Medical Device & Diagnostic Industry Magazine
MDDI Article Index

An MD&DI  January 1998 Column

EMI FIELD NOTES

Radio-frequency interference poses a major threat to sensitive patient cables, but there are several ways to address this problem with success.

Far and away the most difficult electromagnetic interference (EMI) challenge in medical equipment is resolving the interference caused by the patient connection. The statutory limitations on low-frequency leakage currents severely impede effective control of radio interference. Worse yet, there is a strong possibility that FDA will increase radiated immunity limitations in the future. This article focuses on radiated interference issues surrounding patient cable terminations. Moreover, because traditional solutions are not possible with such terminations, the article also discusses ways to address these issues, including diverting the interference currents from the isolated circuit area and inserting series common mode (CM) impedance on the signal line. Chassis ground must exist for these solutions to work.

THE CHALLENGE

Three primary conditions cause radiated susceptibility at cable inputs. First, signal levels from physiological processes are very weak, which makes them vulnerable to radio-frequency interference (RFI). (We often wonder how anyone ever noticed some of the phenomena in the first place, much less figured out how to monitor them.) Second, input signals are difficult to shield adequately. One end of the cable is inevitably connected to the patient and cannot be terminated, limiting shielding effectiveness. Third, low-frequency leakage current limitations make adequate filtering almost impossible, since the most effective method of interference control—capacitive filtering to ground3is not permitted.

These conditions make this EMI challenge tough to overcome. A new dimension makes the problem even more difficult. FDA experiments have shown that the immunity levels of patient equipment are lower when connected to a patient than when connected to a dummy load: the human body acts like an antenna, sending more interference to the receiving device. This issue is being addressed by a group of the Institute of Electrical and Electronic Engineers' Electromagnetic Compatibility Society (C63.8, Patient Connected Working Group, headed by Howard Bassen, chief, Electrophysics Branch, FDA). Industry leaders (including the authors) are examining test methods and levels to more accurately establish real immunity levels. Whatever the results, it seems certain that the radiated immunity test will become tougher in the future.

THE SOLUTIONS

Most patient monitoring requires the use of two or more electrodes placed at widely separated intervals. The cable and human body combine to form an antenna, sending interference currents to the equipment (Figure 1). Typically, both CM and differential mode (DM) interference will be generated, but CM is much harder to eliminate. Left unchecked, the CM currents will overload the front end of the amplifier, resulting in erroneous signal output. The interference currents must be prevented from reaching the amplifier as well as from disrupting a later stage, most notably at the bridge across the isolation barrier.

Figure 1. Model of patient-connected cable.

Curtailing CM currents can be done in several ways. A typical solution would entail adequately shielding the cable. Unfortunately, in this case, the cable shield ends near the patient, leaving one end unterminated and the other connected to the antenna formed by the patient. Alternatively, if unlimited leakage capacitance were available, the interference currents could be diverted. This is not possible either, however, because leakage currents are severely limited. So, the two best methods of blocking RFI are not available, and designers are forced to rely on less-effective shielding and filtering methods.

To eliminate RFI during patient-cable connections, designers must divert as much of the interference currents from the isolated circuit area as possible. With unshielded cables, the only option is to use filters. It is imperative that filters perform properly to ensure success. Most filters fail to do the job because of poor component selection and poor layout.^1,2 At higher frequencies, filter components degrade and coupling factors become increasingly efficient. Therefore, any modeling must account for the parasitic capacitance and inductance within the filter components and between adjacent components and traces. We call this the hidden schematic. If care is not taken to minimize these effects, the filter won't work properly. Unintended filter resonances will occur at distressingly low frequencies, often below 10 MHz, rendering the filter ineffective above resonance.

Filters. Keeping the hidden schematic in mind, the primary filter goal is to divert the interference currents. It is best to stem the CM currents first, then filter DM currents immediately at the cable termination. Filtering CM currents would be easier if they were perfectly balanced. Alas, we live in an imperfect world where components and signal paths don't quite balance, so designers inevitably need CM-to-DM conversion. Filtering DM currents last enables designers to filter both the original DM noise and the converted CM noise.

A second solution entails inserting as much series CM impedance on the signal line as possible to keep CM currents off the board. Because there is never enough CM inductance, designers should use plenty of CM impedance. Any currents that can't be blocked need to be shunted, but this process should minimize the need for shunting. To maximize the use of the allowed leakage capacitance, designers can use a small capacitor from each signal wire to enclosure ground (Figure 2a) or they can use that small capacitance from isolated ground to enclosure ground and larger capacitance to isolated ground (Figure 2b). We prefer the latter approach because it firmly establishes the isolated ground as the reference, protecting the input amplifier from CM and minimizing DM noise as well. The shunt capacitance from the isolated ground to the enclosure ground minimizes the potential interference across the isolation barrier. If the capacitor is not put in this location, the entire burden will rest on that barrier.

Figure 2. Diverting common mode (CM) currents.

Shields. No doubt about it, using a shield is a vastly preferable method, but it is not a cure-all. DM and CM currents are still present, with the cable shield now carrying most of the CM currents. Unfortunately, the CM currents cannot be confined to the shield because the shield is open at the patient end and cannot be completely terminated at the circuit end. This means that some CM currents still end up on the signal lines. Adding a shield makes the issue even more complex.

In principle, a designer could insert a CM choke to the entire cable, including the shield and the signal lines, and terminate the shield to isolated ground. The isolated ground would be subsequently capacitively connected to enclosure ground (Figure 3a). We prefer to connect the shield to enclosure ground and filter the remaining CM to isolated ground (Figure 3b). The sequence is: capacitance to enclosure ground diverts CM currents, the CM choke then blocks noise on the signal lines, and the two capacitors to isolated ground eliminate CM to the amplifier.

Figure 3. Terminating common mode (CM) currents from shield.

It is important to note that both methods require plenty of experimenting. The concern surrounding incomplete shielding is that it gives rise to standing wave problems on cables. In addition to eliminating adverse resonances with circuit design, designers are faced with cable resonances that can't be eliminated and, worse, that are a function of cable length.

CHASSIS GROUND

The premise of these solutions assumes that chassis ground exists, so a definition and some comments are in order. First, the metallic enclosure is the chassis ground. This in no way implies a connection to earth ground; it simply identifies a reference level. A portable device may be completely floating, but it can still have a chassis ground. Second, the enclosure has some shielding capabilities, which could be either a metal box or a plastic box with a conductive coating. In the extreme case, it might be only a metal plane at the bottom of the box. In any case, the ground surface is a means of quickly distributing the interference currents, reducing the current densities to a level where they will do no harm.

Designers should incorporate shielding and provisions for internal grounding in the enclosure box during the design phase. It is false economy to attempt to build the equipment without some level of shielding--the device will almost assuredly fail the radiated immunity test, leaving designers scrambling to find a retrofit.

CONCLUSION

Suppressing EMI at the patient connection requires considerable care and experimentation. Of utmost importance, the filter design has to account for component and circuit board deficiencies (the hidden schematic). With that in mind, designers should strive to divert CM currents from the circuit board by shunting cable shield currents to enclosure ground, block CM currents from the circuit board by inserting ample series impedance, and shunt CM on isolated ground to enclosure ground using a small capacitor. Following these priorities should be the shortest path to successful test results.

REFERENCES

1. Kimmel W, and Gerke D, "Selecting Components to Minimize EMI," Med Dev Diag Indust, 17(1):212­218, 1995.

2. Kimmel W, and Gerke D, "Filtering Analog Signals in Medical Devices," Med Dev Diag Indust, 17(2):88­92, 1995.

William D. Kimmel and Daryl D. Gerke are principals in the EMI consulting firm Kimmel Gerke Associates, Ltd., based in St. Paul, MN.

Copyright ©1998 Medical Device & Diagnostic Industry