Use of on-chip sampling sensor to ealuate conducted RF disturbances propagated inside an integrated circuit M. Deobarro 1, 2 (PhD-2) B. Vrignon 1, S. Ben Dhia 2, A. Boyer 2 1 Freescale Semiconductor 2 LATTIS (INSA Toulouse) mikael.deobarro@freescale.com 19/11/2009
Oeriew I. Issues and goals II. Description of the test ehicle III. On-chip oltage sensor IV. Characterization of sensor V. Experimentation VI. DPI test VII. Conclusion 1
I. Issues and Goals Issues: External measurements do not gie accurate characterization of RF interference (RFI) carried on die Coupling path depends on frequency (PCB, package & IC design) and influences the RFI injected inside an integrated circuit RFI Tip of probe could add parasitic capacitance which influences the impedance of component under test Lack of information about disturbance propagation through an integrated circuit V RFI? Objecties: Improe our knowledge about RFI propagation inside IC Characterize conducted RF disturbances injected on a test ehicle by using on-chip sampling 2
II. Description of test ehicle «Mixed Immunity» test chip 0,25 µm SMAROS technology from Freescale Semiconductor Package: 128 pin Quad Flat Package Calibration SENSORS Sensors LOGIC CORE (x4) Sensors Regulators Sensors I/Os PLL Sensors EMC test board Standard: IEC 62132 (dedicated to susceptibility tests) 150 mm Size: 150 x 150 mm 150 mm Number of layers: 4 Bottom Top 3
III. On-chip oltage sensor Basic architecture Objectie: On-chip measurements of power supply fluctuation Sampling Command Data IN Attenuator 1/3 X 2 RFI Data OUT Attenuator Voltage diider Attenuation ratio = 1/3 Voltage range Є [0 ; 7.5 V] Sampling Cell Sample and hold circuit Bandwidth = 2 GHz Amplifier Type: Non-inerter Gain = 2 4
Principle of acquisition Random acquisition Probability density: Extraction made with «Labiew» Number of bins defined arbitrarily Information deduced from histogram: Sampling command Ex: f 10 khz Sample III. Voltage RFI On-chip oltage sensor +5 V t +5 V t -5 V +5 V t -5 V RFI amplitude (minimum and maximum) on die RFI waeform Probability density (%) -5 Voltage (V) Probability density (%) = (Number of samples / total number of acquired samples) * 100 0 +5 5
IV. Characterization of sensor DC characterization Goal: calibrate the sensor by determining its DC transfer function (sensor gain G SENS and offset V offset ) Setup: V IN_SENS 5 V 0 Voltage ramp 0 5 V t Sampling command (f 10 khz) 0 t Attenuator 1/3 X 2 V OUT_SENS DC transfer function measurements Vout_sensor (V) 4 3.5 3 2.5 2 1.5 1 0.5 0 Conclusion: DC characterization 0 1 2 3 4 5 Vin_sensor (V) G SENS = 0.65 G ATTENUATOR * G AMPLIFIER = 2/3 V offset = 50 mv 6
IV. Characterization of sensor AC characterization Goal: erify the constant gain G SENS (f) oer all the frequency range and determine the sensor bandwidth BW SENS (= the 3 db cut-off frequency F -3dB ) Setup: AC transfer function measurements G SENS (F -3dB ) = 0.325 G SENS_DC = 0.65 V IN_SENS 0 t f = 10 khz 3 GHz Sampling command (f 10 khz) 0 t Attenuator 1/3 X 2 V OUT_SENS F -3dB 2.5 GHz Conclusion: G SENS 0.65 from 10 khz to 2 GHz F -3dB 2.5 GHz 7
Random sampling of basic signals Signal sampled Sampling command (f = 1 MHz) (f 10 khz) 0 t 0 t Attenuator 1/3 X 2 IV. Characterization of sensor Measurement SETUP Probability density (2000 pts) Signal sampled Signal sampled (f = 1 MHz) 0 t Sampling command (f 10 khz) 0 t Signal sampled Attenuator 1/3 X 2 8
Goals: V. Experimentation Characterize propagation of conducted RF disturbances injected inside the IC (power rail of I/O) Compare external & internal measurements Experimental setup: EXTERNAL Measurements Oscilloscope (Sampling rate = 3 GHz) 500 MHz passie probe INTERNAL Measurements Post-processing with Labiew (Probability Density) Signal Synthesizer Power meter Sampling command f 10 khz t sensck % sens out 4 5 6 V P INC P REFL odd sens IN Sensor sens OUT Amplifier RFI Directional Coupler C DPI = 6.8 nf Decoupling Network F SSNIN = 1 MHz ssn IN RPullD R SSN dd ss I/O oss EMC board Ground Chip under test 9
V. Experimentation Goal: Compare external and internal measurements at low frequencies (1 MHz & 50 MHz) P RF = 20 dbm, F RF = 1 MHz P RF = 20 dbm, F RF = 50 MHz EXTERNAL OSCILLO 5.24 V 4.32 V = 5 V ssn IN odd CONCLUSION: = 5 V 5.3 V Good correlation between internal and 4.72 external V measurements @ 1 MHz & 50 MHz 5.1 V 4.9 V ssn IN odd INTERNAL SENSOR Variations of RFI leels (depending on impedance of IC) on power rail are accurately measured with the sensor 10
V. Experimentation Goal: Compare external and internal measurements at high frequencies (500 MHz & 1 GHz) P RF = 20 dbm, F RF = 500 MHz P RF = 20 dbm, F RF = 1 GHz EXTERNAL OSCILLO = 4.6 V = 5.36 V CONCLUSION: = 4.97 V Probe bandwidth (500 MHz) limits accurate characterization of RFI injected inside IC = 5 V INTERNAL SENSOR = 4.61 V On-chip sampling allows accurate = 5.39 V = 4.73 V measurements of internal RFI up to high frequencies = 5.29 V 11
Goals: Compare immunity leels measured with on-chip sensor and oscilloscope + probe Compare immunity graph with S & Z parameters measurements Immunity test setup: EXTERNAL Measurements Oscilloscope (Sampling rate = 350 MHz) 500 MHz passie probe VI. DPI tests INTERNAL Measurements Post-processing with Labiew (Probability Density) Signal Synthesizer Power meter Sampling command f 10 khz t sensck % sens out 4 5 6 V P INC P REFL odd sens IN Sensor sens OUT Amplifier RFI Vectorial Network Analyzer Directional Coupler C DPI = 6.8 nf Decoupling Network F SSNIN = 1 MHz ssn IN RPullD R SSN dd ss I/O S & Z parameters measurements oss Chip under test EMC board Ground 12
VI. DPI tests Susceptibility test (Direct Power Injection) CONCLUSION: S 11 & Z 11 measurements of odd SETUP Probe doesn t influence the input impedance of odd which is more inductie than capacitie External Standard: measurements IEC 62132-4 underestimate internal disturbance from 175 MHz (= Sampling rate of oscilloscope Pin disturbed: odd / 2) power rail (5 V) For Criterion: this case, +/- 10 maximum % of odd (amplitude) of susceptibilities are obsered at each minimum of reflection coefficient P RF MAX = 30 which dbm corresponds to a resonance RESULTS Maximum gap = 16.27 db Z IN OVDD 50 Ω Oscilloscope & probe limitations 13
VII. Conclusion On-chip sampling: Accurate characterization of RFI injected inside IC up to 2.2 GHz (= BW sensor ) Probability density deduced from on-chip sampling gies information about the RF disturbances (signal shape and amplitude) Comparison between external and internal measurements: Good correlation below 100 MHz Oscilloscope sampling rate and probe bandwidth limit DPI measurements at high frequencies Accurate DPI tests need care for the choice of oscilloscope and probe (Oscilloscope sampling rate and probe bandwidth) Internal measurements: Powerful alternatie to external measurements The sampling bandwidth increases with the technology scale-down Use of on-chip sensor could extend DPI measurements to higher frequency (10 GHz in 90 nm) 14
Thank you for your attention 15
Architecture of sensor RFI Sample command Dummy deices Sampled data x2 Sampling command Storage capacitance Attenuator Voltage diider Attenuation ratio = 1/3 Voltage range Є [0 ; 7.5 V] Sampling Cell Bandwidth = 2 GHz Dummy deices to minimize parasitic offset Amplifier Type: Non-inerter Gain = 2 16