A Medical Electronics Manufacturing Fall 1997 Feature

EMC REGULATIONS

The Impact of EMC Regulations on Medical Device Design

William D. Kimmel and Daryl D. Gerke

Assessing the effects of EU and FDA EMI and EMC regulations is critical to the design of medical electronic products.

Because the sensitive analog amplifiers, sophisticated microprocessors, and other components in medical electronic equipment can be adversely affected by electromagnetic interference (EMI), causing serious risks to patients and health-care providers, various regulatory bodies have developed standards and regulations covering both emissions and immunity. In the fourth edition of Designer's Handbook: Medical Electronics (1995), we addressed the EMI requirements and their impact on device design.¹ These EMI regulations are evolving, and it is time for an update. Thus, this article reviews the current and proposed regulations and discusses those areas where the requirements are likely to cause problems for medical device designers.

In the European Union, the Electromagnetic Compatibility (EMC) Directive, which was passed in May 1989 and was a permitted option for medical equipment, is being superseded by product-specific directives. The Active Implantable Medical Devices Directive is already in place, and the Medical Devices Directive (93/42/EEC) will become mandatory on June 14, 1998.

The European EMI requirements to be implemented for medical devices (EN 60601-1-2) are based on the International Electrotechnical Commission standard IEC 60601-1-2 and address both emissions and immunity.² In the United States, FDA has considered adopting IEC 60601-1-2 as a guideline for evaluating the safety and efficacy of products but has identified some deficiencies, most notably, that it does not clearly define failure criteria. As written, the standard allows a manufacturer to define a "pass" result of a required EMC test as "failing safe." FDA would prefer to indicate that the equipment must meet the criterion of maintaining clinical utility. The agency also feels additional tests are needed as well as increased immunity levels for some types of devices. These concerns are being addressed in the next edition of IEC 60601-1-2, however, and it is expected that FDA will adopt the modified requirements in the very near future, probably with some additional guidance.

Emissions Requirements

The primary purpose of emission requirements is to protect reception of common radio and TV receivers. Radio and TV aside, emissions from most electronic equipment are not sufficient to interfere with medical electronics.

Based on CISPR 11, which was developed by the IEC's Special Committee on Radio Interference (CISPR),³ the EN 60601-1-2 requirements address both conducted and radiated emissions from medical equipment. All equipment will be required to meet either Class A or Class B emission limits. Class A requirements are the least stringent and are generally aimed at systems that are permanently installed, while Class B requirements apply to equipment that may operate in close proximity to radio and TV receivers, for example, systems for use in the home or in a patient room in a hospital. If a device generates electrical interference as part of its function, the manufacturer may be granted an emissions limit exemption in certain frequency bands. In those bands, unlimited emissions will be permissible.

The Class A and Class B limits apply to both radiated and conducted emissions. However, because conducted emissions are tested at the power entrance to the equipment, such limits do not apply to battery-driven devices. Radiated tests measure emissions above 30 MHz and conducted tests measure emissions below 30 MHz. This boundary was set in recognition of the fact that low-frequency interference is primarily conducted (low frequencies don't radiate very efficiently) and high-frequency interference is primarily radiated (high frequencies radiate efficiently, but don't conduct well down inductive wires); although, in practice, the effects of various frequencies may overlap.

The difficulty of designing emission controls into a device varies. High-performance processors and high-resolution video running at clock speeds passing 100 MHz and producing harmonic frequencies to 1 GHz and higher are difficult to handle. High-speed switching power supplies, running at 300 kHz or higher, are also becoming increasingly difficult to control. Emissions almost inevitably originate from periodic oscillators, most often clocks and switching power supplies, so controlling edge rates on all clocked signals is the place to start. Shielding and cable signal filtering will probably be needed as well. Switching power supplies typically switch at around 100 kHz, producing significant harmonics up to and, increasingly commonly, above 30 MHz. Careful component placement and line filtering will be needed. Filtering high-order harmonics immediately at the switching transistor is very effective.

Immunity Requirements

The EN 60601-1-2 immunity requirements, which cover electrostatic discharge (ESD), radio-frequency interference (RFI), and a variety of power disturbances, are based on IEC 60601-1-2, which, in turn, cites the basic IEC 61000­4­series standards for immunity levels and test requirements.⁴ (The IEC 801 and the IEC 1000 series have been superseded.)

ESD. IEC 61000-4-2 addresses the human body model of electrostatic discharge, which is the primary problem, although a significant number of ESD events are associated with the use of IV stands and roller carts. The test requirements call for a 3-kV discharge directly to a device's exposed metallic members and an 8-kV air discharge to parts that may be recessed.

The air-discharge test most accurately reflects conditions as they will occur in the field. In this test, a technician charges up an ESD gun to the desired voltage level, then slowly moves the gun's tip to the device under test (DUT) until discharge occurs. This test has bedeviled users from the start, since results are notoriously unrepeatable for reasons stemming from both internal and external conditions. Internally, the DUT is inevitably in a different state each time an ESD pulse is injected. Indeed, with a clock cycling in the megahertz range, it would be remarkable if a device were ever in the same state twice at the instant of discharge. External variables include the vagaries of the atmosphere and the tester's rate of approach. But even if the approach rate were the same each time, the discharge would never follow quite the same path. In practice, because some parts are more vulnerable than others, the test results do tend to settle down to a few failure modes, so the designer can concentrate on those.

The contact test technique was developed to improve test result repeatability. In this test, the ESD gun tip is held in contact with a metallic surface on the DUT when the trigger is pulled. The actual discharge occurs within the gun in a controlled environment, and the current can be injected at the same contact point each time. There are a few drawbacks to the contact test, however. First, the test requires an unpainted metallic contact area, which means that if the paint can't be removed from the receiving surface prior to testing or if there is no metallic surface directly accessible, the test can't be used. Second, the contact test has proved to be more severe than the air-discharge test, which is attributed to its faster rise times. IEC 61000-4-2 has identified nominal equivalencies in a number of categories. For example, an 8-kV air discharge is in the same category as a 6-kV contact discharge, and a 15-kV air discharge is in the same category as an 8-kV contact discharge. Note the nonlinear relationship. These higher levels may seem irrelevant since IEC 60601-1-2 calls for a 3-kV contact test and 8-kV air-discharge test, but many believe the present test levels are too low. In the United States, 15-kV ESD events are not uncommon in the Frost Belt in winter. The IEC is considering increasing the 3-kV contact test level to 6-kV, so manufacturers would be wise to begin testing to that level now.

Table I. Summary of IEC 601-1-2 requirements.

A very brief transient event, ESD generally only has an adverse effect on digital circuits. Analog circuits are generally too slow to respond to the fast ESD pulse, although sensitive analog circuits that are filtered to protect against RFI may have a long enough time constant that sensitive inputs will, in fact, respond. Most digital circuits are fast enough to respond to ESD events, even when protected on a multilayer circuit board, but those lines that go off the board will experience the worst problems. Reset buttons are exceptionally vulnerable to ESD hits, because the pulse has a direct connection to the reset line. Power-up resets are also vulnerable, but even more prone to the effects of electrical fast transients (EFTs, discussed below). Voltage supervisors, now commonly used in electronics, have proven to be vulnerable to transients of any kind and also need to be protected.

The design solution for these problems is to use filters at the board input as well as at the vulnerable on-board circuits. If the device has more than one board, the designer must ensure that the interconnecting ground impedance is low, or else ground bounce will be a problem. A wide ground strap should be used to connect the boards, since ground pins will not be sufficient, and two-sided circuit boards (with no ground plane) should be avoided, because it is essentially impossible to cure ESD problems on such boards. Shielding and filtering at the equipment boundary may be needed, as well.

RFI. IEC 61000-4-3 specifies a modulated RFI test of 3 V/m. Present requirements call for test frequencies of 26­1000 MHz, the lower limit having been selected to include the citizen band, but this frequency range is under revision. Current plans are to increase the lower test frequency to 80 MHz and add a current-injection test for frequencies below that level. This proposed change is based on the recognition that lower-frequency interference is primarily conducted by, or intercepted by, cables, rather than traveling directly to the equipment itself. The IEC Committee has verified that a continuous unmodulated wave is not nearly as much of a problem as modulated waves, so an 80% modulation is specified. Sensitive modulation frequencies, such as those specifically used by the equipment itself, are to be tested.

Some questions have been raised about the appropriateness of these test specifications. While a test using modulated waves is tougher than one using unmodulated waves, there aren't many RFI sources that are amplitude modulated, and frequency modulation looks very similar to an unmodulated wave to any equipment other than FM receivers. Some lower-frequency devices, such as certain CB equipment and the electrosurgical unit, do use AM, and would certainly be a consideration. Another rationale for using a modulation test is the potential for RFI when radio source amplitude suddenly changes, such as when keying a radio on and off or when a vehicle passes in front of the source. There is also a question of what electric field strength to specify. Cellular telephones and handheld transceivers can produce field strengths above 3 V/m, for example. As a result, the test level will probably be increased to 10 V/m for life-support equipment, although that level will be very difficult to meet for sensitive patient-connected devices.

Whatever changes are made in the RFI immunity requirements, designers will have to cope. So what are the principal threats? Low-level analog sensor lines are the most vulnerable to RFI; digital circuits, on average, are fairly immune to RFI at 3 V/m, even though spot sensitivities can occur at resonances. Protecting the analog lines takes considerable care, especially when patient connections are involved. Careful filter design and component placement are mandatory. Cable and enclosure shielding may be needed, but designers must keep in mind that patient cables cannot be effectively shielded. Voltage regulators are also susceptible to RFI. Filters should be used immediately at the regulator, and the feedback path should also be filtered if it has a separate input.

EFTs and Surge. Both EFTs and surge apply interference to equipment power lines. IEC 61000-4-4, EFT test, is directed at transients that arise from an inductive disconnect, usually on the same circuit but possibly on a nearby circuit. Disconnecting an inductive load will result in a showering arc, characterized by a series of high-speed pulse trains, with voltages rising to 2 kV or even higher. The surge test (IEC 61000-4-5) is intended to simulate a lightning discharge or other large transient directly to the line. This test is also to a level of 2 kV, but the speed is much slower than for EFT: EFT is approximately a 60-MHz problem and surge is approximately a 300-kHz problem. Thus, a high-frequency, low-energy test is used for EFTs, while a low-frequency, high-energy test is used for surge. Even though the voltage levels are similar, the design challenges presented by the two conditions are significantly different. Battery-powered equipment is exempt from these tests unless it can also be run with an ac adapter.

During both of these tests, a generator is inserted between the power line and the DUT. Then either the EFT or surge is injected directly into the power line at 1 or 2 kV, which represents the actual transient that can be expected on the device from the line. In addition, a 500-V EFT is injected into any external data cables longer than 3 m via a 1-m-long capacitor clamp. This requirement is intended to represent the ground bounce that might occur when plugging two pieces of equipment into separate grounds. Because this cable test tends to be troublesome, designers should try to keep data cables shorter than 3 m if at all possible.

Overvoltage conditions represent the primary risk of a low-frequency high-energy surge. EFTs, being low energy and high frequency, will attack the power supply electronics and may also pass through or around the supply to attack digital electronics within. All too often power supply designers put in ripple filters and neglect to look at the high frequencies that may be encountered. Depending on component placement, EFT may also bypass the power supply completely, field coupling off the power line to affect the internal electronics. In a fashion similar to ESD, EFTs tend to attack the reset circuit, often through the power supply but occasionally directly to the circuit board.

With regard to external cables, properly terminated cable shields will do a fairly effective job of shunting transients to the equipment enclosure. Poorly terminated cables will lead to problems, since voltage drop across the termination will result in capacitive coupling to the signal lines. Unshielded cables, of course, will be subjected to the full impact of the transient. Thus, the remedy is to use well-terminated shielded cables, or else filtering will be required.

Conducted RFI. IEC 61000-4-6, Immunity to Conducted Disturbances, Induced by Radio-Frequency Fields, is a new test intended primarily to provide protection against conducted RFI and, secondarily, to supplant RFI testing at low frequencies. Basically, the test involves injecting a voltage onto the power line and a bulk current into the data cables. A current probe is clamped around the entire cable bundle and radio-frequency energy is injected. Conducted tests always cover a low frequency range in recognition of the fact that low frequencies don't radiate well and high frequencies don't conduct well. For this conducted test, the frequency range is 150 kHz­80 MHz. Above 80 MHz, the radiated RFI test (IEC 61000-4-3) takes over.

As with radiated RFI, conductive interference attacks analog lines first. The primary defense is filtering. Control can be very difficult to achieve, involving very large filter elements that need to be effective over a very wide frequency range.

Magnetic Fields. Now in draft form, IEC 61000-4-8 is intended to guard against interference from low-frequency magnetic fields, which originate mainly from power line distribution. The principal victims of magnetic fields are ballistic electron devices, notably computer screens (CRTs). The proposed test level of 10 A/m (approximately 125 mG) will cause severe wiggling of the image on a CRT, which will be very difficult to eliminate. Fortunately, wiggling may be tolerable for brief periods when applying the criteria of "clinical utility." Any analog circuits that are sensitive to magnetic fields (certainly Hall effect devices) would be at risk, but no problems are expected with digital electronics.

Power Line Disturbances. A test specification for power-line disturbances is also under consideration (IEC 61000-4-11,Voltage Dips and Interruptions, draft). Power cycle dropout would be covered along with longer-term sags. An additional requirement is for the equipment to be left in a safe condition following long outages.

Designers will need to provide adequate energy reserves in the equipment power supply to ride through the specified sags and outages, at least for critical functions. They will also need to ensure that the equipment doesn't resume full operation in an unsafe condition.

Conclusion

As both the European Union and FDA aggressively pursue EMI regulation, these requirements have become a major consideration for designers of medical electronics. Existing emission and immunity limits are being reviewed and updated, and new immunity requirements are under consideration. Fortunately, it appears that FDA and the EU are coming closer on requirements, so designers may soon be able to follow a single set of rules.

In the past, EMI problems often have arisen after equipment has been produced or marketed. With the current focus on the issue, and to ensure the safety of the device, protection against EMI must be designed in from the start­adding it on later is usually difficult, always expensive, and sometimes impossible. As discussed in this article, there are identifiable problem areas that can be addressed early in the product development process. The use of filters, shielded cables, and other protective components can effectively harden medical electronics to the threats of EMI.

References

1. Gerke DD, and Kimmel WD, "Assessing the Design Impact of Medical EMI Regulations," in Designer's Handbook: Medical Electronics, 4th ed, Santa Monica, CA, Canon Communications, pp 10­16, 1995.

2.Medical Electrical Equipment­Part 1: General Requirements for safety­2. Collateral Standard: Electromagnetic Compatibility­Requirements and Tests, IEC 60601-1-2, Geneva, International Electrotechnical Commission (IEC), 1993.

3. Limits and Methods of Measurement of Electromagnetic Disturbance Characteristics of Industrial, Scientific, and Medical Equipment and Medical (ISM Radio Frequency Equipment), CISPR 11 and Amendments, Geneva, IEC, 1990, 1996.

4. Electromagnetic Compatibility (EMC)­Testing and Measurement Techniques­Section 1: Overview of Immunity Tests, IEC 61000-4-x, Geneva, IEC, 1992­1996.

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

Illustration by Griesbach/Martucci/SIS