Calibration of RF-Voltage on Oscilloscopes

Transcription

1 Calibration of RF-oltage on Oscilloscopes 23 st ANAMET Meeting 3 rd /4 th March 2005 Jürg Furrer Swiss Federal Office of Metrology and Accreditation metas indenweg 50, CH-3003 Bern-Wabern, Switzerland Phone: Mail: juerg.furrer@metas.ch

2 Calibration of RF-oltage on Oscilloscopes Motivation Accreditation assessments of labs, which calibrate oscilloscopes Goal (not yet reached) calibrations based on a harmonized reference document clear definitions of used voltage reference user friendly data within certificate of calibration realistic measurement uncertainies METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 2

3 Calibration of RF-oltage on Oscilloscopes Parameter: Flatness, eveled Sine Wave (Frequency response) Uncertainty calculation (1) Rf oltage depends on reflection coefficients Γ G & Γ Osc Γ Osc (input SWR) of oscilloscope often not specified Generator source match (Γ G ) not exactly known EA10/7 (calibration of oscilloscopes) gives not adequate help:... if reflection coefficient is not negligible high-impedance oscilloscopes: up to the highest frequencies considered, the impedance shall be significantly larger than 50 Ω, otherwise additional uncertainties will be introduced... abs doing oscilloscope calibrations often do not have a NA METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 3

4 Calibration of RF-oltage on Oscilloscopes Parameter: Flatness, eveled Sine Wave (Frequency response) Uncertainty calculation (2) definition voltage often not clear: calibration related to - incident voltage: Inc - total voltage at load: = Inc + Ref feedtrough termination (used for 1 MΩ-oscilloscope input) shifts the reference plane, gives additional phase change used generator (and its leveling) METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 4

5 Traceability of RF voltage up to 100 MHz by TC (Thermal oltage Converters) up to GHz range by RF-power and Impedance Todays Oscilloscope Calibrators (e.g. Fluke 5800, Fluke 9500) in mode eveled Sine Wave RF-oltage source (up to... 6 GHz) based on a built in 50Ω RF generator calibrated in RF power by means of calibrated power-sensor (e.g. 8481A and diode-sensor 8481D (down to 42 dbm or 5 mpp) calculated (& displayed) in volt pp, based on Z = Z 0 = 50 Ω METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 5

6 Oscilloscope calibration setup for frequency response and -3 db bandwidth case a) Oscilloscope input impedance is 1 MΩ a 50Ω-feedthrough termination is used case b) Oscilloscope input impedance is 50 Ω Oscilloscope Calibrator eveled Sine Wave (50 Ω RF-Generator) Feedthrough Terminator 50 Ω Oscilloscope DUT Input 1 MΩ Oscilloscope Calibrator eveled Sine Wave (50 Ω RF-Generator) Oscilloscope DUT Input 50 Ω METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 6

7 Oscilloscope calibration procedure (according to Oscilloscope service manual) - connect... - set oscilloscope... - set reference frequency f ref, e.g. 50 khz (Generator) - set voltage to 1 olt pp (Generator) - take reference reading v (fref) (DUT) - set frequency to f test (Generator) - take reading v (ftest) (DUT) - calculate ratio 20 log ( v (ftest) / v (fref) ) -... up to 3 db point measured frequency response and -3 db point f METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 7

8 (1) What happens exactly by doing this calibration procedure? oscilloscope input has Γ DUT > 0.00 (SWR > 1.00) e.g. input capacity C IN therefore: oading effect: voltage at reference plane (DUT input voltage) change with frequency voltage at reference plane changes, is no longer equivalent to calibrator level Γ DUT Γ G Oscilloscope Calibrator eveled Sine Wave (50 Ω RF-Generator) Oscilloscope Calibrator eveled Sine Wave (50 Ω RF-Generator) reference plane Feedthrough Terminator 50 Ω Oscilloscope DUT Input 50 Ω Γ = 0.2 (SWR 1.5) Oscilloscope DUT Input 1 MΩ C IN = 12pF METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 8

9 (2) What happens exactly by doing this calibration procedure? f ref f test = reference frequency (typical: 50 khz) = test frequency (example: 150 MHz) measured voltage ratio s on oscilloscope (DUT): DUT _ fref DUT _ ftest R fref = R ftest = rp _ fref rp _ ftest voltage at reference plain ( rp ), depending on: rp_ref (P rp_fref, Z 0, Γ DUT_fref, Γ G_fref ) rp_test (P rp_ftest, Z 0, Γ DUT_ftest, Γ G_ftest ) - rp_fref : voltage at refernce reference freq. - rp_ftest : voltage at refernce test freq. - DUT : voltage read on oscilloscope (DUT) - R: voltage ratio (DUT) - P rp : power at reference plane - Γ DUT, Γ G complex reflexion coefficient (DUT, Gen.) DUT frequency response between f ref and f test = R ftest / R fref DUT input conditions vs. frequency: voltage vs. frequency at reference plane is freqency dependent, due to Γ DUT (SWR DUT ) & Γ G METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 9

10 (3) What happens exactly by doing this calibration procedure?... but... Oscilloscope calibration procedure (according to Oscilloscope service manual) set frequency to f ref (Generator) - take reference reading v (fref) (DUT) - set frequency to f test (Generator) - take reading v (ftest) (DUT) - calculate ratio 20 log ( v (ftest) / v (fref) ) -... up to 3 db point... this procedure assumes that the applied voltage at the reference plane is equivalent to the voltage indicated by the generator and therefore frequency independent... only the readings at the oscilloscope (at f ref and f test ) are taken into account applied voltage: at least the generator voltage reading is constant (within a given tolerance) METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 10

11 but my generator is leveled (leveled sine wave) generator / oscilloscope calibrator Sine wave Generator Module Attenuator Module oltage Sensing Module Output? Output down to 5 m pp generator / oscilloscope calibrator Sine wave Generator Module oltage Sensing Module Attenuator Module Output Output voltage error depends on load (DUT) and generator reflexion coeff. (SWR on the line between generator and oscilloscope) METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 11

12 leveled sine wave generator 50 Ω MON Port 3 Power Monitor (Powermeter) Γ DUT G Γ DUT Port 2 Power Splitter Γ G Generator Γ G DUT DUT REF Power Standard Oscilloscope Oscilloscope eveled sine wave generator vs. powersplitter arrangement leveled sine wave generator is calibrated via RF power in Z o do we get the same incident power into an arbitrary Γ DUT, assuming that for both setups the available power in to a ideal load (P Zo ) is equivalent? how reacts the leveling (leveled sine wave generator)? METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 12

13 Absolute value of measured RF voltage level and its uncertainty at frequency f test : What assumes the user of a calibrated scope? calibrated frequency response f ref frequency response between f ref and f test f test f measured voltage level at frequency f test and its uncertainty = a) value of absolute voltage level at f ref and its uncertainty b) + frequency response at f test and its uncertainty METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 13

14 Generator connected to oad (general case) Signal flow graph: allows to determine the voltage on the load a = INC Z 0 wave towards load b = REF Z 0 wave reflected from load a 2 = P INC incident power to load calculated voltages, expressed with Zo Γ G, Γ : b 2 Z0 = PZ 0 Z S = 0 2 available power into ideal matched load Z 0 = PZ 0 Z0 voltage on a ideal load P INC INC PZ 0 = 2 1 ΓG Γ incident power to load Z0 = 1 Γ Γ voltage due to the incident wave Z 1 0 = 1 Γ G G + Γ Γ voltage on the load METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 14

15 = INC + REF = voltage at oscilloscope input (ref. plane) INC Z 0 = 1 Γ Γ G Z 1 0 = 1 Γ G + Γ Γ voltage due to the incident wave (at reference plane) voltage on the load (at reference plane) Γ = Γ oad = Γ DUT = complex reflexion coefficient of oscilloscope Γ G = complex reflexion coefficient of generator Γ G, Γ : usually no phase information available Oscilloscope output is voltage * frequency response calibration of RF-voltage: referenced to or INC referenced to? METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 15

16 influence of Γ and its phase? 1+ Γ = 1 Γ Γ Z 0 simulation of with all phase angles (phase Γ ) G this term is dominating theoretical model of oscilloscope input circuit: assuming 1 MΩ scope input, C IN = 15 pf, with 50 Ω feedthrough terminator, f test = 150 MHz Oscilloscope Calibrator Z = j 1 X C IN simplified (without 1 MΩ ) reference plane feedthrough terminator Oscilloscope Input Γ = Z Z Z + Z Γ = φ -109 f = 150 MHz R par = 50 Ω C par = 15 pf 0 0 Z 0 = 50 Ω METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 16

17 First step: calculation of with / without phase restriction of Γ neglecting 1 - Γ G Γ, because Γ G Γ << 1 Zo = voltage at reference plane into ideal load Z = Z 0 = calibration of the generator voltage (scope calibrator) Zo = 100 % = 1 * 1 + Γ Z 1 0 = 1 Γ + Γ G Γ Γ = φ -109 (50Ω 150 MHz) = 1 + Γ = (94.3% of Zo ) but: phase (Γ ) normaly not measured! with no phase restriction of Γ : = Zo * (1±0.33) - how sensitive is to the phase of Γ? - What is the actual phase (Γ ) vs. theoretical scope input model? METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 17

18 (1) Second step: calculation of with phase (Γ ) restricted to ± 10 Γ = φ -109 (50Ω 150 MHz) φ ± 10 Zo = 100 % Z 1 0 = 1 Γ + Γ G Γ Simulation of formula, complex: 1 - Γ G Γ is no longer neglected Γ = 0.1 Phase (Γ G ) is stepped in 10 steps Γ G Γ phase(γ ) phase(γ ) Γ G Γ Phase(Γ ) Phase(Γ ) Y-Axis: Γ G = 0 (ideal) Γ G = 0 (ideal) φ (Γ ) = ± 10 A φ (Γ ) = ± 0 B METAS J. Furrer, RF- Power aboratory X-Axis: # of calculated phase-steps X-Axis: # of calc. phase-steps φ(γ G )=10, φ(γ )=1 ANAMET 23th Meeting 3. / 4. March 2005 slide 18

19 (2) Second step: calculation of with phase (Γ ) restricted to ± 10 Γ G Γ phase(γ ) phase(γ ) Γ = φ -109 (50Ω 150 MHz) φ -109 ±10 influence of term 1-Γ G Γ interaction between generator & load Y-Axis: Γ G = 0.1 C φ (Γ ) = ± 10 Z 1 0 = 1 Γ + Γ G Γ Y-Axis: Γ G = 0.1 φ (Γ) = ± 180 no restriction of φ( Γ ): normally not measured Γ D X-Axis: # of calc. phase-steps φ(γ G )=10, φ(γ )=1 METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 19

20 comparison of theoretical and measured Γ = Γ scope : 2 oscilloscopes C IN = 9 pf and C IN = 30pF (1 MΩ input, parallel with 50Ω-feedthrough-terminator) Γ SCOPE Γ SCOPE Γ SCOPE Γ SCOPE Freq C IN = 9 pf C IN = 9 pf C IN = 30 pf C IN = 30 pf theoretical measurement theoretical measurement MHz Magin Phase Magin Phase Magin Phase Magin Phase Mag Γ : theoretical model and measurement is quite similar Phase Γ : measured phase is shifted much more than theoretical phase Reason: line length between R p and C IN, C IN is distributed length of 50Ω-feedthrough-terminator Result: oading effect is even worse than theoretical prediction results in higher uncertainty METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 20

21 comparison of measured input reflexion factors Γ left: oscilloscope type X C IN = 30 pf bandwidth (specification) 350 MHz right: oscilloscope type Y C IN = 9 pf bandwidth (specification) 350 MHz of 2 oscilloscopes with switchable 1MΩ and 50Ω input 350 MHz oscilloscope type X C IN = 30 pf 350 MHz oscilloscope type Y C IN = 9 pf Γ = 0.5 1MΩ input // 50 Ω feedthrough term. Γ = 0.5 1MΩ input // 50 Ω feedthrough term. 50 Ω input 50 Ω input Input circuit or measured Γ completely different for 1 MΩ-input and 50 Ω- input: case 1 MΩ - input: capacitive component dominates, Γ can be estimated out of C IN case 50 Ω - input: Γ 0.15 within specified bandwidth METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 21

22 measurement of Γ on 5 different oscilloscopes (each: BW 250 MHz, C IN = pF) (1 MΩ input, parallel with 50 Ω feedthrough-terminator) Graph shows the phase deviation to the theoretical input model (R p = 50 Ω // C IN ) Result: phase deviation is not consistent with increasing C IN phase shifts more than in theoretical model phase deviation (Γ ) to the theoretical input model (deg) METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 22

23 measurement of Γ on 5 different oscilloscopes (each: BW 250 MHz, C IN = pF) phase deviation to the theoretical input model for our model case (R p = 50 Ω // C IN = 15pF) + phase deviation (Γ ) to the theoretical input model (deg) - Phase Offset = -25 ± 15 METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 23

24 Third step: Calculation of with phase (Γ ) according measured phaseshift theoretical model: Γ = φ -109 (50Ω 150 MHz) φ ± 10 practical measurements: Γ = φ: offset to -109 = -25 ± 15 φ -134 ± 15 Z 1 0 = 1 Γ + Γ G Γ Γ G Γ phase(γ ) phase(γ ) METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 24

25 Comparison of and INC (under the same conditions) It is evident that oscilloscope frequency response (or calibration on RF voltage ) is quite depending on the two reference voltages or INC Z 1 0 = 1 Γ + Γ G Γ INC Z0 = 1 Γ Γ G Γ G Γ phase(γ ) phase(γ ) INC Γ G Γ phase(γ ) phase(γ ) METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 25

26 Absolute value of measured RF voltage level and its uncertainty at frequency f test : What assumes the user of a calibrated scope? calibrated frequency response f ref frequency response between f ref and f test f test measured voltage level at frequency f test and its uncertainty = a) value of absolute voltage level at f ref and its uncertainty b) + frequency response at f test and its uncertainty c) + uncertainty of applied input voltage f METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 26

27 calibration of oscilloscopes with 50Ω input it is common practice to calibrate RF-voltage referenced to INC in this case, even if user takes into account the data of certificate of calibration, he measures rf voltage with an additional uncertainty of about ± 100 * Γ Oscilloscope [%] example: typical input reflection coeff. Γ Osc 0.15 (SWR = 1.35) rf voltage, additional uncertainty ± 15 % METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 27

28 Summary: calibration of RF-voltage on oscilloscopes an oscilloscope is a widely used and reliable tool in this paper was not the idea to invent new parameters today s oscilloscopes: increasing bandwidth even with 1MΩ input oscilloscope in metrology? (voltage at reference plane = DUT input) stongly depends on Γ (=Γ DUT ) high uncertainty (... 20%) may result if Γ (mag. and phase) not measured take care with specs of oscilloscope calibrators the exact conditions must be specified (referenced to or INC ) does exist a reference document on this topic? does anybody has more information? update of EA10/7 (calibration of oscilloscopes)? METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 28

29 Thank you for your attention METAS J. Furrer, RF- Power aboratory ANAMET 23th Meeting 3. / 4. March 2005 slide 29