The Feasibility and Effectiveness of a Common Consumer Device as an Electromagnetic Interference (EMI) Source

Transcription

1 James Madison University From the SelectedWorks of George H Baker September, 5 The Feasibility and Effectiveness of a Common Consumer Device as an Electromagnetic Interference (EMI) Source George H Baker, James Madison University Nathan L Olive Joseph M Darragh Joseph D Rudmin Available at:

2 The Feasibility and Effectiveness of a Common Consumer Device as an Electromagnetic Interference (EMI) Source G.H. Baker, J.D Rudmin, N. L. Olive, J.M. Darragh Abstract Because the operation and control of most critical infrastructures are highly dependent on electronics, it is important to understand the feasibility and effectiveness of devices that may be used to intentionally interfere with electronic system operation. This research project investigated () the feasibility of building a high power electromagnetic interference source from a common consumer item and () the effects of this source on the operation of personal computers. We were successful in fabricating a shop-expedient high power microwave (HPM) source by modifying a common microwave oven. Project results confirm that, with straightforward modifications, a common microwave oven can be turned into an effective EMI source that is capable of disrupting personal computer operation at close ranges. INTRODUCTION Determined malefactors have demonstrated the use of common systems as effective weapons against civilian infrastructures, e.g. commercial jetliners used as kinetic weapons and garage door opener electronics used to trigger explosive devices. This undergraduate research project investigated the possibility of using ordinary systems for electro-magnetic interference or disruption. A major element of the present effort was to gauge the ability of undergraduate students, with minimal experience and guidance, to develop an effective electromagnetic interference source. At present, the system effects of high power electromagnetic environments are well recognized by world scientific and military communities. Former CIA Director John Deutch has said that, "the electron is the ultimate precision-guided weapon."[] It has been known for some time that electronic equipment subjected to high electromagnetic field levels will often suffer degradation. In the course of the investigation ofnuclear EMP effects on electronics during the Cold War period, it became evident that garden variety, unprotected electronics would malfunction, in some cases burn out, in the presence of electromagnetic fields in the s to s of volts per meter. These electronic effects could have serious consequences in terms of interruption or termination of critical system operation. Although military systems were the primary concern for Cold War programs, it was clear that the civilian infrastructure was at least as vulnerable to disruption from intense electromagnetic field environments []. SHOP EXPEDIENT SOURCE The term shop expedient applies to any device that can be designed and constructed using common, readily available equipment. Past efforts have demonstrated development of such components from electronic components []. The present effort explored the feasibility of constructing a high power source starting with a common consumer item: a microwave oven. We used an watt,. cubic foot Ewave Model EWFW microwave oven purchased for ~$. U.S.. Source Modifications The oven was modified to allow the radiation of microwaves outside the oven enclosure. The outer oven door was removed along with the outer casing. The oven door interlock mechanisms were easily circumvented using plastic ties as shown in Figure. Figure :Circumvention of door interlocks Research sponsored by NSR, Inc. university grant under contract N9986--C-7 to the Joint Program Office for Special Technology Countermeasures, NSWCDD, Dahlgren, Virginia. Department of Integrated Science & Technology, James Madison University, MSC, Harrisonburg, VA 87, USA

3 . Safety Considerations.. For safety reasons, we made further modifications to enable operation of the oven from outside a copper shielded room enclosing a cubic experimental volume ~ m per side. Remote operation was accomplished by removing the oven s faceplate control panel. The connecting leads were then individually cut and spliced to - meter long wire extension segments. The extension wires were then run through pipe stem apertures in the shield room. A power bus cord was also threaded through a pipe stem. To limit field leakage through the pipe stems, we installed Ferrishield sleeve and cable snap circumferential ferrite limiters on the extension wires at the point where they exited the enclosure. We used a microwave fluence meter to ensure that the exterior fields were within safety limits. TEST VOLUME AND ENVIRONMENT Tests were conducted inside a wood-framed double copper screen enclosure. A shielded door with a finger-stock perimeter seal enabled access. Using an Advantest U6 spectrum analyzer we measured a shielding effectiveness in the range of - db from.5 GHz based on IEEE Standard procedures. An elevated plywood board supported by a dielectric A brace provided a horizontal surface for positioning the microwave source and movable test objects. Figure provides a top view of the test space. Electric Field (V/m) Electric Field Power Density Figure : Field mapping inside test volume EXPERIMENT PROCEDURE The modified microwave source was placed inside the cage at the end of the plywood shelf nearest to the door of the test enclosure. The oven opening faced the far wall. Test objects consisted of eleven used 998 vintage Dell PC systems, each with a CPU, monitor, keyboard, and mouse.. Exposure Parameters The computer systems were initially placed as far as possible ( cm) from the open oven and moved closer as trials progressed to a range of 75 cm and finally, 5 cm. We set identical exposure times for each trial of 5,, and 6 seconds. Exposures started at 5 seconds and progressed to longer durations on subsequent shots. We wanted to find the point of effects onset and recognized that starting at the lowest exposure conditions gave us the best chance of discovering initial effects thresholds. If a subsystem burned out during testing, it was replaced with another operational subsystem and the test continued. Surviving systems were re-used in subsequent trials. Power Density (mw/cm ). Effect Categories Figure : Test volume, top view. We used a NARDA EMR- omnidirectional field sensor to map the field and fluence along the centerline of the oven door opening with results as shown in Figure. Data point values are averages over five second intervals of oven operation. Observed system effects were recorded using the following numbered categories: No effects observed during and after exposure Minor disruption in one or more of the PC components during This was usually in the form of monitor interference, viz. scrolling and/or size distortion. Level indicates there were no residual effects and no system reset was needed.

4 Moderate upset including system shutdown during exposure with automatic recovery after the microwave source was turned off. No residual effects were observed and no manual reset was needed. Significant upset including monitor blackout during and after exposure and no automatic reboot. One or more of the system components was permanently damaged. The system would not operate properly without replacing the damaged subsystem. 5 EXPERIMENT RESULTS The most common effect observed was monitor interference. During most exposures screen image distortion was observed including scrolling, bending or flickering. During longer exposures, monitor distortion was followed by complete monitor blackout. In most of these cases, the monitor returned to normal operation after A level three upset usually meant that the monitor screen was black until the system was manually reset. We observed a total of five permanent monitor failures. CPUs also experienced significant interference. This was usually in the form of power disruptions. During exposure, CPU power indicator lights switched on and off in a cyclic manner. Post-trial, CPUs almost always rebooted automatically. Several level upsets occurred requiring a manual CPU reset. In several instances we observed effects on the CDROM drive. In these cases, the drive opened spontaneously during We observed two permanent CPU failures. We did not have the resources necessary to perform post-test diagnostics on the exposed computer systems other than a cursory functional checkout. 5. System Effects Data Figure is a composite plot of system effects from all experiment trials. The exposure times are not differentiated in this graph. Note the absence of category points we observed effects on each trial. Because of the uniformity of the test enclosure electromagnetic environment, there was not a strong effects range dependence. The most severe effects were observed at the position closest to the source. Interestingly, the midrange position yielded the least severe effects. Based on our field mapping (Figure ), this is explained by Effect scale Effect vs. Range Distance from microwave (cm) Figure. Composite graph of effects vs. range from microwave source. Effect Effect cavity effects which produced higher fluences near the walls. The five-second trial results are included in Figure 5. The relatively low number of level and effects for short duration exposure indicates the possibility of cumulative effects, i.e., failures due to energy build-up over a period of seconds. Effect scale Five sec pulse 6 8 Figure 5. Range-to-effect for 5 second Figure 6 shows the data for ten-second trials. The graph is generally consistent with the 5-second exposure results. There were two additional burnouts, again lending credence to cumulative damage. Effect Scale Ten sec pulse 6 8 Effect Effect Effect Effect

5 Figure 6. Range-to-effect for second Figure 7 shows the 6-second exposure data. The highest numbers of and level upset effects occurred at this longest exposure duration. The data imply that the probability of serious system effects increases with exposure time. The continuing presence of significant numbers of level effects indicates that some systems have greater inherent hardness making them immune to non-recoverable failures at these fluence levels. This also implies the possibility of screening electronic systems by exposing them to long duration, high power signals in order to weed out weak sisters. Effect Scale Sixty sec pulse 6 8 Effect Effect Figure 7. Range-to-effect for 6 second 5. Data Analysis Statistics on the relative incidence of effects are provided in Figure 8. The most prevalent effect was the level upset requiring manual reboot. About ½ of our trials resulted in effects that were followed by automatic recovery. The other ½ of the trials resulted in effects requiring human intervention, i.e., manual reset, or subsystem replacement. Our experiments indicate that the ~KV/m field level represents a rule-of-thumb level for 5% incapacitation of an electronic system population for ~ minute exposures. Percent of Distribution of Effects for All Trials Effect level Effect level Effect level Effect level Effect level (burn out) A graph of the cumulative numbers of level and effects is provided in Figure 9. The incidence of both effects is logarithmic with exposure time over the 5-6 second domain. Cumulative Failures Level : N =.9Ln(t) R =.986 Level : N =.9Ln(t) R =.98 Exposure Time (seconds) Figure 9. Cumulative Number of Level and Effects. 6 CONCLUSIONS It is possible with minimal knowledge and effort to modify a common microwave oven to create intentional EMI effects, including upset and damage, at close ranges. Our results indicate that effects are enhanced on electronics inside a shielded room. Results also indicate that effects are cumulative in time and exhibit a logarithmic dependence on exposure duration. Acknowledgments We greatly appreciate the grant from National Security Research, Inc., and the advice and encouragement of their program manager, Dr. Stephen Dillingham. We could not have done the experiments without funding and instrumentation loans from the U.S. Navy JPO-STC, and the support and encouragement of Mr. John Latess and Dr. Dave Stoudt. We are also grateful to the IT Department at James Madison University for providing the computer test objects. References [] Congressional Hearing, Intelligence and Security, Chairman Jim Saxton, Joint Economic Committee, June 7, 997 [] Preliminary Study Regarding the Resistance of Critical Societal Systems to High Intensity Electromagnetic Radiation, M. Backstrom et al, Royal Swedish Defense Research Agency, Report FOA-R SE, August Figure 8. Effects Summary