Transcript Document
Near field scan immunity measurement with RF continuous wave A. Boyer, S. Bendhia, E. Sicard LESIA, INSA de Toulouse, 135 avenue de Rangueil, 31077 TOULOUSE cedex, France. E-mail : [email protected] 1/18 Outline 1. Introduction : immunity methods overview 2. Description of the near filed scan immunity method 3. Modeling of the aggression 4. Case studies 5. Conclusion 2/18 Introduction : immunity methods overview Methods Advantages DPI IEC 62132-3 (localized method) Simple model Frequency limit 1 GHz Low power, Low cost BCI IEC 62132-2 (localized method on N pins) Cable injection Frequency limit : 400 MHz No specific test board Weak coupling Test on actual production boards TEM/GTEM IEC High frequency method Weak coupling Complex coupling Special board High frequency method Far field Complex model Much space and expensive 62132-4 Drawbacks (Global aggression) Mode Stirred Chamber IEC 62132-6 (Global aggression) Objective : Find a simple method to characterize and investigate the immunity of each part of a circuit at high frequency 3/18 Description of the near field scan immunity method • Reuse of near field scan in emission (IEC 61967-3) Few mm • Injection of RF disturbances by a miniature near field probe Magnetic probe RF generator Near field probe Field produced by the probe Lead frame device under test Parasitic induced currents Test bench Principle Advantages : Localized disturbances coupled on lead frame of packages Produce immunity cartography or scan No specific test board or test fixture required Frequency limitation : several GHz, linked with the resonances of the antenna 4/18 Description of the near field scan immunity method General set-up of near field aggression experiment Signal synthesizer Amplifier Pforw Prefl Directional coupler Failure detection : Near field probe Oscilloscope Device under test Pattern exit • Generation of continuous wave aggression. • The use of a directional coupler allows to know the forward power in the injection loop. Results are given in terms of forward power. • For each frequency, the forward power is increased until the failure is detected • The forward power that creates a failure is stored. • The probe is then moved to a new position 5/18 Modeling of the aggression Electrical modeling of the probe Tangential magnetic field probe L=0.13m, εr=2.2 Line with losses Inductive load SPICE model for the magnetic probe • Validation of the model up to 10 GHz • No antenna resonance below 10 GHz H i dS 2 S M 12 I 0 I1 I1 Electrical modeling of the coupling 2 RF Generator L1 I1 n dS Ground plane Hi k L2 Inductive coupling M 12 L1 L2 • Interesting method to evaluate the mutual coupling coefficient : PEEC method 6/18 Modeling of the aggression Partial Element Equivalent Circuit to find mutual coupling Ib Ia Conductor A 2. 1. All conductors are meshed in elementary filaments Conductor B Inductive coupling between conductors is computed by adding the influence of all filaments on each other : 0 la lb Lab dVB dVA 4 S a Sb A B ra rb 3. Gives directly an electrical model which can be simulated under SPICE. La Lab Lb 7/18 Modeling of the aggression Electrical modeling of the coupling: Validation Signal synthetizer Validation case : coupling to a micro-strip line 40dB Amplification Scan on X z εr=2.2 h=1mm y x 50Ω Spectrum analyzer Pmeasured (dBmW) +20dB/dec. Coupling vs. position Coupled power vs. probe position Transmission coef. vs. frequency • Efficient coupling at high frequency • Good correlation for maximum coupling • Model valid until 6 GHz 8/18 Modeling of the aggression Modeling of the radiated magnetic field z q 1 2 o2 1 Hr j Ib 2 cosq j 3 3 e j r 4o o r or o Hq r b I Hr Ej Based on the approximation of the elementary loop crossed by a current 1 o2 1 1 Hq j Ib 2 sin q j 2 2 j 3 3 e j r 4o o r or o r o y Elementary loop x Comparison with measurement – Calibration of the injection loop RF Generator Injection probe z h=1mm x y Hy(f) for h=1mm Measurement probe (calibrated) Spectrum analyzer Hy(x) for h=1mm and f=500MHz Calibration of the injection loop to determine the radiated field as a function of the incident power, the frequency and the distance. Good correlation until 2GHz 9/18 Modeling of the aggression Development of a tool under IC-EMC : • Compute coupling between a magnetic probe and the package leads IC-EMC • Build a SPICE-compatible electrical model of the aggression • Compute H field Immunity simulation flow : Fo no Po Probe dimensions SPICE netlist of DUT+aggression Ibis file Package geometry H field computation Transient SPICE simulation Immunity criteria checked ? DUT SPICE model yes Pinject ed Extract forward power Results Po IC-EMC 10/18 fo freq Susceptibility threshold Case studies Aggression of the PLL of a 16 bit microcontroller Scanned area • Tangential H field • Frequency 490 MHz • Scan height 0.25 mm • Criteria : 5% variation of the frequency of the bus clock Hy Immunity cartography Apparition of a weakness zone located on the digital supply of the PLL pin (VddPLL) Quartz susceptibility 450 MHz VddPLL aggression vs. frequency 450 MHz Correlation between immunity threshold and impedance between VddPLL and ground of core Vss Impedance between VddPLL and Vss core 11/18 Case studies Aggression of the PLL of a 16 bit microcontroller A measurement in a TEM cell has been tried : no failures detected. H field generated close to the probe above a pin of a TQFP 144 package: Htot (A/m) Pforw = 31dBm F=480MHz H=1mm Hmax=9A/m Hmean=3.6A/m Theoretical H field generated in TEM cell : if Pinc 31dBm on 50 then Vin Pinc 50 7.9V ETEM Vin 7.9 E 176 V / m H TEM TEM 0.47 A / m dTEM 0.045 For an equivalent power, the maximum H field generated by the probe is 20 times greater than in TEM cell. 12/18 Case studies Aggression of a 10 bit ADC of a 16 bit microcontroller Scanned area • Tangential H field VSSA • Frequency 500 MHz • Scan height 0.25 mm AN0 Hy • Criteria : LSB modification Highlights 2 susceptible areas located on the analog ground of the ADC pin (VSSA) and on the input of the ADC channel (AN0). VSSA influence Aggression of the input of the ADC Weakness at low frequency (in-band aggression) which depends on conversion clock In-band aggression Second weakness linked with VSSA susceptibility (see next slide) High frequency susceptibility Weakness at high frequency (800MHz-1.4GHz) AN0 aggression vs. frequency 13/18 Case studies Aggression of a 10 bit ADC of a 16 bit microcontroller Aggression of the analog ground of the ADC Weakness at low frequency (in-band aggression) In-band aggression Weakness around 500MHz 500 MHz VSSA aggression vs. frequency • Correlation between immunity threshold and impedance between VSSA and supply rails of core Vdd/Vss. 450 MHz • These weaknesses are linked with supply impedance resonances Impedance between VSSA and Vdd/Vss core 14/18 Case studies Aggression of an input port of a 16 bit microcontroller DPI aggression of an input port Near field aggression of an input port • Two different injection methods, two different results. • Only one common point : susceptibility level decreases with frequency above 1 GHz. • Does the same model predicts these 2 results ? Currently, only DPI injection modeling has been established. Near field injection is on going. 15/18 Case studies Aggression of an input port of a 16 bit microcontroller Reuse of the ICEM model built for emission. Useful blocks to build a susceptibility model in DPI : Block behavior model Injection path model measure Measure/given Susceptibility SPICE model Passive Distribution Network IBIS model Measure/ICEM Measure/given • Model of DPI injection valid up to 1.8 GHz • Z model shows the influence of the different parameters 16/18 Case studies Aggression of an input port of a 16 bit microcontroller Comparison DPI injection measurement/simulation Simulation problem Comparison measure/simulation of forward power Comparison measure/simulation of transmitted power • Good correlation until 900 MHz. • Model built from first order parameters, without any confidential data • High influence of the injection path and of the PDN and IO model. Essential parameters for a future ICIM model. 17/18 Conclusion • A method of susceptibility characterization of ICs using near field has been presented. • Main advantages : Valid until 6 GHz Help to detect susceptible pins of the integrated circuits Simple inductive model • A modeling software have been developed to predict the coupling, the radiated field and build an electrical model for susceptibility. • Several cases have been presented which shows different effects of near field aggression. • Future work : propose this method as an extension of BCI standard method to higher frequencies 18/18