Characterization of Integrated Circuits Electromagnetic Emission with IEC 61967-4 Bernd Deutschmann austriamicrosystems AG A-8141 Unterpremstätten, Austria bernd.deutschmann@ieee.org Gunter Winkler University of Technology Graz A-8010 Graz, Austria gunter.winkler@tugraz.at Roland Jungreithmair austriamicrosystems AG A-8141 Unterpremstätten, Austria roland_j@sbox.tugraz.at Abstract - The electromagnetic compatibility (EMC) of integrated circuits (IC) has permanently gained more and more significance during the last years. At present, ICs are moving towards operation frequencies in the GHz range with rise and fall times in the order of a few pico seconds. The high transition rates, as well as the increased complexity of ICs, lead to high electromagnetic emissions and weak susceptibility, with the result that modern electronic devices often fail to meet the EMC requirements. Therefore, EMC of ICs has to be treated as a challenging issue. New, well-suited EMC measurement methods (IEC 61967 and IEC 62132) for the characterization of EMC on chip-level enable us to perform in-depth measurements of radiated and conducted emissions and the susceptibility of ICs. In this paper the measurement results of the so-called 150 Ω method (IEC 61967-4) are compared to the results of two other methods (IEC 61967-2 and IEC 61967-6). We will see, that using the 150 Ω method, slew-rate controlled output drivers appear to have much higher emissions compared to the other methods. I. INTRODUCTION Manufacturers of electronic systems use standards as a basis for the specification and qualification of new electronic products. Established standards already describe the EMC requirements, measurement procedures, maximum emission and susceptibility limits of electronic systems and modules (EMC system-level). But such established standards are not yet available for ICs used within those systems. Accepted EMC measurement procedures on chip-level will close that gap. Although EMC measurements on chip-level will never be able to completely eliminate EMC measurements on system-level, they will allow the identification of the main contributors to the overall emissions. Direct correlation between chip-level and system-level EMC measurements is impossible because of the different technical ambient parameters. A final approval is only possible after a successful EMC measurement of the whole system. Often some suppression of interference is needed, which can be done by filters and metal shielding. Whether those various EMC measures will be implemented within the IC and/or outside of the IC on the PCB is very much dependent on cost aspects, feasibility and available space. The subcommittee 47A (Integrated circuits) of IEC s technical committee 47 (Semiconductor devices) is just working on two standards: The IEC 61967 for the characterization of the emission and the IEC 62132 for the characterization of the susceptibility of integrated circuits. II. EMISSION MEASUREMENT METHODS Under the general title IEC 61967: Integrated circuits Measurement of electromagnetic emissions 150kHz to 1GHz this new standard consists of six parts. Part 1: Part 2: Part 3: Part 4: Part 5: Part 6: General conditions and definitions Measurement of radiated emissions TEM-cell method Measurement of radiated emissions Surface scan method Measurement of conducted emissions 1Ω/150Ω direct coupling method Measurement of conducted emissions Workbench Faraday cage method Measurement of conducted emissions Magnetic probe method The aim of the IEC 61967 standard is to describe general conditions in order to establish a uniform measuring environment to measure the RF disturbances from ICs. The measurement results can be used for comparison or other purposes. The results of the conducted or radiated emission measurements yield information about the potential for RF disturbances in an application, in which the IC is operated. Due to the complex emission mechanism and electromagnetic behaviour of ICs, there is no single measurement method, which can be used to measure and to estimate all the elements of the electromagnetic 1
emissions integrally. Each of the measurement methods has its advantages and limitations concerning the measuring techniques. Therefore, semiconductor users and manufacturers in this area shall select the one of these measurement methods, which best fits their requirements. II.1 IEC 61967-2 (TEM-Cell method) The TEM-cell method is used to measure the electromagnetic radiation from an IC [1], [2], which is mounted on an IC test printed circuit board (PCB) [3]. This test PCB is clamped to a mating port, which is cut in the top or bottom of a TEM- or wideband TEM- (GTEM) cell. In this case, the test PCB is not as usually in the cell, it becomes a part of the cell wall. This test board has to be designed in a way that all connecting leads within the cell are eliminated. All the connecting leads should be located on the backside of the board outside the cell. Only the operating IC is allowed to be inside the cell. The TEM-cell has two 50 Ω ports. One of these ports is terminated with a 50 Ω load. The other 50 Ω port of the TEM-cell (or the single 50 Ω port of the GTEM-cell) is connected to the input of a spectrum analyser to measure the RF emissions emanating from the integrated circuit and impressed onto the septum of the cell. An EMC receiver can also be used instead of a spectrum analyser. Figure 1 shows an example of the described TEM-cell with a mounted PCB. RF voltage measurement on single pins with the 150 Ω method Applying the 150 Ω method, the contribution of a single pin or of multiple pins to the emission of the IC can be evaluated. The 150 Ω coupling network simulates a long PCB trace (>10cm) or antenna with a typical impedance of 150 Ω connected to the pin of the IC. To match the network to the input of the test receiver or spectrum analyser, a 50 Ω resistor has to be included to the coupling network. This matching network is shown in figure 2 (general test configuration). Figure 2 General test configuration. II.3 IEC 61967-6 (Magnetic probe method) Figure 1 TEM cell with mounted PCB. This part of the standard specifies another method to evaluate the RF emission of an IC pin. The magnetic field caused by the current flowing through the strip-line connected to an output or the power supply of an IC is measured by a triplate magnetic probe [6]. The frequency range is specified from 150kHz to 1GHz and higher. For the measurements described in IV.3 the frequency range is 150kHz to 2GHz. In figure 3 the standardized test setup to measure the magnetic field strength over an output or a power supply line is shown. A high degree of repeatability is provided by this test method if the magnetic probe is placed accurately. II.2 IEC 61967-4 (1 Ω/150 Ω method) In this part of IEC 61967, two methods for evaluating the conducted electromagnetic emission (EME) of integrated circuits are specified. With the 1 Ω method, the direct radio frequency (RF) current is measured by the use of a 1 Ω resistive probe, with the 150 Ω method the RF voltage is measured using a 150 Ω impedance matching network [4], [5]. Due to the simple measurement setup, these two methods guarantee a high degree of repeatability and correlation of EME measurements. 2
Characterization of IC Electromagnetic Emission with IEC 61967-4 IV. MEASUREMENT RESULTS The radiated and conducted emission of the test chip is measured with three different test methods: IV.1 IEC 61967-2 (TEM-cell method) Using the TEM-cell method to measure the radiated emissions of the test chips a significant difference in the emission of the slew rate controlled and the conventional output driver can be found (see figure 5, 6). Note that in this case only the emissions of the die, the lead frame and the package of the IC are characterised. Figure 3 Setup for emission testing (magnetic probe). III. EMC TEST CHIP Two different test chips have been designed, with a focus on the reduction of the electromagnetic emissions by controlling the slew rate of the output signal in combination with a reduction of ground bounce and ringing. Every test chip consists of 8 inputs and 8 outputs. Each input is directly connected to its corresponding output via a variable delay block, to control the propagation time. Two different output drivers have been used. Whereas for test chip 1 eight conventional output drivers without slew rate control, ground bounce, and ringing reduction were used, test chip 2 consists of eight slew rate controlled output drivers. A picture of one of the test chips is shown in figure 4. The total die-size of the test chip is about 1.5mm x 1.5mm. Both test chips were manufactured by using a standard 0.35µm CMOS technology. The dies were mounted into a ceramic DIP 24 package. Figure 5 Emission of the slew rate controlled output driver (IEC 61967-2 TEM-cell). Figure 6 Emission of the conventional output driver (IEC 61967-2 TEM-cell). IV.2 IEC 61967-4 (150 Ω method) Each of the three measurement methods has its own assumptions concerning measuring model and technique. Therefore, direct comparisons of the emissions measured with different methods are impossible because of the different Figure 4 Picture of EMC test chip 2. 3
technical ambient parameters. The measurement results of the 150 Ω method to measure the conducted emissions do not show the same significant difference as seen with the TEM-cell method. The measurement results of the 150 Ω method are shown in figure 7 and 8. capacitor. As expected, the measurement results of figure 9 and 10 show that the emissions of slew rate controlled output drivers are lower than the emissions of conventional output drivers. Again, a significant difference is obvious. Figure 7 Emission of the slew rate controlled output driver (IEC 61967-4 150Ω method). Figure 9 Emission of the slew rate controlled output driver (IEC 61967-6 Magnetic probe method). IV.3 Figure 8 Emission of the conventional output driver (IEC 61967-4 150Ω method). IEC 61967-6 (Magnetic probe method) To ensure that there is really a significant difference between conventional and slew rate controlled output drivers, another method to characterise conducted disturbances is used. Due to the absence of the coupling network, which is used with the 150 Ω method, the influence of the measurement to the output driver is negligible. A magnetic probe measures the magnetic field of the current flowing in the I/O stripline (PCB-trace of the standardised test board), which is loaded with a 10kΩ resistor in parallel to a 50pF Figure 10 Emission of the conventional output driver (IEC 61967-6 Magnetic probe method). V. 150 Ω METHOD APPLIED TO SLEW-RATE CONTROLED OUTPUT DRIVERS As we know, the electromagnetic emissions of an IC can be reduced when slew rate controlled output drivers are implemented. By controlling the slew rate of the output signal, the rise and fall time can be varied to reduce the high frequency content of the frequency spectrum in order to reduce the electromagnetic emissions. Additionally, in some of the advanced output drivers different techniques, for example to eliminate the crowbar current by using break before make 4
techniques in their output stages, or to reduce the over- and undershoot (ringing) of the output signal, are implemented. The coupling network of the 150 Ω method interacts with the internal structures of the output drivers i.e., with the structure to control the slew rate, the structure to eliminate the crowbar current, and the structure to reduce ringing. As can be seen in the lower trace of figure 11, the output signal of the slew rate controlled output driver, which originally has got a rise and fall time of 40ns), is changed to a stepped signal, having much shorter rise and fall times. This results in higher emissions according to the additional frequency content. VII. REFERENCES [1] IEC 61967-2, Integrated circuits - Measurement of electromagnetic emissions, 150 khz to 1 GHz Part 2: Measurement of radiated emissions, TEM-cell method and wideband TEM-cell method (150 khz to 8 GHz), 47A/619/NP, New Work Item Proposal, Date of proposal: Jul. 2001 [2] Andy Engel, Model of IC Emission into a TEM Cell, IEEE International Symposium on Electromagnetic Compatibility, pp. 197-202, 1997 [3] IEC 47A/552/NP, Universal testboard for measurement of EMC of ICs, SC47A/WG9, New Work Item Proposal, Date of proposal: Feb. 1999 [4] IEC 61967-4: Integrated circuits - Measurement of electromagnetic emissions, 150 khz to 1 GHz - Part 4: Measurement of conducted emissions 1Ω / 150Ω direct coupling method, 47A/636/FDIS, Final Draft International Standard, Distributed on 2002-01-18 VI. Figure 11 Output of the 150 Ω probe, conventional and slew rate controlled output driver CONCLUSIONS In this paper the measured electromagnetic emissions of a slew rate controlled output driver is compared to the emissions of a conventional output driver using three different methods proposed in IEC 61967. The emissions of ICs with slew rate controlled output drivers are typically much lower than the emissions of ICs with conventional output drivers. This can only be observed with measurement methods using loads that do not interact with the internal structure of the output drivers in a different way than a regular load does. The coupling network of the 150 Ω method (IEC 61967-4) does not fulfil this requirement in all cases. For that reason care should be taken if this method is applied to slew rate controlled output drivers. [5] Franco Fiori, Sergio Pignari, Analysis of a Test Setup for the Characterization of Integrated Circuits Electromagnetic Emissions, IEEE International Symposium on Electromagnetic Compatibility, vol. 1, pp. 375-378, 2000 [6] IEC 61967-6: IEC 61967-6: Integrated circuits, Measurement of Electromagnetic Emissions, 150 khz - 1 GHz Part 6: Measurement of conducted emissions, Magnetic probe method, 47A/608/CDV, Committee Draft For Vote, Date of circulation 2000-11-10 5