Things to Look for After the Schematics Are Done but Before the Layouts Are Started

9.6.1.1 High-Frequency Voltages and Currents

High frequency is defined as > 10 kHz or so, but the threshold varies among designs. Unfortunately, many potential sources of such signals exist; a few of the more prominent threats are:

• Loop filter capacitors associated with integrated bus clock synthesizers. These often have signals at a multiple of the bus frequency, and very short rise and fall times. Because synthesizer loop filters usually receive current from an up/down charge pump, it is important to also consider how these signals are bypassed to both V_dd and V_ss.

• External buses, especially serial ports. This includes not just the clocks, but the data line(s) as well. Because these signals can travel a relatively long physical distance around the circuit board, they can be serious EMC threats; they are prime candidates for system-level EMC reduction (e.g., by not operating them when sensitive victim circuits are active). External Bus Interfaces (EBIs) and external memory buses also fall into this category. A good circuit board layout will minimize the length of these buses.

• Voltage multipliers, switching converters, and regulators. Often these circuits switch relatively large voltages and currents at frequencies in the tens to hundreds of kilohertz, making them important circuits to place properly in a network node layout. As with external buses, time spent in the system design phase finding ways to allow switching converters to shut down during sensor or receiver operation will pay great dividends near the end of the product design cycle.

• The microcontroller clock itself. It is important to consider this high-frequency circuit as both a potential threat and a potential victim. It is typically a very high-impedance circuit, subject to disruption with the application of relatively small signals; disruption of the microcontroller clock is, of course, usually fatal to device operation.

• External signals. Although all sources of high-frequency energy should be noted, special attention should be paid to potential sources from outside the network node itself — particularly when the node is using a host microcontroller, or is residing in another device, such as a cellular telephone. The host likely was not designed with the EMC concerns of the wireless sensor network node designer in mind, so it is good engineering practice to examine the host for the presence of high-frequency signals that, although they may not interfere with the host's performance, may seriously degrade the performance of the network node. An example is the use of switching voltage converters in the host. The host may have a system design that activates its switching converters only when the host is asleep; however, it is likely that the host and the wireless sensor network node are asynchronous, meaning that it is possible for the host's switching converter to become active while the network node is active, resulting in an EMC problem. A second example is mains noise entering through the power supply of a mains-powered node. This can be particularly difficult in industrial environments, where large amounts of noise can be placed on the mains from operating equipment. Finally, one should keep in mind that most countries regulate the conducted emissions of mains-powered devices; it is important to ensure that the network node itself does not cause interference to other services.^[14]

• Fast transitioning voltages/currents. Almost by definition, high-frequency sources are sources of fast transitioning voltages and currents, so this may seem to be a distinction without a difference. Some signals, however, have rapid transitions at a relatively low frequency; such signals can become relevant to EMC when, for example, the transitions are rapid, not because they occur in a short amount of time, but because they change a relatively large value. In the coupling equations, it is the di/dt or dv/dt ratio that matters, not necessarily the value of di, dv, or dt alone. Circuits that commonly cause difficulty are:

  • Voltage multipliers, switching converters, and regulators. Because they switch relatively large signals relatively quickly, these circuits can be significant EMC problems. They will need to be placed as far as possible from an internal antenna and from sensitive RF and analog circuits; the reactive element of switching converters can be one of the largest components in the network node, so its location relative to the antenna, also one of the largest coponents, should be considered during the early stages of product design.

  • Transducer drivers. Transducers can require very large currents; almost any controller of a physical object, such as a heating, ventilation, and air-conditioning (HVAC) air damper, must switch hundreds of milliamperes. Particularly difficult are large audio sources, such as sirens and alarms; these often require large, square-wave input (for maximum drive efficiency), with very large harmonic content. The drivers are often in a bridge configuration, for maximum output power.

  • Display backlights. Many technologies are suitable for display backlighting, including electroluminescent panels and cold-cathode fluorescent tubes. These can be particularly difficult EMC problems due to their relatively high drive voltages and their necessarily large physical size.

• Areas of high current density. Areas of high current density are important to identify because they can be sources of conducted coupling. Common points of difficulty are:

  • Voltage multipliers, switching converters, and regulations. Particular areas to identify are the connections to the inductor in inductive switching converters, and to the capacitor(s) in capacitive switching converters, because the entire output power of the converter is transferred through these points — often in a harmonic-rich waveform. One should recognize that, in the case of the switching reactive elements, two current loops must be considered — one when the converter switch is closed, a second when it is open — and both loops need to be as small as possible, with minimal current density. Two loops that are associated with the converter input and output filter capacitors must also be considered. In the case of the input capacitor, one loop is associated with the high-frequency currents placed on it by the source (which may not be an issue if it is a constant-voltage source like a battery, but can be significant in some energy scavenging sources) and the second with the return currents of the converter itself. In the case of the output capacitor, one loop is associated with the high-frequency currents placed on it by the converter (which is nearly always significant) and the second with the return currents of the load. When the layout is done, the current paths between the converter and its input and output filter capacitors should be separate from the current paths of other circuits in the node. Note that the production of a low output current at a high output voltage must require a high input current if the input voltage is low. The current density around the input circuits of voltage multipliers is often overlooked.

  • Microcontroller V_dd and V_ss pins. Significant current will pass through the supply pins of the microcontroller when it is active. Due to the large number of pins that must be connected to V_dd or V_ss logic levels in a typical microcontroller, it is often difficult at first glance to identify the main current sinks in a typical chip; often, several are labeled V_dd or V_ss, and some sleuthing is necessary to identify the right pins. One should ensure that the schematic has a bypass capacitor between V_dd and V_ss and recognize that current from other pins, such as those supplying onboard analog-to-digital converters, will need to be routed away from this area of high current density to avoid conductive coupling effects.

  • Transducers. In addition to the di/dt and dv/dt concerns associated with transducers and their drivers, because they can switch large currents, a major concern is their potential for conductive coupling. This potential grows if their large currents get constricted to a small physical area on the circuit board, as can happen, for example, near a connector. If transducers are present in the network node design, the schematic should be examined to identify how their current reaches them from the supply, and how it is returned. This loop will often need to be a separate path; a common EMC error is to combine return currents from the transducer and other circuits in a common connector "ground" pin. Because the contact resistance of a connector can be significant, this is to be avoided, and a separate pin placed on the schematic for the transducer return current.

  • Transceiver V_dd and V_ss connections. The power supply rejection ratio (PSRR) of low voltage (e.g., 1-Volt) RF and analog circuits is typically not as good as that of circuits that can operate at a higher supply voltage. In addition, a given level of noise is a higher percentage of the supply voltage, as the supply voltage is decreased. One must, therefore, pay particular attention to the transceiver supply to minimize its impedance over as wide a bandwidth as possible, by bypassing as appropriate.

Due to its physical size, and the large amount of amplification that follows it in the receive signal path, the location of an internal antenna and, to a lesser extent, the location of the connector for an external antenna, is critical in proper EMC design. The antenna is probably the component with the most contradictory requirements placed on it. It must be big enough to be effective, but small enough to fit in the product; it must be sensitive to very weak external RF signals, but not to any internally generated ones. The antenna must be kept away from lossy components and materials, like batteries, yet must be close to transceiver components to prevent undue RF losses between antenna and transceiver. It is worth spending some analysis time considering possible antenna locations, and the trade-offs involved with each. As discussed in Chapter 8, Section 8.22, an internal antenna will need a "keep out" region surrounding it in all directions, out to a distance inversely proportional to the frequency of operation. A fixed rule for the size of the keep out region cannot be given, as it depends on the physical shape and RF loss of the offending material, and the level of antenna efficiency one is willing to tolerate, but the radius of the keep out region will be on the order of λ/100 for circuit chip components, and λ/10 for larger but passive sources of loss, such as a AA cell.

9.6.1.3 Power Source Placement

The placement of the power source is important because it is likely that many different current paths will have to converge there. There should be space available so that the currents in these paths mix only at the terminals of the power source itself, instead of, for example, a narrow circuit board runner. The power source placement is also important because it is often a source of electrical noise, or at least composed of lossy materials; it should, therefore, be some distance away from the antenna.

9.6.1.4 Sensor Placement

Sensors can be sources of electrical noise, but some types can also be EMC victims if placed too close to the transmitting antenna. The designer must rely on experimentation and previous experience to determine the potential EMC problems of each sensor.

9.6.1.5 Placement of Oscillators

One source of potentially interfering signals inside a wireless sensor network node is, of course, the oscillators associated with the microcontroller and transceiver. Because they typically have high harmonic content, microcontroller clock oscillator circuits can be a problem for RF transceiver circuits and sensitive analog circuits associated with sensors. Conversely, because they are often very high impedance circuits (quartz or ceramic resonators in CMOS inverter strings), they can be corrupted by strong signals from a nearby RF transmitting antenna or power amplifier. Quartz reference oscillators associated with transceiver synthesizers have similar difficulties; although they are often higher power circuits, making them more resistant to interfering signals, they are often required to meet higher spectral purity requirements, meaning that even small degradations in performance can be unacceptable. In addition, they are typically operated at higher frequencies than microcontroller clock oscillators (in low power applications), so lowerorder (and, therefore, higher-energy) harmonics of the synthesizer reference oscillator may reach RF frequencies used by the transceiver.

RF filters, receiver low noise amplifiers, and transmitter power amplifiers were saved for the end of the list, because they are some of the few circuits that should be placed close to the antenna. It is most correct to say, however, that only one port of these circuits (the port connected to the antenna) should be placed close to the antenna. A very great reduction in filtering effectiveness can result if the opposite port of an RF filter is visible by the antenna; instability can result in both LNAs and power amplifiers, if sufficient energy can couple from the LNA output into the antenna, or from the antenna into the power amplifier input. Good engineering practice is to bring these circuits radially outward from the antenna, so that their connections to the antenna are as short as possible, for minimum loss, although their other ports are as far from the antenna as possible, for minimum coupling. When both LNA and power amplifier are integrated, as is usually the case in wireless sensor network design, the job is to place the chip in the proper orientation so that the LNA input and power amplifier output are as close as possible to the antenna. The external signal path from IC to antenna, which may include an antenna switch, balun, RF filter, and perhaps even antenna tuning or other circuits, should be as short and as straight as possible.