Look for the Antennas

By William D. Kimmel, P.E.
and Daryl D. Gerke, P.E.

Kimmel Gerke Associates, Ltd.

Radiated emissions and susceptibility inevitably involves an antenna, but antennas are not always obvious – sometimes you have to look for them. Let’s take a look at the unintentional antennas that may be in your equipment and some methods of dealing with them.



Antennas

Currents flowing near any metallic/dielectric boundary will serve as an antenna, both for radiation and reception. This includes any wire and cable, enclosures with openings, printed circuit boards and on-board circuit elements, as well as any non-electrical metallic member.

For fairly low frequencies, the antenna is almost always the power and data cables attached to the equipment. At higher frequencies, openings in the box or metallic elements inside the box become effective antennas. At still higher frequencies, we look at traces on the circuit board or even the leads inside the chip itself.

So what does low frequency-high frequency mean? It’s a matter of dimensions and frequency of the metallic elements. A metallic element starts to become an efficient antenna at about 1/20 wavelength, and becomes very efficient at path lengths of a quarter or half wavelength (depending on whether it is end fed or center fed)

C = fl, where c is speed of light, f is frequency and l is the wavelength.

We can convert this to an easier relationship:

300 = fl, where f is frequency in MHz and l is the wavelength in meters.

With this relationship, we can calculate the free space wavelengths of a 1/4 wavelength, a 1/2 wavelength, and 1/20 of a wavelength at various frequencies, producing the data listed in Table I.

In spite of what you may interpret from reading antenna efficiency from a log-log scale, antenna resonances are not all that sharp – plotting on log-log scales is a bit deceiving. To a first approximation, antenna efficiency below resonance is linear with frequency.

Antenna efficiency, generally specified as “Antenna Factor,” reaches a maximum at the first resonance (1/4 or 1/2 wavelength). While it is certainly possible to get higher efficiencies at higher resonances, that takes carefully controlled geometries and is not encountered in haphazard antenna formation that is found in randomly spaced cables, etc. So the principal antenna effect occurs a fraction of a wavelength of the recipient or transmitter, and additonal contribution due to additional length is not significant.

Looking at the table, we find the first resonance will be in the 1/4 to 1/2 wavelength range, depending on the termination and stray loading effects, and the last column, 1/20 wavelength, represents the approximate onset of antenna behavior. The 1/20 criteria is a bit arbitrary (others will say from 1/50 to 1/10 wavelength), but the key issue is that the dimensions are comfortably below resonance – these cases usually dominate at the lower frequencies. Above that, you are starting to close in on the the lowest frequency resonances, near 1/4 and 1/2 wavelength. Additional resonances at 1/2 wave intervals may well occur. Thus, an emission or immunity band of, say 300 MHz, indicates a resonance due to some metallic member of 25 to 50 cm in length, or 10 to 20 inches. This might be from slots in the enclosure due to excessive fastener spacing, or perhaps internal cabling or wire harness.

What are the antennas? The dipole, loop and slot antennas are the most common in EMI.

Types of Antennas and Their Characteristics

We can’t avoid antenna effects without completely enclosing the electronics in a complete metal box – an icebox. But there are things that can be done to reduce the effects. You need to look at the basic geometry of the particular antenna, and determine what parameters can be altered to reduce antenna effects. If you can’t eliminate (or at least reduce the effectiveness of) the antenna, you need to eliminate the energy source.

Wherever there is a conductor, there is a possible antenna. But the most common ones are:

Dipole – this is the traditional antenna recognized by most people. It is simply a length of wire attached to your equipment, and includes power and data cables and let’s not forget the shield – if currents get on the shield from any source, you have a dipole antenna. External wires and cables, having the longest dimensions, will become effective at the lowest frequencies – radiation at 30 MHz is almost always from the external cables. Circuit boards also make dipole antennas (even though the flatness may obscure the fact) – having smaller dimensions, these will become effective at higher frequencies, usually well into the hundreds of MHz.

Up to the first resonance, the antenna efficiency is proportional to the length of the cable, so cable length should be no longer than necessary to make connection. Often, cable length is not a controllable parameter. Your only real control is to use a shield, with circumferential terminations at both ends – no pigtail connections or single point grounds allowed for high frequency shielding.

Loop – this is the most common antenna inside the box and would include internal ribbon cables, power harness, circuit board traces and lead frame/bond wires in the chip. Loop antennas are often the most controllable. Simply put, the antenna efficiency is proportional to the loop area. Anything you can do to reduce the effective loop area directly reduces antenna efficiency. Thus, internal cables should be routed close to the ground plane. Avoid crossing slots in the ground plane. Shield internal cables if necessary, and ground both ends. Internal crosstalk can be further reduced if you separate noisy from sensitive lines, avoid running in parallel, and, if lines must cross, route them orthogonally.

Loops can be reduced on the circuit board, also. Make sure all critical signal lines have an adjacent return path to reduce loop area. Avoid crossing slots or gaps in ground (or power) planes. In the GHz range, we are also concerned about the loop areas in the chip. Most people are not in a position to work internal to the chip package, but you can still reduce loop areas by placing a ground patch immediately underneath the package. Chip shields are effective, they need to be grounded to circuit ground around the perimeter. Heat sinks need to be grounded in all four corners, as a minimum.

Slot – this is the analog of the dipole antenna. You have the currents bending around the slot and exciting the slot. Your solutions are to minimize the longest dimensions of the slot or to reduce the current flowing over the slot.

Patch – (or radiating plane) typically occurs where a metal patch is elevated from the circuitry, such as with a heat sink. The solution is to ground the patches at frequent intervals, every corner as a minimum.

Summary

When chasing down a radiated emission or susceptibility problem, look for possible antennas. Problem areas found in frequency bands are almost always due to a resonance, and this will occur near a quarter or half wavelength. If you can’t reduce the effectiveness of the antenna, you will need to reduce the available energy feeding the antenna.