EMI/EMC Compliance

Problem Definition

Designs for automotive applications have to meet stringent EMI/EMC requirements. See Ford's requirements for EMI/EMC compliance . Every automotive OEM has some flavor of such EMI/EMC specifications. From a user point of view, you would not want to have your car malfunction whenever you drive past a cell-phone tower or an electrical transmission tower. From a manufacturer's standpoint, you don't want to fix your car for issues that may arise in the field that results in massive recall costs. From a design perspective, a premium-class automobile has 70 to 100 microprocessor-based electronic control units networked throughout the car and has about 100 million lines of software code. How does one ensure that all of the electronics works well together and does not cause any malfunction? As you can imagine, finding an issue or troubleshooting such complex systems is quite a challenge.

In order to forestall such issues at the system level, it is better to find and fix design issues at the block level. Hence, system designers specify that each electronic block or module meets stringent criteria for electromagnetic interference and electromagnetic compatibility. By doing so, EMI/EMC issues are pushed down to a lower-level block that is less complex. Hence, the problems can be solved more easily (or at least that is the intent!). Before a sensor or module is used in the car, it has to be tested and validated to show that it can meet these specifications.

There are two parts to the EMI/EMC specifications. The EMI part is addressed by tests, where the device under test (DUT) is subjected to various forms of electromagnetic energy. This can be classified as two types of tests - conducted immunity and radiated immunity. In conducted immunity, the EM energy is conducted into the DUT via the cables that are used to connect it to the system. In radiated immunity, the part is irradiated with EM waves and is required to function within a specified tolerance. The EMC part of the tests also consists of two parts - conducted emissions and radiated emissions. The DUT is checked to see that it does not radiate energy into space or does not transmit energy through its cables that it is not supposed to. (Of course, this is only in frequency bands that are not used in regular operation. For instance, a radio is expected to be sensitive to and pick up the EM signals!). Though most articles and analysis in the industry club these together, they are different - and a design is required to comply with both aspects - though some design changes for one aspect may also improve the other.

How to mitigate EMI Issues

EMI specifications are intended to ensure that your sensor or module does not cause any other unit to fail in the system. If there are unintended radiation from your module, this may be picked up by other blocks and cause a malfunction. Many modules may share the same power-line that delivers power from the battery. Hence, emissions on the power-line can cause other sensors or modules to malfunction.

If your design is failing radiated emissions, then you have some circuitry in your design that is radiating energy at unwanted frequencies (obviously!). You can follow these guidelines to mitigate these issues:

1. Reduce the power: Total radiated power is both a function of the energy supplied to the antenna and the directional gain of the antenna. By reducing the power in the circuit (this could be using weaker line-drivers or changing the clock frequency by spread-spectrum techniques), one can reduce the amount of radiated power.
2. Change the antenna gain: The most likely scenario is either a transmission line or loop in the circuit is transmitting the energy. By reducing the loop area or shielding the transmission line, the effective antenna gain of the radiator is reduced. This can also lower the radiated emissions and enable the design to comply with the specifications.
3. Line termination: Unterminated lines or improper terminations give rise to reflections or standing waves. These are the characteristics of a good antenna. Proper terminations will ensure that there are no reflections, lowering the radiated power.
4. Bypass capacitors: Using effective by-pass capacitors provides a low-impedance path from the supply to ground. When done effectively, this reduces the loop area for high frequencies. Hence, this can also improve radiated emissions.

If the design is failing conducted emissions, the problems are different. This implies that the circuit is drawing power or emitting conducted energy at higher frequencies. The following techniques can help:

1. Spread-spectrum: If the clock in the system uses spread-spectrum, the energy is spread over a wider band. Hence, the peak-emissions on the power line or the transmission lines can be reduced.
2. Use of power-line or signal filters: Adding ferrite-beads and by-pass capacitors on the power line increases the AC-impedance on the power lines - looking out from the circuit. This is a low-pass filter on the power line and hence removes high frequency energy from going out of the module. For a signal line, an RC filter can also be used.

Reducing EMC issues

EMC specifications are intended to avoid problems when system integration is done and other blocks are added into the system. There can be some unintended energy from the other systems - due to the physical proximity or shared busses. These issues are more severe in a sensor. The signal levels are very low at the sensing elements. Any coupling of energy into the sensor nodes will obviously cause a big deviation at the output, as the injected noise is amplified with the sensed signal. The following techniques can be used for mitigating EMC issues:

1. Shielding: EMC shields or faraday cages are used in many designs to reduce the energy coupled through radiated pickup. However, it can be expensive to add an EMC shield or a faraday cage. In some cases, there may not be physical space to add such a shield.
2. Short leads: Reducing the length of wires that carry sensitive signals can also reduce the amount of stray pickup on these lines.
3. Signal filters: Use low-pass filters on the sensor outputs, so that only the desired signals are transmitted. Signals outside the bandwidth of the sensor should be filtered out by the filter.
4. Power-line filters: Adding ferrite-beads and EMI filters on the power supply can lower the noise on the supply lines for the signal-conditioning electronics. This will make the design more robust and improve the margin of compliance.
5. PSRR: Use amplifiers that have improved PSRR specifications. Better PSRR numbers indicate that these amplifiers reject supply noise more effectively and hence improve the circuit performance.
6. Differential circuits: Use of differential circuits can improve noise that is coupled as common-mode. This technique is widely used in communication circuits, where the receiver is physically located far from the transmitter. In these cases, noise that is coupled to both lines equally is rejected as common-mode noise.
7. Uniform sampling: Many sensors use asynchronous sampling, where the signal is only sensed when there is a need. In such scenarios, it becomes difficult to debug or filter the noise. If the signal is sampled at uniform time-intervals, digital filters or software filters can also be used.
8. Digital filters: If the signal is sampled at a multiple of the Nyquist rate, then information is only present at a fraction of the bandwidth of the digitized signal. Any noise that is folded outside the bandwidth of the desired signal can be filtered out by digital signal. If the signal is only sampled at the Nyquist rate, this is extremely difficult - as any noise is aliased into the signal band and is indistinguishable from the sampled signal. (Acoustic noise cancellation or adaptive filters can still work under these circumstances!!)

Observations

The above techniques were used in recent automotive sensor design. A 200x improvement was observed in the test response from before and after the EMC related changes. This enabled the design to change from being non-compliant to being compliant to the specifications.

Conclusion

A designer needs to put up-front thought to meet EMI/EMC requirements of the system. Thinking of EMI/EMC as an afterthought will result in complete design revisions, impacting the cost and design cycle time. Multiple levels to testing and qualification add to the design cost as well. It is inevitable that a design meeting more stringent EMI/EMC requirements will cost more - due to the additional components required. By designing for EMI/EMC from the get-go, a system can be designed to be more tolerant - and hence require less mitigation efforts. This not only results in a more robust design, but also is more cost-effective. Though the article above gives some general guidelines, each design and its EMI/EMC specifications require that we tailor the solutions to meet optimal design and cost requirements. If you have difficulty meeting your specifications, we can provide design consultation and help.