Weeds are simply the wrong plants in the wrong place. You can look at electronic noise and interference in the same way. But if properly treated, noise and interference won’t come back—weeds will.
Interfering signals affect circuitry in a finite number of ways: conduction via the input, output, or power and ground lines; magnetic coupling into chokes or other less well-defined loops; or capacitively by proximity to high-impedance circuit nodes.
That’s not to say preventing or eliminating noise is easy, but there are a number of proven techniques to use. For example, filtering reduces corrupting signals selectively by frequency. A low-pass filter eliminates sharp corners and spikes with a significant high-frequency content.
The term suppression usually applies to preventing or minimizing interference. Examples include surge-arresting varistors used to protect telecommunications systems and snubbing circuits connected across switch or relay contacts to reduce arcing.
In contrast, shielding attempts to isolate and protect an uncorrupted signal from the source of interference. Operating on the principle of a Faraday cage, a shield shunts interference around sensitive circuitry and typically diverts it to ground. Alternatively, a shield can be applied to the source of radiated energy, inhibiting it from interfering with sensitive circuitry outside the shield.
Choosing Your Weapons
Surge-suppression varistors, zeners, gas-filled spark gaps, diode clamp circuits, and metal oxide varistor (MOV) devices are in the first line of defense. These components protect your circuitry from potentially destructive interference caused by lightning strikes, power line surges, or simply high voltages accidentally applied to the input of a sensitive instrument. You use them to ensure survival, not to improve circuit performance.
Negative temperature coefficient thermistors present a high impedance when cold. A soft-start characteristic is provided because as the device heats up, its resistance reduces, allowing the full operating current to flow after a short time. Thermistors also are part of interference prevention rather than cure.
Other types of devices operate when the voltage across them exceeds a specified limit. Zeners or avalanche parts, like transient voltage suppression diodes, maintain a constant voltage while dissipating a relatively small amount of power. They are good choices for short overvoltage spikes in circuits with appreciable source impedance. Similarly, MOVs or multilayer varistors (MLVs) dissipate short overvoltage spikes, although their energy ratings can be much higher than those of zeners.
At higher current levels, four-layer thyristors are a better choice. They don’t start to conduct until a threshold voltage is exceeded. If they do begin to conduct and the voltage continues to rise, they switch to a low-impedance, fully conducting mode. In this state, these four-layer devices handle high currents because the voltage drop across them is small (Figure 1, see June 2000 issue of EE).
Finally, for relatively high voltages, gas surge arresters are preferred. They have very small capacitance and essentially zero leakage current when not conducting. This means that they have minimal effect on the circuitry they are protecting. However, when conducting, gas discharge tubes can handle up to 10,000 A for a short time. In applications where they may be subjected to longer overvoltage transients, special fail-safe switches can be added that are triggered by the high temperature of the overstressed gas tube.
Suppression components direct or dissipate the energy in interfering signals. Ferrites have become very popular as signal speeds increased first in PCs and more recently in mobile phones.
Ferrite materials can be designed to be lossy above certain frequencies. There is no deliberate, wired connection to the material. Instead, the high-frequency energy in the field surrounding a current-carrying wire is attenuated by placing the ferrite around the wire; for example, a ferrite bead that is slipped over a component lead.
Sometimes small-value resistors are placed in series with digital signals. This forms resistor-capacitor (RC) low-pass filters with the stray capacitance associated with the signal conductor. The approach works well and is very low cost, but it has an effect even at DC. In contrast, a ferrite bead may provide similar losses at high frequencies but has almost no effect on the circuit below several kilohertz.
Capacitors can be added in shunt to direct high frequencies to ground. To be effective, their impedance in the desired frequency range must be small compared to that of the node being filtered. Using capacitors in this way presupposes a good ground.
Special-purpose three-terminal capacitors are used for their very low-inductance properties. These components resemble two-terminal feedthrough capacitors, but the bulkhead connection is the third terminal that is usually attached to ground.
The maximum effectiveness of suppression components is obtained only when good layout practices have been followed. Typically, these include the use of power and ground planes, properly terminated transmission lines, and attention to the placement of decoupling capacitors in digital circuits.
The intention is to make performance as good as it can be before adding suppression components or shielding. If interference occurs, the preferred approach is to eliminate it at its source by using suitable filtering/suppression components.
At very high frequencies, parasitic inductance inherent in discrete resistors and capacitors limits filter performance. Integrated RC networks eliminate much of the inductance resulting in predictable multigigahertz performance (Figure 2).
Shielding
In contrast to filtering and suppression components that become part of the circuitry, shielding does not affect the operation of the circuit. Several factors determine whether shielding the source or the affected circuitry makes the most sense. For example, how many sources of interference are there that may corrupt a sensitive node in your circuit, and over how many do you have control?
If you can’t control the sources or if there are several, then it probably makes more sense to shield the sensitive circuitry. That also may be the better approach based on the strengths of the signals involved. Signal strength falls off with distance from the source of interference, so shielding the affected circuitry means that you may be dealing with a signal many decibels smaller than had you tried to shield the source. Consequently, the effectiveness, cost, and perhaps weight of the solution will improve.
Commenting on PCB-mounted shields, Jack Black, director of sales at Boldt Metronics International (BMI), said, “A circuit or component that radiates unwanted electromagnetic energy, if not controlled, can interfere with a component or circuit next to it on the board. By placing an EMI metal shield over the radiating component, the unwanted electromagnetic energy hits the metal shield and becomes an induced current that rides on the surface of the shield.
“We find ferrous alloys are best,” he continued, “because materials with low magnetic permeability, such as aluminum and copper, have a tendency to reflect more of the unwanted EMI. Conducted problems usually are related to the capability to absorb unwanted low-frequency EMI. The best materials for this kind of shielding are ones with high magnetic permeability, often ferrites.”
Shielding a source may cause additional problems. Radiated signals can reflect from the internal walls of the shield with unwanted effects. To reduce this possibility, Emerson & Cuming and Martek provide dissipative/absorptive material engineered for use within shielded areas. The shield doesn’t allow the signal to interfere with outside circuitry, and the absorber reduces the level of the radiated signal within the shield by damping cavity resonances.
But, shields are not perfect. Depending on the material, its thickness, the frequencies involved, and whether magnetic or electric coupling is involved, only a finite reduction in signal strength is provided. The extent of the shield—is it an airtight box or are there a few holes for wires and cooling air—also can make a difference. This is the reason that extremely critical applications may involve multiple layers of shielding.
“Beyond intelligent board design, shielding is the only means of suppressing EMI or reducing susceptibility that does not slow down the operation of high-speed systems,” according to Gary Fenical, EMC technical support engineer at Instrument Specialties. “Shielding not only limits radiated electromagnetic fields, but it also provides an isolated ground reference which effectively reduces internal crosstalk and circuit path coupling as well as common-mode coupling.”
Return to EE Home Page
Published by EE-Evaluation Engineering
All contents © 2000 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.
June 2000