ICONIC 2007 St. Louis, MO, USA June 27-29, 2007 A NEAR FIELD INJECTION MODEL FOR SUSCEPTIBILITY PREDICTION IN INTEGRATED CIRCUITS Ali Alaeldine 12, Alexandre Boyer 3, Richard Perdriau 1, Sonia Ben Dhia 3, Mohamed Ramdani 1, and Etienne Sicard 3 1 ESEO Angers - 4, rue Merlet-de-la-Boulaye - BP 30926-49009 Angers Cedex 01 - France 2 IETR - INSA de Rennes - 20, avenue des Buttes de Coësmes - 35043 Rennes Cedex - France 3 LESIA - INSA de Toulouse - 135, avenue de Rangueil - 31077 Toulouse Cedex 04 - France Abstract This paper introduces a complete electrical simulation model of a near-field injection setup, used to measure the immunity of integrated circuits to radiated near-field aggression. This model includes the measurement setup itself, as well as the integrated circuit under test, its environment (printed circuit board, power supply), and finally power losses. Therefore, the amount of power injected through a near field probe, triggering a malfunction of an integrated circuit according to a given criterion, can be identified and predicted at any frequency up to 1 GHz. The validation of this model is ensured through a comparison between measurement and simulation results. Keywords: EMC, integrated circuit, near-field aggression, immunity measurement, immunity simulation. 1 INTRODUCTION Within the recent years, many digital and analog integrated circuits (ICs) have become more and more susceptible, due to an increased number of interfaces, higher data rates, decreased node capacitance, and a steady reduction in power supply voltage and, consequently, noise margin. In order to characterize the electromagnetic behavior of these ICs, several measurement methods have been developed by the International Electrotechnical Commission (IEC), including near-field [1]. However, this standard is limited to emission, although near-field can be used for immunity as well, by injecting power into the pins of an IC through a near-field probe and observing its behavior according to a given criterion [2]. This method is proposed as an extension to the existing BCI test methodology [3] for frequencies above 400 MHz. Moreover, the characterization of this behavior by simulation, during the design phase of the IC, is time- and money-saving. Therefore, this paper introduces a susceptibility-oriented electrical model of a near-field injection set-up, making it possible to predict the immunity of an IC on a given printed circuit board (PCB). Corresponding author - E-mail : ali.alaeldine@eseo.fr
2 NEAR-FIELD MEASUREMENT SET-UP The measurement set-up chosen for the characterization of the susceptibility of an IC to near-field interference closely mimics the Direct Power Injection (DPI) set-up [4]. It is depicted in Fig. 1. Fig. 1. Near-field injection set-up For this experiment, a digital IC (CESAME), designed by ST-Microelectronics in 0.18 µm technology, is mounted on a 10x10 cm 4-layer PCB [5], and powered by a 9 V battery through a 1.8 V regulator. This circuit consists of 6 cores, differing only by their power supply strategy, and was initially developed for the validation of low-emission design rules. Each core includes 240 identical synchronous base cells, the output of one of them being made available on a pin. The IC is fed with 20 MHz clock and 10 MHz data signals, in order to ensure normal operation. Then, a near-field aggression is injected through a magnetic probe located at 1 mm above the V dd pin of the NORM core. This probe is fed by a RF generator and a 10 W power amplifier through a directional coupler, allowing the measurement of incident and reflected powers through a dual-channel power meter. Therefore, an inductive coupling is established between the magnetic probe and the V dd pin. 3 COMPLETE ELECTRICAL MODEL OF THE NEAR-FIELD SET-UP Since the evaluation of the immunity of an integrated circuit under test requires an exact knowledge of the power actually injected into the circuit, it is compulsory to model the whole near-field set-up very accurately. Therefore, each part of the set-up must be modeled as equivalent passive elements; these individual models must then be combined in order to obtain the whole equivalent model.
3.1 Modeling of the injection system The magnetic probe used in the near-field injection system is essentially a coaxial cable with a copper core. Some measurements performed with a vector network analyzer (VNA) demonstrate that its model is inductive, with a low series resistance. Therefore, the probe is modeled as a series RL circuit, while the cable and the directional coupler are replaced by a transmission line adapted to 50 Ω, with 10385 ns delay time. Then, the inductive coupling between the loop of the magnetic probe and the pin under influence is represented by a mutual inductance (in the theoretical model) or a coupling factor K M (in the Spice electrical model of the whole system under test). This mutual inductance M 12 is given by Eq.1: M 12 = φ 2 = µ S H i d S (1) I 1 I2 =0 I 1 The value of this coupling is computed with the IC-EMC software developed at LESIA-Toulouse [6]. 3.2 Modeling of power losses During a near-field experiment, only a small amount of the injected power enters the IC, the remainder being either dissipated in other discrete components or radiated in free space. From 10 MHz to 1 GHz, these power losses are due to many different phenomena : losses in the directional coupler and the cables, radiated losses around the magnetic probe, skin effect of the probe. Therefore, an equivalent R loop resistance is added in the model, and takes into account these different power losses (Fig. 2, left). R loop is deduced from measurement data obtained using the VNA. A comparison between measurement and simulation, including the internal impedance of the magnetic probe is shown in Fig.2 (right). The measurement also demonstrates that the highest power losses come from the cables. The whole electrical model simulated with Spice includes this resistance. R loop is frequency-dependent [4] and represents the real part of the impedance. 3.3 Modeling of the PCB under test In this study, the PCB under test includes its own power supply, composed of a 9 V battery and several regulators powering the digital core and IOs of the IC. The 1.8 V regulator, the battery and the PCB tracks (including vias) are modeled by series RLC networks from measurements of their equivalent impedances with the VNA. 3.4 Modeling of the CESAME integrated circuit In order to enhance the accuracy of the simulation, the CESAME integrated circuit has to be modeled thoroughly, from package to die.
Equivalent loss resistance versus frequency Impedance of the magnetic probe versus frequency Fig. 2. Characterization of the magnetic probe The electrical model of CESAME s TQFP144 package was obtained from a 3D electromagnetic simulation with HFSS R (Ansoft R ) [7] and verified with ASERIS-EMC2000 R (EADS-CCR R ) [8] at LESIA. It includes the leadframe, the bonding and the pads. Then, the electrical model of the die, including power rails, is added. In order to speed up time-domain simulation, only the transistor netlist of one base cell (reference cell) is included in the whole netlist and simulated with Eldo R (Mentor Graphics R ) [9]. The remaining cells are replaced by an equivalent parallel RC model representing the impedance of all CMOS transistors, and the reference cell is fed with clock and data signals. 3.5 Electrical model of the whole set-up By assembling all the models previously computed, a complete electrical SPICE model of the nearfield setup can be established, and is depicted in Fig.3. The upper left part models the injection path, including the generator, the coupler, the cables and the magnetic loop (R loop represents equivalent losses); the power supply model is located in the lower left corner of the figure, and is connected with the package model, the latter being inductively coupled with the loop (K3); finally (lower right corner), the model of the IC includes its passive power supply network (pads and rails) as well as the active core itself (rightmost part). Since power generators can not be modeled with SPICE, the generator used for measurements is replaced by a sine-wave voltage generator for simulation purposes; the amplitude of the resulting signal increases from 0 to 22.36 V (40 dbm on a 50-Ω load). The incident power is thus given in Eq.2 : P inc = (V in+i in Z c ) 2 2 (2) 50
Fig. 3. Complete electrical model of the near-field injection setup, coupled to the device under test mounted on its test board 4 RESULTS A first immunity simulation was performed from 10 MHz to 1 GHz, including the frequencydependent loss model (R loop ). This time-domain simulation is performed for each frequency step (10 MHz); a failure in the IC is characterized by the undulation of the output signal reaching 20 % of the logic "1" voltage level, or by the jitter of this output signal reaching 10 % of the period. Fig.4 depicts the comparison between experimental measurements and simulation results. It can be noted that the IC is immune to a 10-watt incident power below 200 MHz. Above this frequency, the circuit is becoming more and more susceptible. An encouraging correlation between measurement and simulation results can be observed. The difference is greater then 3 dbm only at 300 and 500 MHz. The discrepancies (including a 10-MHz frequency shift below 500 MHz) may be due to small differences in the impedance of the whole model in simulation and measurement, the disadaptation of the magnetic probe, and the parasitic coupling of the magnetic probe with adjacent pins, which has to be characterized with better accuracy. 5 CONCLUSION In this paper, a complete electrical model of a near-field injection set-up was presented. Each part of the system was characterized and modeled, including frequency-dependent power losses. Immunity simulations were performed using the complete electrical model, and compared with experimental results. They demonstrate that there is an encouraging correlation between measurements and simulations. In a near future, other immunity tests such as very fast transmission-line
Fig. 4. Immunity simulation of the near-field model (dashed line) and measurement results (solid line) pulsing (VF-TLP) will be performed on the same PCB and IC, and their results will be compared with those obtained in this paper. References 1. IEC EMC Task Force. IEC61967-3 : Integrated circuits, measurement of electromagnetic emissions, 150 khz to 1 ghz - part 3: Measurement of radiated emissions - surface scan method. Draft technical report, IEC. 2. J.J. Laurin, S.G. Zaky, and K.G. Balmain. EMI-induced failures in crystal oscillators. IEEE Transactions on Electromagnetic Compatibility, 33(4):334 342, November 1991. 3. IEC EMC Task Force. IEC62132-2 : Immunity test to narrowband disturbances by bulk current injection (BCI), 10khz- 400mhz. Draft technical report, IEC, 2001. 4. M. Ramdani, A. Alaeldine, and R. Perdriau. Power modeling for susceptibility prediction in integrated circuits. In Workshop, EMC Europe 2006, Barcelona, September 2006. 5. B. Vrignon, S. Ben Dhia, E. Lamoureux, and E. Sicard. Characterization and modeling of parasitic emission in deep submicron CMOS. IEEE Transactions on Electromagnetic Compatibility, 47(2):382 387, May 2005. 6. LESIA-INSA Toulouse. IC-EMC. http://www.ic-emc.org/. 7. Ansoft Corporation. HFSS. http://www.ansoft.com/. 8. EADS-CCR. ASERIS-EMC2000. http://www.aseris-emc2000.com/. 9. Mentor Graphics Corporation. ELDO. http://www.mentor.com/.