measurement uncertainty: -20 *log (.158 +.039) = 14.1 d B (-1.9 d B) -20 *log (.158 -.039) = 18.5 d B (+2.5 db) (.891)(.316)(.126)(.316)(.891) =.010 (.891)(.024)(.891) =.019 Worst-case error =.010 +.010 +.019 =.039 Low-loss bi-directional devices generally require two-port calibration forlow measure ment uncertainty
Transmission Example Using Response Cal RL = 14 db (.200) RL = 18 db (.126) Thru calibration (normalization) builds error into measure ment due to source and load match interaction Calibration Uncertainty = (1 ± ρ S ρ L ) = (1 ± (.200)(.126) = ± 0.22 db
Filter Measurement with Response Cal Source match = 14 db (.200) DUT 1 db loss (.891) 16 db RL (.158) Load match = 18 db (.126) 1 (.126)(.158) =.020 (.126)(.891)(.200)(.891) =.020 Total measure ment uncertainty: +0.60 + 0.22 = + 0.82 db -0.65-0.22 = -0.87 db (.158)(.200) =.032 Measurement uncertainty = 1 ± (.020+.020+.032) = 1 ±.072 = + 0.60 db -0.65 db
Measuring Amplifiers with a Response Cal Source match = 14 db (.200) DUT 16 db RL (.158) Load match = 18 db (.126) 1 (.126)(.158) =.020 (.158)(.200) =.032 Total measure ment uncertainty: +0.44 + 0.22 = + 0.66 db -0.46-0.22 = - 0.68 db Measurement uncertainty = 1 ± (.020+.032) = 1 ±.052 = + 0.44 db -0.46 db
Source match = 35 db (.0178) Filter Measurements using the Enhanced Response Calibration Effective source match = 35 db! DUT 1 db loss (.891) 16 db RL (.158) Load match = 18 db (.126) Calibration Uncertainty = (1 ± ρ S ρ L ) = (1 ± (.0178)(.126) = ±.02 db 1 (.126)(.158) =.020 (.126)(.891)(.0178)(.891) =.0018 (.158)(.0178) =.0028 Measurement uncertainty = 1 ± (.020+.0018+.0028) = 1 ±.0246 = + 0.211 db -0.216 db Total measurem ent uncertainty: 0.22 +.02 = ± 0.24 db
Source match = 35 db (.0178) Using the Enhanced Response Calibration Plus an Attenuator 10 db attenuator (.316) SWR = 1.05 (.024 linear or 32.4 db) Analyzer load match =18 db (.126) DUT 1 db loss (.891) 16 db RL (.158) Effective load match = (.316)(.316)(.126) +.024 =.0366 (28.7dB) 1 (.0366)(.158) =.006 (.0366)(.891)(.0178)(.891) =.0005 (.158)(.0178) =.0028 Calibration Uncertainty = (1 ± ρ S ρ L ) = (1 ± (.0178)(.0366) = ±.01 db Measurement uncertainty = 1 ± (.006+.0005+.0028) = 1 ±.0093 = ± 0.08 db Total measurem ent uncertainty: 0.01 +.08 = ± 0.09 db
C alculating Measurement Uncertainty After a Two-Port Calibration Corrected error ter ms: (8753ES 1.3-3 GHz Type-N) Directivity = 47 db Source match = 36 db Load match = 47 db Refl.tracking =.019 db Trans.tracking =.026 db Isolation = 100 db Trans mission uncertainty = 0.891 ±.0056 = 1 db ±0.05 db (worst-case) DUT 1 db loss (.891) 16 db RL (.158) Reflection uncertainty 2 11m 11a D 11a S 21a 12a L 11a RT S = S ± ( E + S E + S S E + S ( 1 E )) 2 2 = 0158. ± (. 0045 + 0158. *. 0158 + 0. 891 *. 0045 + 0158. *. 0022) = 0.158 ±.0088 = 16 db +0.53 db, -0.44 db (worst-case) S = S ± S ( E / S + S E + S S E E + S E + ( 1 E )) 21m 21a 21a I 21a 11a S 21a 12a S L 22a L TT 6 2 = 0. 891 ± 0. 891( 10 / 0. 891 + 0158. *. 0158 + 0. 891 *. 0158*. 0045 + 0158. *. 0045+. 003)
Response versus Two-Port Calibration Measuring filter insertion loss CH1 S21&M log M A G 1 db/ R E F 0 db CH2 MEM log M A G 1 db/ REF 0 db Cor Aftertwo-port calibration Afterresponse calibration Cor Uncorrected x2 1 2 STA RT 2 000.000 MHz S T O P 6 000.000 M H z
ECal:Electronic C alibration (85060/90 series) Variety of modules cover 30 khz to 26.5 GHz Six connector types available (50 Ω and 75 Ω) Single-connection reduces calibration time makes calibrations easy to perform minimizes wear on cables and standards eliminates operator errors Highly repeatable temperaturecompensated terminations provide excellent accuracy σα 85093A Electronic Calibration Module 30 khz - 6 GHz Microwave modules use a transmission line shunted by PIN-diode switches in various combinations
Adapter Considerations reflection from adapter leakage signal Coupler directivity = 40 db desired signal ρ measured = Directivity + adapter + ρ ρ DUT Adapter DUT Termination D U T has S M A (f)connectors W orst-case System Directivity Adapting from APC-7 to SMA (m) AP C-7 calibration done here 28 db APC-7 to SMA (m) S W R:1.06 17 db APC-7 to N (f) + N (m) to SM A ( m) S W R:1.05 S W R:1.25 14 db APC-7 to N (m) + N (f)to SMA (f) + S M A (m) to (m) S W R:1.05 SWR:1.25 SWR:1.15
Calibrating Non-Insertable Devices W hen doing a through cal, normally test ports m ate directly cables can be connected directly without an adapter resultis a zero-length through What is an insertable device? has same type of connector, but different sex on each port has same type of sexless connector on each port(e.g. APC-7) What is a non-insertable device? one that cannot be inserted in place of a zerolength through has same connectors on each port(type and sex) has differenttype of connector on each port (e.g., waveguide on one port,coaxial on the other) What alibration hoi es do I have for non- DUT
Swap Equal Adapters Method Port 1 DUT Port 2 Adapter Port 1 A Port 2 Accuracy depends on how well the adapters are matched -loss, electricallength, match and impedance should all be equal 1. Trans mission cal using adapter A. Port 1 Adapter Port 2 B 2. Reflection cal using adapter B. Port 1 DUT Adapter Port 2 B 3. Measure D UT using adapter B.
AdapterRemovalCalibration Calibration isvery accurate and traceable Infirmware of8753, 8720 and 8510 series Also accomplished with ECal modules (85060/90) Uses adapterwith same connectors as DUT Must specify electricallength of adapterto within 1/4 wavelength of highestfrequency (to avoid phase ambiguity) Port 1 DUT Port 2 Cal Port 1 Adapter Port 2 Adapter B C al Set 1 Port 1 Cal Adapter Adapter B Port 2 C al Set 2 [CAL][MORE] [MODIFY CAL SET] [ADAPTER REMOVAL] 1. Perform 2-port cal with adapter on port2. Save in cal set 1. 2. Perform 2-port cal with adapter on port1. Save in cal set 2. 3. Use ADAPTE R RE M O V AL to generate new cal set. Port 1 DUT Adapter B Port 2 4. Measure DUT without cal adapter.
Thru-Reflect-Line (TRL) Calibration W e know about Short-Open-Load-Thru (SOLT) calibration... What is TRL? A two-port calibration technique Good fornoncoaxial environm ents (waveguide, fixtures, wafer probing) Uses the sa m e 12-term error model as the more com mon SOLT cal Uses practical calibration standards that TRL was developed for noncoaxial microwave are easily fabricated and characterized Two variations: TRL (requires 4 receivers) measurements and TRL* (only three receivers needed) Other variations: Line-Reflect-Match (LRM), Thru-Reflect-Match (TR M), plus many others
Agenda What measurements do we make? N etwork analyzer hardware Error models and calibration Example measurements Appendix
Frequency Sweep - Filter Test CH1 S 21 log MAG 10 db/ REF 0 db CH1 S 11 log MAG 5 db/ REF 0 db Cor 69.1 db Stopba nd rejectio n START.300 000 MHz STOP 400.000 000 MHz Cor CENTER 200.000 MHz CH1 S 21 log MAG 1 db/ REF 0 db m1: 4.000 1000 G Hz - 0.16 db m2-ref: 2.145 234 G Hz 0.00 db ref 2 SPAN 50.000 MHz Return loss Insertion loss Cor x2 1 2 START 2 000.000 MHz STOP 6 000.000 MHz
Optimize Filter Measurements with Swept-List Mode PRm CH1 S21 log MAG 12 db/ REF 0 db Segment 3: 29 ms (108 points, -10 dbm, 6000 Hz) Linear sweep: 676 ms (201 pts, 300 Hz, -10 dbm) Swept-list sweep: 349 ms (201 pts, variable B W's & power) PASS Segment 1: 87 ms (25 points, +10 dbm, 300 Hz) Segment 5: 129 ms (38 points, +10 dbm, 300 Hz) START 525.000 000 MHz STOP 1 275.000 000 MHz Segments 2,4: 52 ms (15 points, +10 dbm, 300 Hz)
Power Sweeps -Compression Output Power (dbm) Compression region Linear region (slope = small-signal gain) Saturated output power Input Power (dbm)
Power Sweep -Gain Compression CH1 S21 1og MA G 1 db/ REF 32 db 30.991 db 12.3 db m 0 1 1 db co mpression: input power resulting in 1 db drop in gain START -10 dbm CW 902.7 MHz STOP 15 dbm
AM (db) PM (deg) Amplitude A M to PM Conversion Measure ofphase deviation caused by a mplitude variations Power sweep Mag(Am i n ) DUT A M can be undesired: supply ripple,fading,thermal A M can be desired: modulation (e.g. QAM) Test Stimulus Time Amplitude Q AM -PM Conversion = Mag(Pm Mag(Am ) out ) in (deg/d B) AM (db) PM (deg) Output Response Time Mag(AM o ut ) Mag(Pm o ut ) AM to PM conversion can cause bit errors I
Measuring AM to PM Conversion 1:Transmission Log Mag 1.0 db/ Ref 21.50 db 2:Trans mission /M Phase 5.0 deg/ Ref -115.7 deg Ch1:Mkr1-4.50 db m 20.48 db Ch2:Mkr2 1.00 db 0.86 deg 1 2 Use trans mission setup with a power sweep Display phase of S21 AM -PM = 0.86 deg/db 2 Start-10.00 db m Start-10.00 db m 1 C W 900.000 MHz CW 900.000 MHz 1 Stop 0.00 dbm Stop 0.00 dbm
Agenda W hat measurements do we make? N etwork analyzer hardware Error models and calibration Example measurements Appendix Advanced Topics time domain frequency-translating devices high-power a mplifiers extended dynamic range multiport devices in-fixture measurements crystalresonators balanced/differential Inside the network analyzer Challenge quiz!
Time-Domain Reflectometry (TDR) impedance Zo What is TDR? time-domain reflectometry analyze impedance versus time distinguish between inductive and capacitive transitions With gating: analyze transitions analyzer standards capacitive transition inductive transition non-zo transmission line time
TDR Basics Using a Network Analyzer start with broadband frequency sweep (often requires microwave VN use inverse-fouriertransform to compute time-domain resolution inversely proportionate to frequency span Time Do main Frequency Do main F -1 CH1 S22 Re 50 mu/ REF 0 U t f Cor 20 GHz 6 G Hz t 0F(t)*dt Integrate 1/s*F(s) TDR F -1 t f C H1 STA RT 0 s ST O P 1.5 ns
Time-Domain Gating TDR and gating can remove undesired reflections (a form of error correction) Only usefulfor broadband devices (a load orthru for exa mple) Define gate to only include D UT CH1 S11& M log M A G 5 db/ R E F 0 db Use two-portcalibration PRm Cor CH1 MEM Re 20 mu/ REF 0 U PRm Cor RISE TIME 29.994 ps 8.992 m m 1 1: 48.729 mu 638 ps 2: 24.961 mu 668 ps 3: -10.891 m U 721 ps Gate 2 1: -45.113 db 0.947 GHz 2: -15.78 db 6.000 GHz 2 3 C H1 STA RT 0 s Thru in time domain ST O P 1.5 ns 1 START.050 000 000 GHz Thru in frequency domain, with and without gating STO P 20.050 000 000 GHz
Ten Steps for Performing TDR 1. Set up desired frequency range (need wide span for good spatialresolution) 2. Under SYST E M, transform menu, press "setfreq low pass" 3. Perform one- or two-port calibration 4. Select S11 measurement * 5. Turn on transform (low pass step) * 6. Set format to real* 7. Adjust transform window to trade offrise time with ringing and overshoot * 8. Adjust start and stop times if desired 9. For gating: set start and stop frequencies for gate turn gating on * adjust gate shape to trade offresolution with ripple * 10. To display gated response in frequency domain * If using turn two transform channels off(leave (even ifcoupled), gating these on)* parameters must be set independentlyfor change format second to log-magnitude channel *
Time-Domain Transmission RF Input RF Output Main Wave Leakage CH1 S21 log M A G 15 db/ REF 0 db Cor Triple Travel CH1 S21 log MA G 10 db/ REF 0 db Cor RF Leakage Surface Wave Triple Travel Gate off Gate on STA RT -1 us STOP 6 us
Time-Domain Filter Tuning Deterministic method used fortuning cavity-resonator filters Traditionalfrequencydomain tuning is very difficult: lots oftraining needed m ay take 20 to 90 minutes to tune a single filter Need VNA withfast sweep speeds and fasttimedo main processing
Filter Reflection in Time D o m ain Set analyzer s center frequency = centerfrequency ofthe filter Measure S 11 or S 22 in the time domain Nullsin the time-domain response correspond to individual resonators in filter
Tuning Resonator #3 Easierto identify mistuned resonator in time-domain: null#3 is missing Hard to tell which resonatoris mistuned from frequencydomain response Adjustresonators by minimizing null Adjust coupling apertures using the peaks in-between the dips
Frequency-Translating Devices Medium-dynamic range measurements (35 db) 8753ES 1 2 Ref In Attenuator Filter Attenuator FREQ OFFS ON off LO MENU DOWN CONVERTER UP CONVERTER RF > LO 8753ES High-dynamic range measurements (100 db) Ref in Ref out Reference mixer Start:900 MHz Stop: 650 MHz Fixed L O: 1 G Hz LO power:13 db m Start:100 MHz Stop: 350 MHz RF < LO VIE W MEASURE RETURN Filter Attenuator CH1 CONV MEAS log MAG 10 db/ REF 10 db Attenuat or DUT Attenuator START 640.000 000 MHz STOP 660.000 000 MHz ES G-D4000A Powe r splitt er
High-Power Amplifiers 8753ES Prea mp Ref In Source Prea mp AUT DUT R A B AUT +43 db m max input (20 watts!) 8720ES Option 085 85118A High-Power Amplifier Test System
High-Dynamic Range CH1 MEM LOG 15 db/ REF 3 db Measurements CH2 MEM LOG 15 db/ REF 3 db PRm Cor Avg 10 PRm Standard 8753ES Cor Avg 10 8753ES SpecialOption H16 C H1 START 775.000 000 M Hz C H2 START 775.000 000 M Hz STOP 1000.000 000 M Hz STOP 1000.000 000 M Hz
Multiport Device Test PRm Cor Hld PRm Cor Hld 1 CH1 S 21 log MAG 10 db/ REF 0 db 1_ -1.9248 db CH2 S 12 log MAG 10 db/ REF 0 db 1_ -1.2468 db Duplexer Test - Tx-Ant and Ant-Rx CH1 START 775.000 000 MHz CH2 START 775.000 000 MHz 8753 H39 1 839.470 000 MHz 1 880.435 000 MHz PASS PASS STOP 925.000 000 MHz STOP 925.000 000 MHz 2 Multiport analyzers and testsets: improve throughput by reducing the number of connections to D U Ts with more than two ports allow simultaneous viewing oftwo paths (good fortuning duplexers) include mechanical or solid-state switches, 50 or 75 ohms degrade raw performance so calibration is a must(use two-port cals whenever possible) Agilent offers a variety of standard and custom multiport analyzers and test sets
87050E/87075C Standard Multiport Test Sets For use with 8712E family 50 Ω: 3 M H z to 2.2 G Hz, 4, 8, or 12 ports 75 Ω: 3 M H z to 1.3 G Hz, 6 or 12 ports Test Set Cal and SelfCal dra matically improve calibration times Systems offer fully-specified performance attest ports Once a month: perform a Test Set Cal with external standards to re m ove systematic errors in the analyzer,test set,cables, and fixture Test Set Cal Fixture SelfCal DUT Once an hour: automatically perform a SelfCal using internal standards to remove systematic errors in the analyzer and test set
Test Set Cal Eliminates Redundant Connections of Calibration Standards Reflection Connections Through Connections 12-port 12-port 8-port 8-port 4-port 4-port 0 100 200 300 400 0 25 50 75 Test Set Cal Traditional VN A Calibration
In-Fixture Measurements Measurement proble m: coaxial calibration plane is notthe same as the in-fixture measurement plane Calibration plane Measurement plane Fixture E D E S DUT E T Loss Error correction with coaxial calibration Phase shift Mis match
Characterizing CrystalResonators/Filters E5100A/B Network Analyzer Ch1 Z: R phase 40 / REF 0 1: 15.621 U 31.998 984 925 M Hz Min Cor 1 START 31.995 MHz SEG START STOP POINTS POWER IFB W STOP 32.058 MHz 1 31.995 MHz 32.008 MHz 200 0 db m 200Hz > 2 32.052 MHz END 32.058 MHz 200 0 db m 200Hz Exa mple of crystalresonator measurement
Balanced-Device Measurements ATN-4000 series (4-porttest set + software) measure tough singled-ended devices like couplers measure fully-balanced or single-ended-to-balanced DUTs characterize mode conversions (e.g. com mon-to-differential) incorporates 4-port error correction for exceptional accuracy works with 8753ES and 8720ES analyzers more info at ww w.atn m icrowave.co m σα Channel Partner
Traditional Scalar Analyzer Incident DUT Transmitted SOURCE Reflected processor/display source INCIDENT (R) REFLECTED (A) SIG N AL SEPARATION TRANSMITTED (B) RECEIVER / DETECTOR PROCESSOR / DISPLAY Example: 8757D/E requires external detectors, couplers, bridges,splitters good forlow-cost microwave scalar applications RF R A B RF R A B D etector D etector Reflection Bridge DUT Termination DUT Trans mission D etector
Directional Coupler Directivity Coupling Factor (fwd) x Loss (through arm) Directivity = Isolation (rev) Directivity (db) = Isolation (db) -Coupling Factor (db) - Loss (db) 50 db 20 db Examples: Test port Directivity = 50 db - 20 d B = 30 db 50 db 30 db 10 db Test port Directivity = 50 db - 30 d B - 10 db = 10 db 60 db 20 db 10 db Test port Directivity = 60 db - 20 d B - 10 db = 30 db
One Method of Measuring Coupler Directivity 1.0 (0 db) (reference) Coupler Directivity 35 db (.018) Source short.018 (35 db) (norm alized) Directivity = 35 db - 0 db = 35 db Source load Assume perfect load (no reflection)
Directional Bridge 50 Ω 50 Ω Detector 50 Ω Test Port 50-ohm load at test port balances the bridge -- detector reads zero Non-50-ohm load imbalances bridge Measuring magnitude and phase ofimbalance gives complex impedance "Directivity"is difference between maximu m and minimu m balance
NA Hardware:Front Ends, Mixers Versus Sa mplers SOURCE INCIDENT (R) Incident Reflected REFLECTED (A) DUT SIG N AL SEPARATION TRANSMITTED (B) RECEIVER / DETECTOR Transmitted PROCESSOR / DISPLAY ADC / DSP Sa mpler-based front end S ADC / DSP Mixer-based front end Itis cheaper and easier to make broadban d front ends using sa mplers instead of mixers Harmonic generator f frequency "comb"
Three Versus Four-Receiver Analyzers Source Source Transfer switch Transfer switch R R1 A B A B R2 Port 1 Port 2 3 receivers more economical TRL*,LR M* cals only includes: 8753ES 8720ES (standard) Port 1 Port 2 4 receivers more expensive true TRL, LRM cals includes: 8720ES (option 400) 8510C
W hy Are Four Receivers Better Than Three? TRL TRL* 8720ES Option 400 adds fourth sa mpler, allowing full TRL calibration TRL* assumes the source and load match of a testport are equal (port sym metry between forward and reverse measurements) thisis only a fairassumption for three-receiver network analyzers TRL four receivers are necessary to make the required measure ments TRL and TRL* use identical calibration standards In noncoaxial applications, TRL achieves better source and load match correction than TRL* What about coaxialapplications? SOLT is usually the preferred calibration method coaxial TRL can be more accurate than SOLT, but not com m only used
Challenge Quiz 1. Can filters cause distortion in co m m unications systems? A. Yes, due to impairment ofphase and magnitude response B. Yes, due to nonlinear components such as ferrite inductors C. No, only active devices can cause distortion D. No, filters only cause linear phase shifts E. Both A and B above 2. Which statem ent about trans mission lines is false? A. Usefulfor efficienttransmission of RF power B. Requires termination in characteristicimpedance forlow VSW R C. Envelope voltage of RF signalisindependent of position along line D. Used when wavelength of signalis smallcompared to length ofline E. Can be realized in a variety offorms such as coaxial, waveguide, microstri 3. Which statement about narrowband detection is false? A. Is generallythe cheapest way to detect microwave signals B. Provides much greater dynamic range than diode detection C. Uses variable-bandwidth IF filters to set analyzer noise floor D. Provides rejection of harmonic and spurious signals E. Uses mixers or samplers as downconverters
Challenge Quiz (continued) 4. Maximu m dyna mic range with narrowband detection is defined as: A. Maximum receiverinput power minus the stopband ofthe device under te B. Maximum receiverinput power minus the receiver's noise floor C. Detector 1-dB-compression point minus the harmonic levelofthe source D. Receiver damage level plus the maximu m source output power E. Maximum source output power minus the receiver's noise floor 5. With a T/R analyzer,the following error terms can be corrected: A. Source match,load match,transmission tracking B. Load match,reflection tracking,transmission tracking C. Source match,reflection tracking,transmission tracking D. Directivity, source match,load match E. Directivity,reflection tracking,load match 6. Calibration(s) can re m ove w hich of the following types of measure ment A. Systematic and drift B. Systematic and random C. Random and drift D. Repeatability and systematic E. Repeatability and drift
Challenge Quiz (continued) 7. Which statem ent about TRL calibration is false? A. Is a type oftwo-porterror correction B. Uses easilyfabricated and characterized standards C. Most com monly used in noncoaxial environments D. Is not available on the 8720ES family of micro wave network analyzers E. Has a specialversion forthree-sa mpler network analyzers 8. For which co m p o nent is ithardest to get accurate trans mission and reflection measure ments w hen using a T/R network analyzer? A. A mplifiers because output power causes receiver compression B. Cables because load match cannot be corrected C. Filter stopbands because oflack of dynamic range D. Mixers because oflack of broadband detectors E. Attenuators because source match cannot be corrected 9. Power sweeps are good for which measurements? A. Gain compression B. AM to PM conversion C. Saturated output power
Answers to Challenge Quiz 1. E 2. C 3. A 4. B 5. C 6. A 7. D 8. B 9. E