A Method for Evaluating EMI/RFI Gasket Material in PCMCIA Cards

David A. Case and Michael J. Oliver

Figure 1. The PCMCIA assembly was tested at horizontal (pictured) and vertical polarizations.

Evaluating the effectiveness of an RFI shielding gasket is usually straightforward. One simply tests the product with and without gasket material installed and compares the shielding effectiveness graphs or numeric profiles. If several types of gasket materials are involved, you can repeat the test for various combinations. This method may be simple if you have a completed project. But what if shielding effectiveness must be determined before the electrical design is complete?

This was the problem confronting Instrument Specialties Co. (Delaware Water Gap, PA) and Aironet Wireless Communications (Fairlawn, OH). The product in question was a 2.4-GHz spread spectrum transmitter and radio fabricated on a PCMCIA card form factor. Before the unit was produced, it was necessary to determine what gasket material was the best type to use prior to the system being sonically welded together.

The solution was found with an easily repeated test method to evaluate the shielding gasket material. The preliminary test method was based on a method used at the now defunct ZDS EMC lab to evaluate electronic cabinet designs. The big change was the physical size of the unit, since we were attempting to evaluate a PCMCIA card rather than a computer cabinet. The final test configuration was a modified MIL-STD-285 developed at Instrument Specialties, using a test protocol developed by Aironet.

Test Parameters

The problem with performing an evaluation of gasket material on a radio-based product, or even a computer, is that temperature variances and changes in the operating state of the device can make repeatability difficult. For example, duty cycles of the radio transmitter and slight variations in power output over temperature can change the test results. Even if the transmitter has been modified to run at 100% duty cycle, there are still enough other factors to add additional uncertainty to your measurements. Finally, when the transmitter is not yet designed, how can you test the gasket material within the frequency bandwidth of interest?

Testing Solution

Our answer to this dilemma was to design and fabricate a dipole antenna on a PCMCIA card and install the PC antenna into the PCMCIA shield. The card itself was cut to match the size and outline of the planned product. The antenna was milled out of the top side of the board, leaving a conductive trace around the outer edge. The back side of the card incorporated a solid sheet of copper as a ground plane. We tested two versions of the card, one without a ground plane; we obtained the best results with the ground plane.

As the radio design evolved, the PC antenna card layout changed to match it. The 50-() cable (antenna lead) center conductor was soldered to one side of the dipole and the cable shield soldered to the other side and the ground plane; an SMA connector was attached to the other end of the cable. After many measurements and setup configurations, it became clear that soldering the cable shield to the ground plane on the back of the antenna card was an important step in achieving accurate and repeatable data. It was very important to make sure that the ground connection was well established on the ground plane under the dipole antenna. If this ground plane was left floating (no electrical potential), or had a poor electrical connection, there would be no difference between measurements with or without a shielding gasket. This effect is caused by radiated energy coupling to the bottom lid of the PCMCIA card housing and radiating as an antenna element due to near-field effects. Adding a small ferrite core at the connector end near the PCMCIA card lid helped reduce unwanted emissions radiating from the coax cable. To avoid unwanted emissions, low-loss cables (greater than 90 dB isolation) connected the generator to the spectrum analyzer.


Figure 2. Left-side setup for each polarization.	Figure 3. The signal generator and spectrum analyzer.

A fixture was designed and fabricated to retain the PCMCIA card assembly (antenna, gasket, and lids) and to establish repeatable compression of the gasket configuration when under test. We tested the assembly at horizontal and vertical polarizations (Figure 1), with left side, right side, and front setups for each polarization (Figure 2). To establish an initial reference level, the dipole antenna was tested without top and bottom lids. The signal generator (transmitter) and spectrum analyzer (receiver) are shown in Figure 3.

To evaluate shielding gasket material, a control (in this case, reference level minus the PCMCIA card without a shielding gasket) must be compared with the test specimen (reference level minus the PCMCIA card with a shielding gasket). This experiment determines whether the gasket material is providing the required shielding effectiveness. Thus, the total shielding effectiveness (dB) of the test specimen is determined by a reference level (dBm) minus the test level (dBm) when the gasket is in place.


Figure 4. The dipole antenna elements were insulated from the shielding gasket with plastic tape.	Figure 5. Ferrite and conductive tape shielded open areas of the PCMCIA card housing.

The dipole antenna is shown in Figure 4. The dipole antenna elements were insulated from the shielding gasket with plastic tape. The electrical contacts of the dipole antenna at the base of the antenna card were insulated from shorting to the top lid with plastic tape. The tested unit included ferrite around the antenna lead and conductive tape on the back and front sides to shield open connector areas of the PCMCIA card housing, as shown in Figure 5. By loading the receiving cable with a 50-() load, and transmitting at a power level of 10 dBm, ground-floor noise levels were recorded per the test bandwidth of 30 MHz to 1000 MHz. Calibrated instrumentation was used in measuring all test parameters.

The dynamic range of the test system configuration (maximum output in dBm minus noise floor level) was established at approximately 120 dB. The antenna card without lids was tested and its response zeroed out, and identified as 0 dB shielding, as shown in Figures 6–9. The system was tested at predetermined test points from 30 MHz to 1 GHz. (Though our main concern was digital emissions, we also looked from 1 to 3 GHz to determine shielding effectiveness at the receiver's local oscillators and transmit frequencies.)

Test Specimens

We evaluated two electrically conductive elastomer test gasket materials and configurations for this project. The first specimen configuration was a molded-in-place gasket made of silver aluminum silicone elastomer. This specimen was molded to the inside perimeter of the top and bottom lids of a PCMCIA card frame. The specimen made electrical contact from the lids to the copper trace around the outer edge of the antenna test card. The second specimen configuration was an extruded gasket made of silver silicone elastomer that was cut to a length of 2.4 in. The specimen was press fitted onto the copper trace around the outer edge of the antenna test card and it made electrical contact from the trace to the top and bottom lids.

Figure 6. PCMCIA shielding effectiveness for the horizontal polarization, right side of PCMCIA card.

Figure 7. PCMCIA shielding effectiveness for the horizontal polarization, left side of PCMCIA card.

Figure 8. PCMCIA shielding effectiveness for the vertical polarization, right side of PCMCIA card.

Figure 9. PCMCIA shielding effectiveness for the vertical polarization, left side of PCMCIA card.

Data Evaluation

The characteristic shielding properties for the molded-in-place gasket performed well in comparison to the extruded gasket. For all test positions and polarizations, the molded-in-place gasket provided an average of 20 dB more shielding from 180 to 280 MHz and 600 to 1000 MHz. At the remaining frequencies, 30 to 155 MHz and 300 to 550 MHz, the gaskets performed relatively the same, with each signature trace interweaving.

At 900 to 1000 MHz the extruded gasket tends to be equal to or less in shielding effectiveness than the configuration with lids but no gasket. This low-shielding characteristic from 900 to 1000 MHz was not exhibited by the molded-in-place gasket. Instead, it showed an increase in shielding of about 20 to 30 dB. This result is due to superior electrical contact with the PCMCIA lids of the molded-in-place gasket compared to that of the extruded gasket.

In summary, the molded-in-place gasket provided an average of 20 dB higher shielding effectiveness than the extruded gasket for various frequency ranges and test positions within 30 to 1000 MHz. The extruded gasket averaged approximately 15 dB in shielding effectiveness at various frequency ranges and test positions compared to the configuration with no gasket.

Conclusion

After evaluating the gasket material, our next step was to see how it compared in a real-world test with production samples. A series of radiated tests for Class B compliance were conducted at 3m at a couple of test sites. For this testing, the radio was analyzed with a laptop computer on an extender card, eliminating any shielding benefits the laptop computer could have provided. Results showed that the radio without gasket material had the worst profile—in fact, it was out of FCC Class B at several frequencies. The radio with the extruded gasket material had about a 10 dB improvement in the profile across the board and readily met FCC Class B. When the over-molded gasket was compared to the extruded gasket, we saw again about a 7–15-dB improvement over the extruded gasket (below 1GHz). These results were very similar to what we saw when we tested the test board with the various materials, and they helped prove out the test method. Despite some minor differences, the method worked, providing useful engineering design data.

This same test method can be used for evaluating system cabinets. One would need to make the antenna on a board roughly the same dimensions as the system motherboard, or whatever the suspected emittor of RF noise. Additional thought on making a better match for the antenna may help improve results. (For the next round of testing, the plan will call for the coax to be terminated by 50 () before connecting to the antenna to help reduce mismatch.)

Our best recommendation is to try laying out all the parameters before you begin testing. Although the method is far from quick (at the start), it can provide a repeatable baseline if done correctly.

David A. Case, NCE, is the senior compliance and reliability engineer responsible for worldwide product approval at Aironet Wireless Communications Inc. (Akron, OH), a firm with expertise in spread spectrum technology. His responsibilities include radio type approvals, as well as EMC design and safety approvals for spread spectrum—based products. He also handles product reliability studies, EOS/ESD evaluation, and antenna and power supply evaluation for WLAN systems. Case is NARTE-certified in the fields of telecommunications and EMC and ESD control, and he currently chairs the IEEE EMC Society Representative Advisory Committee. He can be contacted by e-mail at dcase@aironet.com.

Michael J. Oliver is a product development electrical engineer at Instrument Specialties Company, Inc. (Delaware Water Gap, PA). His responsibilities include electromagnetic shielding development, project management, and testing. Oliver has expertise in EMC shielding technology with a background in military shelter electrical systems and high-power antenna/radome design. He currently serves as vice chairman of the ASTM D09.12.14 Electromagnetic Interference Subcommittee, as a member of the Executive Committee for the SAE AE4 Electromagnetic Compatibility Committee, and as a member of the IEEE EMC Society. He can be contacted by e-mail at mike_oliver@instr.com.

Back to March/April Table of Contents