Direct comparison technique - CMIrfmw.cmi.cz/documents/meetings/directcompar/02... · 2016, METAS e.g., R&S NRPCxx 29 Practical considerations. EMPIR 15RPT01 workshop, 7. 11. 2016,

Direct comparison technique

Martin Hudlička, CMI

Murat Celep, UME

Contact: [email protected]

EMPIR 15RPT01 workshop, 7. 11. 2016, METAS

mailto:[email protected]

• Introduction

• Direct comparison method

• Practical considerations

• Conclusion

Contents


• power sensor calibration

linearity of the sensor

– single point adjustment at zero input (sensor zero) – uses stored data to characterize detector gain and linearity (no RF reference used)

– two point adjustment – zero input and mid- or full-scale RF power reference (calibration for gain and offset, factory linearity data used for transfer function shape)

– multi-point adjustment – series of RF power values (can replace or enhance the stored factory data)

Introduction

EMPIR 15RPT01 workshop, 7. 11. 2016, METAS3

• power sensor calibration

frequency response

– power sensor calibrated at series of frequency points to generate a table of correction values

– effective efficiency or a calibration factor

Introduction


Introduction


Rohde&Schwarz:Voltage and PowerMeasurements,app. note, 1999

5

• calibration factor used to describe both the eff. efficiency and the mismatch

• used for a calibration transfer from a reference standard to an unknown power standard

• traceability: unbroken chain of comparisons from the standards of lower level to the primary standard

• calibration factor is typically frequency dependent and can be stored in an EEPROM memory of the sensor after calibration (together with temperature effect)

• power sensor and power meter calibrated separately (either can be interchanged without invalidating the calibration)

Introduction

𝐶𝐹 = 𝜂𝑒 1 − |𝛤|2


• calibration of RF sensor is transferring the eff. efficiency or cal. factor from

– a primary standard to a secondary standard (NMI -> NMI)

– a secondary standard to a reference standard (NMI -> secondary lab.)

– a reference standard to a power sensor (secondary lab. -> industry)

• parameter transfer– through comparison

– calibration one against the other

Introduction

signal

generator

reference power

standard

GG

power sensor to

be calibrated

GDUT

Gstd


• general mismatch between generator and load

Introduction

generator sensorpower

meter

Pincident

Preflected PRF

Pdissipated

PDC substitution

a

b

𝑃𝑖 =|𝑎|2

𝑍0𝑃𝑟 =

|𝑏|2

𝑍0𝛤 =

|𝑏|

|𝑎|

𝑃𝑖 = 𝑃𝑍01

|1 − 𝛤𝐺𝛤𝐿|2 𝑃𝑟 = 𝑃𝑍0

𝛤𝐿2

|1 − 𝛤𝐺𝛤𝐿|2

various sensor losses


GL GG

bL

aL

b

a bs 1

1

8

• Introduction



• Conclusion

Contents


• direct comparison method– very old technique, used since 50’s :

R. W. Beatty, A. C. Macpherson: Mismatch Errors in Microwave Power Measurement, Proc. of IRE, Vol. 41, No. 9, pp. 1112-1119, Sept. 1953

G. F. Engen: Amplitude Stabilization of a Microwave Signal Source, IRE Trans. Microwave Theory Techn., Vol. 6, No. 2, pp. 202-206, Apr. 1958

R. F. Desch, R. E. Larson: Bolometric Microwave Power Calibration Techniques at the National Bureau of Standards, IEEE Trans. Instrum. Meas., Vol. 12, No. 1, pp. 29-33, Jun. 1963

Direct comparison method


• power dissipated to the reference and DUT sensors


signal

generator

reference power

standard

GG

power sensor to

be calibrated

GDUT

Gstd

𝑃𝑠𝑡𝑑 = 𝑃𝑖 − 𝑃𝑟,𝑠𝑡𝑑 = 𝑃𝑍01 − |𝛤𝑠𝑡𝑑

2 |

|1 − 𝛤𝐺𝛤𝑠𝑡𝑑|2

𝑃𝐷𝑈𝑇 = 𝑃𝑖 − 𝑃𝑟,𝐷𝑈𝑇 = 𝑃𝑍01 − |𝛤𝐷𝑈𝑇

2 |

|1 − 𝛤𝐺𝛤𝐷𝑈𝑇|2


• power dissipated to the reference and DUT sensors

• if Gstd = GDUT, exactly the same amount of power is dissipated in both sensors (ideal case)

• available source power:


signal

generator

reference power

standard

GG

power sensor to

be calibrated

GDUT

Gstd


𝑃 = 𝑃𝑍01 − 𝛤𝐷𝑈𝑇

2


12

• generator reflection coefficient 𝛤𝐺 = 0

only magnitudes of 𝛤𝐷𝑈𝑇 and 𝛤𝑠𝑡𝑑 are required, otherwise all three magnitudes and phases required

• phase angles of reflection coefficients unknown (scalar measurement)




2

1 ± 𝛤𝐺 𝛤𝐷𝑈𝑇2

13

• transfer of the effective efficiency hstd from the reference standard to the effective efficiency hDUT of the DUT power sensor:


𝜂𝐷𝑈𝑇𝜂𝑠𝑡𝑑

=

𝑃𝐷𝐶,𝐷𝑈𝑇𝑃𝑅𝐹,𝐷𝑈𝑇𝑃𝐷𝐶,𝑠𝑡𝑑𝑃𝑅𝐹,𝑠𝑡𝑑

𝜂𝐷𝑈𝑇 = 𝜂𝑠𝑡𝑑𝑃𝐷𝐶,𝐷𝑈𝑇𝑃𝐷𝐶,𝑠𝑡𝑑

1 − |𝛤𝑠𝑡𝑑|2 |1 − 𝛤𝐺𝛤𝐷𝑈𝑇|

2

1 − |𝛤𝐷𝑈𝑇|2 |1 − 𝛤𝐺𝛤𝑠𝑡𝑑|

2


• similarly, transfer of the calibration factor CFstd from the reference standard to the calibration factor CFDUT of the DUTpower sensor:


𝐶𝐹𝐷𝑈𝑇 = 𝐶𝐹𝑠𝑡𝑑𝑃𝐷𝐶,𝐷𝑈𝑇𝑃𝐷𝐶,𝑠𝑡𝑑


|1 − 𝛤𝐺𝛤𝑠𝑡𝑑|2


impedance mismatch between the source and power sensor(s) is the main uncertainty source

15

• accuracy improvement: inserting passive components improves source mismatch effect and the DUT mismatch effect

– attenuator – improves mismatch, but decreases dynamic range

– power splitter – holds effective source output power constant

• reflection coefficient of a generator difficult to measure (varies with time, many old signal generators and amplifiers have non-linear output impedance)

equations derived using linear network theory do not work




2


16

• solution 1: levelled source

• a feedback circuit maintains the output from the detector at a constant predetermined level



signal

generator

to load

Z0

Z0

amplifier

1

2

3

offset

detector

17

• solution 1: levelled source

• suitable for unstable sources, generator characteristics eliminated from the measurement



• reduction in amplitude noise• correcting for slow drifts

• noisy detector in the levelling loop may worse theS/N ratio (amplitude modulation of the generator)

• frequency of the source must be stable (frequency dependence of a coupler/splitter)

18

• solution 2: monitoring the power level

• correction of the level at the DUT arm for the power measured at the STD arm

• use of normalized powers (dividing the reading of power meters to be compared by the reading of the monitoring meter)



signal

generator

monitoring power

sensor + meter

Z0

Z0

1

2

3

to load

19

• solution 2: monitoring the power level

• normalized powers = readings which would be obtained if the source were levelled to maintain the monitoring power meter reading constant and equal to 1

• levelling and monitoring methods are equivalent if the power sensors are linear



does not eliminate effects of rapid changes in signal level or noise (monitoring and measured power meter do not necessarily respond in the same way)

20

• best solution: combination of monitoring and levelling (non-ideal feedback loop is corrected by monitoring), port 2 of the power splitter becomes the effective source output


signal

generator reference power

standard

GEG

power sensor to

be calibrated

GDUT

Gstd

monitoring power

sensor + meter

Z0

Z0

level

control

P3std when ref. sensor connected

P3DUT when DUT sensor connected

1

2

3


• resistive power splitter S-parameters (Z0 = 50 W)


1

2

3Z0

Z0

𝑆 =0 0.5 0.50.5 0.25 0.250.5 0.25 0.25


• power splitter’s effective source match

𝛤𝐸𝐺 = 𝑆22 − 𝑆21𝑆32𝑆31

• resistive power splitters



• broadband (DC to GHz)• better impedance match than coaxial couplers

• temperature dependent properties• 6 dB loss (high dissipated power in

high-power measurements)• errors due to splitter asymmetry

• calculating 𝛤𝐸𝐺 directly from S parameters sensitive to small measurement errors, alternative measurements proposed:


J. R. Juroshek: NIST 0.05-50 GHz direct comparison power calibration

system, Conference on Precision Electromagnetic Measurements

(CPEM2000), pp. 166-167, 14-19 May 2000, Sydney, Australia

K. Yhland, J. Stenarson: Measurement Uncertainty in Power Splitter

Effective Source Match, IEEE Trans. on Instrum. Meas., pp. 669-672, Vol.

56, No. 2, April 2007


K. Shimaoka: A new method for measuring accurate equivalent source

reflection coefficient of three-port devices, Conference on Precision

Electromagnetic Measurements (CPEM2010), pp. 589-590, 13-18 June

2010, Daejeon, Korea

24

• instead of the power splitter, one can use other types of coupling devices (large powers, waveguides, ...)



• for the equivalent source match calculation, one needs a complete s-matrix of the coupling device

• if difficult to measure – approximation can be used if only specified (datasheet) magnitude parameters are known or if only a scalar measurement is possible



𝛤𝐸𝐺 = 𝑆22 − 𝑆21𝑆32𝑆31

𝛤𝐸𝐺 ≤ 𝑆22 +𝑆21𝑆32𝑆31

26

• Introduction



• Conclusion

Contents


monitoring/levelling power meter

• in theory may be of any type (bolometric, thermoelectric, diode)

• in practice, bolometric meters are the most suitable– mostly used is the thermistor principle

– they are RF/DC substitution devices

– relative stable characteristics

– substituted DC power applied without the need to disconnect the power meter from the coupler

Practical considerations


e.g., HP 8478B (coaxial)or HP R486A (waveg.)

28

commercial solutions

• highly linear thermistor power sensor + 2R power splitter in one block, calibration of customer’s power sensors + update of the data in EEPROM memory, traceable to PTB

• other vendors have similar solutions (similarly expensive )


e.g., R&S NRPCxx

29



• vector corrections etc. are fine, BUT bad connector repeatability can spoil all the effort

• research conducted by NPL & Agilent in 2001, power sensors and power splitters with different connectors evaluated

– old precision connector

– new precision connector (hex nut)

– metrology grade connector

• calibration factor may change significantly with bad connectors

• need to measure connector repeatability for every connector pair forming part of a measurement system






• Introduction



• Conclusion

Contents



• traceability transfer from standard sensor (calibration factor determined in a calorimeter)

• combination of monitoring and levelling improves the system stability

• vector corrections should be applied, provided a full vector information about the coupling element is known

• worn-out sensor connectors could introduce measurement error, measurement for more positions of the connector ( 3) recommended

Conclusion

Thank you for attention


EMPIR project “15RPT01 Development of RF and microwave metrology capability”http://rfmw.cmi.cz/

http://rfmw.cmi.cz/

Documents

Direct comparison technique - CMIrfmw.cmi.cz/documents/meetings/directcompar/02... · 2016, METAS e.g., R&S NRPCxx 29 Practical considerations. EMPIR 15RPT01 workshop, 7. 11. 2016,