Novel Techniques of RF High Power Measurement

Transcription

1 Novel Technques o RF Hgh Power Measurement Ovdu D. Stan Department o Electrcal and Computer Engneerng COLORADO STATE UNIVERSITY EE PhD Dssertaton Deense 2007

2 Why s RF Hgh Power Measurement s Important? Semconductor and lat panel ndustry requres RF Power level up to 50kW wth better then 1% power accuracy, or requences 1-200MHz Process control Repeatablty Faster dagnostcs Increased yeld Reduced development tme Exstng RF power methods are typcally 1-3% accurate nto real mpedance (50 Ohms) Accuracy s decreasng wth VSWR We lack a natonal (e.g. NIST) or nternatonal RF hgh power standard RF hgh power calbraton s perormed usng only ndrect methods

3 RF Hgh Power Measurement- Topcs Background on RF Hgh Power Measurement New Solutons or Real Impedance Lnes and Loads Double dode detector wth dgtal correcton Multpler wth dgtal correcton New Solutons or Complex Impedance Lnes and Loads Drect dgtal samplng Calbraton Methods Summary and Future work

4 Real Impedance RF Hgh Power Measurement Applcaton Advantages: Cost eectve Impedance o the lne s xed Z50+j0 tan(φ)0 Dsadvantages: Lmted plasma process normaton Error o exstng power measurement methods s 1%-3% Proposed methods have 3X better accuracy

5 Complex Impedance RF Hgh Power Measurement Applcaton Advantages: Better plasma process control Feedback dagnostcs Dsadvantages: Expensve Impedance o the lne s complex and varable ZR+jX tan(φ)x/r Error o exstng methods or power measurement s >1% and >1.5% or mpedance measurements New method has 3X better accuracy or power and mpedance

6 Hgh Power RF Power Measurement - Orgn o Errors There are 4 categores o errors n RF power measurement: RF level lnearty (Power/Impedance vs. RF Sgnal Level) Frequency error (Power/Impedance vs. Frequency) VSWR error (Power/Impedance vs. VSWR) Envronmental Stablty ( Accuracy vs. Humdty and Temperature) All above errors compound nto the total error o the RF measurement method The RF components have parastc characterstcs and stablty ssues, thereore any accurate RF measurement requres ether deal components or a proper correcton (analog and/or dgtal)

7 Anatomy o the typcal RF Hgh Power Measurement Advantages: Mature technology V/I Sensor Drectonal Coupler RF Sensor Harmonc Rejecton Heterodyne Freq. Selecton Amply Flter Multply V rms Analog Crcut Compute Calbrate Data nterace Dgtal Crcut Dsadvantages: Less Accurate. Accuracy s best nto 50Ohm and s degradng wth VSWR Power Error >1% Impedance Error >1.5% Complex Accuracy s changng wth temperature and humdty Typcal RF Hgh Power Instrument s analog whle the new technque s dgtal

8 Proposed Enhanced Dgtal RF Measurement Method Advantages: Better Accuracy Power Error 0.3% Impedance Error 0.3% Smaller sze ( 50%) Extra eatures Frequency agle (2-64Mhz) Plasma process eedback Envronmentally stable Dsadvantages: Cost o Development V/I Sensor or Drectonal Coupler RF Sensor Flter Analog Crcut Correct Compute Freq. Selecton Data Com. Dgtal Crcut

9 RF Measurement Method or Real Impedances Background P load V*I*cos(φ) For Real Impedance Loads and Lnes cos(φ)1 thereore Power Measurement s smpled to V rms measurement: P load V rms2 /Z 0 or I rms measurement: P load Z 0 I rms 2 However, on RF we use Forward and Relected Power as ollows: P load P wd -P rl wth the ollowng equatons: P wd V wd2 /Z 0 P rl V rl2 /Z 0 Error o exstng power measurement methods s 1%-3% Proposed methods have 3X better accuracy

10 Illustratve RF Peak Dode Detector Crcut Hstorcally the most basc desgn or RF voltage measurement Advantages: Smple Wde-band Dsadvantages Non-lnear Does not work at low level sgnals

11 Improved Double Dode Detector Crcut Advantages: Smple Wde-band Works at low level sgnals Dsadvantages Non-lnear at hgh sgnal level

12 Calculatons Requred or Calbratng a Double Dode Detector Drawback: RF Power calculaton requres a second degree polynomal correcton Power vs. Voutput- Double Dode Detector Desgn y x x R 2 1 Power (W) Vout (V) Analog multpler soluton s smple to correct dgtally

13 Calculaton Requred or Multpler Implementaton o Power Measurement Advantage: Power vs. Voutput- 50Ohm Multpler Desgn Multpler wll provde the voltage squared RF Power measurement s lnear and requres only scalng and oset correcton Standard Power (W) y x R Vout (V) Besdes power scalng there are urther correcton requrements

14 P RF Measurements Other Correcton Requrements:Thermal and Bandwdth Temperature Stablty Requrement: Tests done n a envronmental Chamber conrmed drt o analog components Wdeband Requrement: For nstrumentaton wth passve nput lters a correcton o ampltude wth requency s mandatory Both requrements along wth RF Power correcton can be solved by DSP

15 Illustratve Dgtal Correcton Get VF&VR N10? No Subtract Oset Yes Get VFoset& VRoset N? Oset? Correcton Factors? Measurement Results? Dgtal Correcton wll enable: Correcton o measurement wth temperature Correcton o measurement wth requency bandwdth Correcton o measurement wth ampltude o nput sgnal Seral Interace Runnng Averagng Correcton Calc. Other Dgtal Functons: Seral communcatons Averagng o the sgnal All the dgtal correctons are mprovng the RF Power accuracy

16 Expermental Error Analyss or Multpler and Double Dode Detector Standard (W) Vout (V) Oset (V) Slope (W/V) Standard (W) Vout(V) Oset (V) Slope (W/V) Analog Multpler results RF Power equaton: x A/ D _ Double Dode Recter results RF Power equaton: readng P Oset + x * Slope P A* x 2 + B* x + Oset Oset W Slope W/V Max Error0.05% Oset 7.293W A B Max Error0.22%

17 Accuracy Results or the Analog Multpler wth Dgtal Correcton Calbraton results over sx devces that employed a multpler wth a dgtal correcton delvered: RF Power Error < 0.20% or W sgnal level RF Power Error < 0.30% wthn 15 ο C-45 ο C Results were compared to a Reerence Standard calbrated on a calormeter In concluson

18 Far more challengng are RF measurements nto complex mpedances Conclusons or Real Impedance Measurement Methods Two mproved methods were developed or RF hgh power measurement 1) Double dode detector technque wth dgtal correcton 2) Multpler technque wth dgtal correcton Both desgns employ a dgtal correcton or requency bandwdth and temperature varatons Both desgns acheved a power error < 0.3% versus the exstng state o the art that has a typcal error 1-3%

19 RF Measurement or Complex Impedances Bascs P load V*I*cos(φ) For measurements nto complex mpedance loads and lnes the most dcult element to measure (and major source o errors) s: cos(φ) Advantages: Better plasma process control Feedback dagnostcs Dsadvantages: Expensve

20 New Method or Measurng Complex Impedances Advantages: Better Accuracy Power Error 0.3% Impedance Error 0.3% Smaller sze ( 50%) Extra eatures Frequency agle (2-64Mhz) Plasma process eedback Envronmentally stable V/I Sensor or Drectonal Coupler RF Sensor Flter Analog Crcut Correct Compute Freq. Selecton Data Com. Dgtal Crcut Dsadvantages: Cost o development

21 New Method or Measurng Complex Impedances Drect Dgtal Samplng Analog crcut topology conssts o: Balun Transormer Low pass lter Forward and relected channels are smlar and parallel sampled

22 Drect Dgtal Samplng Dgtal Sgnal Processng Schematc e Samplng req. w Wndow req. Sn / Cos s Input sgnal requency Correcton Matrx (4x4) 2 2 c V V _ Q + V _ I 2 c V V V _I S V _I V _I c V V _Q S V _Q V _Q c V r ndex o the sample X Calculus Module R V r _I S V r _I V r _I c X V r V r V r _Q S V _Q V _Q c P Samplng Hann Wndow Fourer Transorm Calbraton RF Calculus

23 Drect Dgtal Samplng Ampltude and Phase Processng I We know the requency s V Asn( ω t + ϕ ) S Dgtal Samplng V Asn( ω t + ϕ ) S *sn (ω s t ) *cos(ω s t ) V V _ I _ Q Asn( ω t S Asn( ω t S + ϕ )*snω t + ϕ )*cosω t S S Trgonometrc transorm V V _ I _ Q 0.5A[cosϕ 0.5A[snϕ cos(2ω t S + sn(2ω t S + ϕ )] + ϕ )] Both I and Q components are summed over the observaton wndow

24 Drect Dgtal Samplng Ampltude and Phase Processng II _ 1 1 _ N N Q V N Q V I V N I V )] sn(2 [sn 0.5 _ )] cos(2 [cos 0.5 _ S S t A Q V t A I V ϕ ω ϕ ϕ ω ϕ ) sn(2 2 sn 2 _ ) cos(2 2 cos 2 _ N N S N N S t N A N A Q V t N A N A I V ϕ ω ϕ ϕ ω ϕ A Q V A I V ϕ ϕ sn 2 _ cos 2 _ Ater DFT we stll have the sgnal Ampltude and Phase Inormaton These Sum Terms are Zero because: a) Observaton Perod s large compared to the RF sgnal requency: w << s N s the number o samples acqured durng the Hann Wndow b) Hann uncton s gong to attenuate the begnnng and end dscontnutes Next, both I and Q components are corrected by a calbraton matrx

25 Hann Wndow Eect In ths example the Hann wndow has 1µSec perod, the nput sgnal has 13.56MHz and the samplng rate s 100Ms/Sec 2π w( ) cos( ) N 0,1,...N 1.5 Hann Wndow Eect Input Sgnal Hann Wndow (Input Sgnal) * (Hann WIndow) Ampltude Sample n

26 Four Channel Dgtal Calculaton o Correcton or Drect Dgtal Samplng K s a scalng actor or ampltude, does not aect phase V _I c V _Q c V r _I c V r _Q c K * a11 a12 a13 a14 a21 a22 a23 a24 a31 a32 a33 a34 a41 a42 a43 a44 1_ 0 _ 0 _ 0 0 _1_ 0 _ 0 0 _ 0 _1_ 0 0 _ 0 _ 0 _1 * V _I V _Q V r _I V r _Q aj,j1 to 4 are calbraton actors or phase correcton V _Q c V r _Q c V ϕ c V _I c V r ϕ rc V r _I c In deal case (no phase dstorton) the correcton s the dentty matrx Corrected values or I and Q wll provde V, V r and ϕ

27 Dgtally Corrected Ampltude and Phase or Drect Dgtal Samplng The corrected values or I and Q are provdng real readngs o the nput sgnal V _Q c V 2 2 c V V _ Q + V ϕ c V r Vr _ Qc + Vr _ Ic _ I 2 c V _I c V r _Q c V r ϕ rc V r _I c ϕ ϕ c rc tan 1 tan 1 V V V V c c rc rc _ Q _ I _ Q _ I ϕ ϕ c ϕ rc

28 Determnaton o Calbraton Matrx or Complex Impedance Measurements I FWD RFL X Input Block Calbraton Factors A K * V _I V _Q V r _I V r _Q I we know the Matrx Y and the Matrx X (as reported by the RF Measurement) n at least 16 cases then we can calculate calbraton Matrx A Y a11 a12 a13 a14 a21 a22 a23 a24 a31 a32 a33 a34 a41 a42 a43 a44 V _I c V _Q c V r _I c V r _Q c Y K* A * X How to determne the 16 elements o matrx A?

29 Determnaton o Calbraton Matrx or Complex Impedance Measurements II DUT reports Matrx X or every sngle load Changng the settngs o the Varable Match to 113 derent loads wll generate 113 equatons: Y K* A * X Determnng Calbraton Matrx A reduces to solvng an over-speced system wth 113 equatons and 16 unknowns The RF match s calbrated (known mpedance) nto 113 loads We Measure the Power Level We know the RF sgnal phase (rom the load) and we measure the RF Power, thereore we know Matrx Y

30 Test Results or Drect Dgtal Samplng nto Complex Impedances Test bench setup Red dot Target mpedance, measured wth a Impedance Analyzer Green Crcle Impedance Measured by the new measurement system 1) Impedance Error <1% or loads up to VSWR5 2) 6 derent DUT s had an Error<0.1% nto a xed load Z17+j2.7

31 Drect Dgtal Samplng. Power Measurements_Test Results n 50Ω RF Power (W) DUT Power (W) Reerence Standard Power (W) Error UM W W Power Error < 0.71W compared to a calbrated Reerence Standard Power Error < 0.34%* compared to a calbrated Reerence Standard * 0.15% out o the Power Error was dented as systemc calbraton ssue related to the match losses measurements W W % % % % % % %

32 Drect Dgtal Samplng Conclusons New RF Hgh Power Measurement Technque wth: Better Accuracy Power Error 0.3% vs. typcal 1% o exstng methods Impedance Error 0.3% vs. typcal 1.5% o exstng methods Accuracy s consstent over large VSWR Smaller sze ( 50%) Extra eatures Frequency agle (2-64Mhz) Plasma process eedback Envronmentally stable Drectonal Coupler RF Sensor Flter Analog Crcut - Correct - Compute -Freq. Selecton - Data Com. Dgtal Crcut RS232 An accurate measurement method should be complemented by a good reerence

33 Calbraton Technques or Hgh Power RF. Overvew RF Power Hgh Power (>100W) does not have a NIST, nor any other Internatonal Standard There are only Indrect Methods to calbrate RF Power Instruments, typcally usng substtuton methods PV*I*cos(φ) In RF, Voltage level s not a good method o Power Measurement

34 Calbraton Technques or Hgh Power RF. Wet Calormeter RF #2 Error: I DC measurement RF Standard to be Calbrated I DC Wet Calormeter compares the temperature ncrease n the same load by an RF source and a DC source. H 2 O Pump DC H 2 0 T IN #1 Error: T T out -T n measurement V DC H 2 0 T OUT When the temperatures are equal we conclude that the DC powers s equal to RF delvered power

35 Calbraton Technques or Hgh Power RF. New Technque usng Dry Calormeter I Dry Calormeter compares the RF power level between two nstruments, one as a reerence and the other one unknown, usng a calbrated Drectonal Coupler P K 10*log P 3 2 Calbrated Power Reerence A 20dB coupler would have 100:1 Power Rato (K100) Drectonal Coupler wth a calbrated power couplng coecent, K K P P 2 3

36 Novel Technques o RF Hgh Power Measurement. Summary It s more practcal to employ the proper dgtal correcton technque then to research deal components. 1) Two Improved Technques o Power Measurement or Real Impedances were presented Multpler ollowed by dgtal correcton (3X accuracy mprovement) Double dode detector ollowed by dgtal correcton 2) A New RF Measurement Technque or Complex Impedances was presented Drect dgtal samplng (3X mprovement) A new calbraton method or complex mpedance nstruments Results on all the above methods have lower errors then all prevous methods 3) Research results nto RF hgh power calbraton methods Wet Calormeter Error Analyss New Dry Calormeter Method

37 Novel Technques o RF Hgh Power Measurement. Future Work It s Impractcal to Calbrate all RF Instruments on the RF Calormeter; Transer Standards are employed DC Voltage DC Current NIST Traceable RF Calormeter DC to RF Cal O Transer Standard RF Instrument Calbraton RF Calormeter can mprove the absolute accuracy o the measurement 1) Reduce Calbraton Tme on the RF Calormeter by usng extrapolaton technques

38 Novel Technques o RF Hgh Power Measurement. Future Work It s mpractcal to calbrate all RF Instruments on the RF Calormeter; Transer Standards are employed DC Voltage DC Current NIST Traceable RF Calormeter DC to RF Cal O Transer Standard RF Instrument Calbraton 1) Reduce calbraton tme on the RF Calormeter by usng extrapolaton technques 2) Increase relablty and the RF Power lmt on the RF Calormeter In ths moment maxmum RF calbrated power s 3500W 3) Research the lmtatons o the Drect Dgtal Samplng Method

39 Novel Technques o RF Hgh Power Measurement Questons?