CTF3 Low-Level RF measurements TBTS/TBS lines and 12 GHz testand (Xbox 1) Stéphane Rey* - BE/RF Luca Timeo - BE/RF Ben Woolley - BE/RF Wilfrid Farabolini - CEA October 7th, 2013 - Rev 1.0

Calibrating RF measurement paths Why ? To ensure each device is rated for the frequency and power applied To get best measurement accuracy of RF powers To measure RF path characteristics (insertion & return losses, linearity, reflections, …) To check and compensate long term drifts To provide RF systems diagnosis How ? Several methods : Method 1 : Generator + powermeter  S21 Method 2 : Generator + 4 powermeters  S11, S21, S22, S12 Method 3 : VNA 1-port  S11, Method 4 : VNA 2-ports  S11, S21, S22, S12 Method 5 : TDR (Time Domain Reflectometry) using a VNA  S11 + post-processing (inverted FFT x virtual pulse)  S11, S21, S22 Method 6 : TDR using a Time Domain Reflectometer  S11, S21, S22 Method 7 : TWT + CTF3 Acquisition system and powermeters  S21, S11, S22

Calibrating RF measurement paths Typical RF path at CTF3 Semi-rigid cables Diodes ADC Coaxial attenuators IQ N to N transition Klystrons Gallery CLEX 15-35m 3/8’’ coaxial cable waveguide to N adapters waveguide ~10 dB attenuators 50 dB coupler

(or Murphy law variation) The true life (or Murphy law variation) Different measurement methods gives different results Measurement instruments are less accurate than the expected measured accuracy Devices are never specified correctly… if they are specified ! … and are anyway never used within their nominal specified range RF Powers are always above expectations ‘Low-phase’ coaxial cables are phase dependent Fixed attenuators are not A 5 watts rated equipment exhibits a derating starting from 1 watt and burns at 4 watts Symmetric devices are not ‘Perfectly matched’ devices have a return loss > -10 dB … which fluctuates over time Controlled temperature means ‘almost controlled’ Screwing a connector again with the same torque never gives the same return loss ‘Do not touch’ warning signs are usually inviting people to touch

Instruments accuracy (or uncertainty) Multimeter Agilent 34401A – 6½ digits (source 34401A datasheet) Powermeter Agilent E4416 (source E4416 datasheet) Absolute accuracy = ± 0.04 dB within ± 5°C from its calibration temperature Absolute accuracy = ± 0.08 dB from 0 to 55°C

Instruments accuracy (or uncertainty) 20 GHz VNA Agilent E5071C 6-20 GHz corrected system performances with 3.5mm 85052D SMA calkit (source E5071C datasheet) Transmission uncertainty for transmission coef. between +10 and -40 dB = ± 0.2 dB @ 23°C ± 5°C with < 1°C deviation from calibration temperature Phase uncertainty for transmission coef. between +10 and -40 dBm = ± 2 degrees Magnitude temperature drift = ± 0.04 dB/°C Phase temperature drift = ± 0.8 degrees/°C

Instruments accuracy (or uncertainty) (source DC270 datasheet) Acquiris DC270 digitizer : 8-bits / 1 GSPS / 250 MHz (source DC270 datasheet) Accuracy ±2% Full scale range Linearity ±1% Full scale range ENOB (number of effective bits) : DC-20 MHz : > 6.8 20-100 MHz : > 6.4 Instruments last calibration ? Calkits verification ? Probes checking ?

What are uncertainties ? Uncertainties are not only offsets ! They are made of : Frequency dependence Gain & Offset errors Linearity, Hysteresis errors Gain and offsets drifts over temperature Noises (thermal, flicker, …) No easy calibration to implement ! Global error estimation (Gaussian distr.) :

Distinguishe DUT from instruments errors 6-sigma method to determine equipment capability 6-sigma : methodology from Motorola (80s) to increase quality and process efficiency by measuring equipment capability (accuracy & repeatability) Statistic parameters , CP and CPk characterize the error distribution Estimate the number of parts out of specified limits Lpl + Lpu First, characterize the measurement system and then the population

Distinguishe DUT from instruments errors 6-sigma method to determine equipment capability Non-conformities vs CP, CPk

Generator + powermeter(s) Measurement method 1 Generator + powermeter(s) ~ ~ Power meter Power meter Step 1 Step 2 Measure S21 directly Pros : Easy to do even without access to the two hands of the cable Direct measurement of DUT without extra instruments (couplers, splitters, …) Cons : Measure only S21 Measurement error if there are bad return losses Limited possible path attenuation to 40-50 dB due to generator power output limit (usually +10 dBm) and powermeter noise floor (~ -40 dBm)

Generator + couplers + load + powermeters Measurement method 2 Generator + couplers + load + powermeters Power meters Power meters ~ ~ Power meter Step 1 Step 2 Measure all parameters Pros : Measure all S-parameters : S11, S21, S22, S12 Cons : Longer calibration process : calibration of couplers with VNA (uncertainties) Needs several instruments Add extra equipment in the measurement path (source of uncertainties) Limited measurable insertion losses to 40-50 dB due to generator power output limit (~ +10 dBm) and powermeter noise floor (~ -40 dBm)

Measurement method 3 VNA 1-port VNA Measure S11 Pros : Cons : Measure only S11 S22 or any intermediate return loss effects seen might be seen on S11 With high attenuation paths, not possible to see a bad output return loss VNA uncertainties

Measurement method 4 VNA 2-ports VNA Measure all S-parameters Pros : Cons : Impossible if no access to the two hands of the DUT With high attenutation paths, not possible to see an intermediate bad return loss (damaged cable) VNA uncertainties

TDR using VNA 1-port + post-processing Measurement method 5 TDR using VNA 1-port + post-processing VNA + Measure S11 Pros : Enable to see intermediate return losses and identify damaged cables Enable to measure distances (or times) Cons : Accurate only if return losses are negligible No real time measurement : very slow process. Two measurements needed (loaded and shorted) With 30~40 dB attenuation paths, not possible to measure a bad output return loss VNA uncertainties

Measurement method 6 TDR TDR Measure S11 Pros : Cons : Enable to see intermediate return losses and identify damaged cables Real-time Higher dynamic range, resolution and accuracy than VNA 1-port method Enable to measure distances (or times) Cons : S22 or any intermediate return loss effects seen might be seen on S11 With high attenuation paths, not possible to see a bad output return loss VNA uncertainties

TWT+ couplers + powermeters Measurement method 7 TWT+ couplers + powermeters Power meters Power meters TWT Measure all parameters Pros : Measure all S-parameters : S11, S21, S22 Measure the actual behavior with RF power as in application Cons : Longer calibration process : Needs several instruments Add extra equipment in the measurement path (source of uncertainties) TWT output impedance is varying vs RF power making S11 measurement uncertain

Measurement methods Synthesis All the methods are providing different results. The choice of the method depends on what exactly need to be measured. For S21 the generator + powermeter gives the best accuracy but this require to check first with a VNA or TDR that return losses are correct For S11 measurement, the VNA is easier to use but provide a less accurate measurement. But is there really a need for accuracy in RL measurement ? For time measurement as well as to check quickly quality of RF Path, TDR is the most suitable. In order to fully characterize correctly a RF path it might be necessary to combine several methods.

CTF3 measurement results Layout Diodes / Log ADC IQ Coaxial S11 ~7-15 dB Klystrons Gallery CLEX Coaxial S11 ~7-15 dB Attenuator S22 ~10dB Coupler S34 ~0dB Attenuator S11 ~20dB Coupler S14 ~50dB Uncertainty ?

CTF3 measurement results Coupler + attenuator + coaxial cable Calibrated Not Calibrated WR90-N Adapter S21=~0dB = 98W + Attenuator S22=-10dB (10% = 0.3-1.96W) - Coax S11=7-15dB (3-20%=3-19.6W) = 78.4…95W =0.3..1.96MW =3..19.6MW =98MW Attenuator S11 + WR90-N S11 + Coax S11 - Attenuator S22 =3.8..18.6 MW = 3.8 to 18.6 % Attenuator S21=-10dB = 99,0W - WR90-N adapter S11=-20dB (1%=0.99W) = 98W =0.1MW =99MW Coupler S31=-50dB = 1 KW - Attenuator S11=-20dB (1%=10W) = 990W =1MW =100MW RF power = 100 MW

CTF3 measurement results Coupler + attenuator + coaxial cable Possible solution :

CTF3 measurement results IQ Crates BW = ~150 MHz !

CTF3 measurement results Diode Crates

CTF3 measurement results ½ Diode crate chanels cross-coupling 3dB 3dB CAL in CAL out 6dB 6dB Group similar amplitudes on same 6dB splitter To channels 5-8 3dB #1 3dB 3dB #2 3dB RFin 3dB #3 3dB 3dB #4 3dB

CTF3 measurement results XBOX 1 calibrations – method 7 with TWT

CTF3 measurement results TBTS calibrations – method 7 + method 4 S11 varies with power : TWT output impedance or RF path drift

CTF3 measurement results Temperature logging MKS 03 temperature :

CTF3 measurement results Conclusions Layout modifications for detectors dynamic optimization & general improvement Full RF paths characterizations with several methods giving different results are needed instruments & layout mismatches leads to uncertainties Next ? Methods validations in lab on reference cable Continue equipment characterizations (automatic tests bench, climate room) : Replace our detectors technology Current designs : Down-mixing detector (design for XBOX2) New log detectors crate Investigate on direct acquisitions closer to the machine Investigate on RF & clock distribution quality (phase noise, spurs)