Click here for the enhanced PDF version of this article

FM radio receiver modules are a standard feature in most modern mobile phones. Short-range FM transmission (Tx) has recently become a popular means of transferring audio from a portable MP3 player to a home or car radio, a feature that will be available shortly for mobile phones. Laird Technologies has developed a built-in antenna concept for FM radio reception in mobile phones, the RadioAnt, which provides performance similar to that of the obsolete earpiece cord antenna by integrating the radiating element with a co-designed pre-amplifier. This approach has several advantages compared to traditional passive solutions.

One advantage is that the requirement of an antenna impedance of 50 is effectively removed. This is important at FM frequencies, where the achievable radiation resistance is approximately 1 m. While the intrinsic radiator-amplifier impedance interface is not on the order of 50 in the active antenna concept, the output can be adjusted to any impedance level, including 50 single-ended or 200 differential impedances, for a proper connection to the receiver input.

The gain of the pre-amplifier suppresses the noise contribution of the FM receiver, which is about 6 dB. This is equivalent to using a passive antenna with 6 dB higher gain. This high gain of the active antenna provides more suitable signal levels to the FM receiver, due to the limited dynamic range of the automatic gain control of standard receivers. While the higher gain does not improve the signal-to-noise ratio (SNR) at RF frequencies (as noise and signal are amplified equally), it does significantly improve SNR at the downconverted audio frequencies. Yet, the amplifier does not need to be unconditionally stable, eliminating the need for resistive loading that would severely reduce the gain and increase the noise of the antenna.

This active antenna does have drawbacks, but these are manageable. Specifically, the design and characterization is more complex, and the pre-amplifier consumes power and PCB area. Also, the active element must be protected from ESD without degrading the sensitivity. Most important, stability and linearity must be achieved without resistive loading, even though the antenna will present a nearly open or short-circuit impedance at the amplifier input.

The main figure of merit for active antennas is the total gain (antenna + amplifier) normalized by the total output noise temperature, G/T^[1] (referred to as “G over T”). Now, if the amplifier gain is increased, the output noise will increase, and there is no improvement in terms of G/T. For example, the G/T for a lossless, perfectly matched short-dipole or loop antenna (with directivity of 1.8 dBi) in room temperature is -22.8 dB/K (1.8 dBi — 10 log (290 K)). The concept presented here for G/T degradation relative to a perfectly matched lossless short-dipole antenna is similar to that of the noise figure (NF) in that the SNR is compared at two different nodes, but without the requirement of a matched source at 290 K noise temperature at the input (as is defined for the NF metric). Typically, as most electrically small antennas have a directivity of 1.8 dBi, the gain G is considered as an “average gain” over all angles, which is identical to the standard antenna efficiency (therefore, 0 dB or 100%, is maximum). Throughout this article, gain is used synonymously with efficiency and, therefore, does not include directivity. A G/T degradation of 10 dB, for example, gives a system performance equivalent to that of a passive antenna with -10 dB efficiency (if both antennas are connected to noise-free receivers).

The G/T degradation value in a real application is, besides the antenna properties, influenced by two external effects: the surrounding noise temperature T_a, which will increase the output noise, and the noise figure of the receiver, NF_rec, which will increase the antenna's output noise T_out (and hence reduce the G/T). The value of T_a has been shown to be significantly higher than room temperature, T₀, at FM frequencies (for example, 290 K or -174 dBm/Hz), due to man-made RF noise^[2]. The increased noise level means that the effect of the noise contributions from the active devices and resistors is reduced unless, as in the case of built-in antennas, the gain of the radiating element is so low that the physical temperature of the antenna dominates the noise temperature. Also, the high background noise level means that the efficiency requirement of the radiating element can be reduced without as significant a reduction of G/T as for the ideal low-noise case. This can be understood qualitatively by noting that a highly efficient antenna will receive a larger signal level than an antenna with low efficiency, but it will also receive more noise. Hence, the SNR at the antenna output is not significantly better.

The second effect, NF_rec, also contributes noise to the antenna output, but can be made insignificant by selecting a sufficiently high gain of the amplifier (G_amp > NF_rec), thus improving the system NF performance compared to using a passive antenna. It should be noted that the two effects of background noise and NF_rec are not in general separable, as for example a high background noise temperature can make the noise figure of the receiver irrelevant, and vice versa.

The G/T degradation of an active antenna can be calculated if the efficiency of the radiator and the gain G_amp of the amplifier are known (assuming that the antenna is at room temperature, T₀, and “sees” an ambient noise temperature of T_a ) by the following equations (in which temperature is in units of Kelvins):

The SNR at the output of the active antenna compared to the SNR at the input (the noise figure) is given by:

The SNR at the output of the passive lossless ( = 1) reference antenna compared to at the input is given by:

with the G/T degradation being the quotient between Eq. 1 and 2 as follows:

In general, however, the efficiency of the radiator and the gain of the amplifier are not known separately (at least not through measurements, but simulations or analytical models can be used to obtain such data). Instead, the G/T degradation is obtained by measuring the total output noise power of the antenna when placed in a specified environment (for example, where T_a = T₀ in an anechoic chamber) and the gain is measured by a method such as comparison to an antenna with known gain. Care must be taken so that noise contributions from the measurement equipment are removed by calibration, and that no metallic objects (coaxial measurement cables or voltage supply wires) are attached during these measurements for reasons previously explained.

To support these requirements, Laird Technologies has developed a cable-replacement system based on fiber optics, which together with a battery-driven pre-amplifier facilitates proper characterization of electrically small antennas (Figure 1). The gain of monopole antennas of different lengths protruding from the chassis has been measured with a coaxial cable and with the optical-fiber system. The measurement error is more than 20 dB for the coaxial system for lengths below about 10 mm, which are realistic values for built-in antennas.

Finally, it should be noted that the gain of small antennas at FM frequencies is enhanced by the presence of the human body, in particular if the antenna or chassis is touched by the user. This is due to the human body being a fairly efficient antenna at around 100 MHz, as a half-wavelength is around 1.5 m and the human tissue is conductive at such low frequencies. This is in contrast to cellular antennas, which can lose more than 10 dB gain in talk position. The positive human body effect is illustrated in Figure 2, where the output spectrum of the receive antenna is shown with and without a user touching the antenna. The gain is much higher in the hand-touching case, and it can also be shown that the G/T degradation is improved by 10 dB to 15 dB in this case.

The design of the RadioAnt active antenna concept is shown in Figure 3. The radiating element is a single-turn half-loop, where the radiator is grounded at one short edge of the chassis and connected to the amplifier at the other. By short-circuiting the antenna at one short-edge and ac shorting (at GSM frequencies) the antenna at the other short edge by the shunt capacitor (to obtain resonance), the antenna is shorted at the E-field maxima of the GSM antenna, thus ensuring low cross-talk. The shunt capacitance is between the gate and source nodes of the amplifier input, which besides increasing the gain also provides a better noise match by increasing the real part of the antenna as seen by the amplifier (improving stability). The amplifier uses a microwave FET transistor configured in a common-source topology to minimize the noise contribution. The complete amplifier consumes 3 mA at 3 V, which gives sufficient gain and linearity for the application. The bias point is stabilized by dc feedback, and the noise contribution from the bias network is reduced to nearly zero by design. As the microwave transistor has a positive gain on the order of 10 GHz and beyond, care must be taken to ensure stability at the source impedance presented by the antenna. The input impedance of the radiator is only sensitive to magnetic materials (as it is a short loop), which is a relatively rare occurrence, so the antenna is not de-tuned by the influence of nearby objects. The sensitivity to GSM crosstalk was measured by placing a reference dipole antenna (824 MHz to 960 MHz, and 1710 MHz to 2170 MHz) adjacent to the phone and connecting it to a high-power CW transmitter. The onset of signal deterioration was detected at approximately +36 dBm at 824 MHz (worst case frequency), which is well above the peak output power of GSM.

The measured G/T degradation and gain of the RadioAnt is shown in Figure 4 for an implementation in a Nokia 6125 mobile phone, shown in Figure 5. This phone can be operated in two modes, open and closed, with somewhat different performance. Typically, the open position is a few dB better than the closed position due to the longer chassis length, but it is anticipated that the closed position will mainly be used during radio listening. Notice that while the gain is highly resonant with an in-band variation of approximately 20 dB, the G/T degradation, which is the important figure of merit for reception and sound quality, is nearly flat, with about 5 dB in-band variation. Tunability is, therefore, not required. However, the RadioAnt module does support frequency tunability (if control signals are available from the FM receiver), which will improve the received SNR level at the band edges by a few dB (in particular, if the complete 76 MHz to 108 MHz band must be covered) and also increase tolerance to strong in-band blocking signals. This is an optional feature and is not necessary to obtain good performance.

For transmission, the radiator is used as a standard passive antenna and is connected through an SPST (on/off) switch (Figure 3). The measured gain of the unmatched half-loop antenna, with the pre-amplifier for Rx mode still connected (but turned off) and fed from a 50 source is shown in Figure 6. A wideband average gain (or efficiency) of -53 dB to -49 dB is achieved, and given that the maximum allowed output power in Europe is 50 nW or -43 dBm^[3], the FM Tx power amplifier (PA) must, therefore, be able to supply around +10 dBm and tolerate the induced voltage and current swing at the output. By introducing a matching network between the switch and the FM Tx module, or by suitable choice of PA output impedance, the output power requirement can be significantly relaxed.

Field tests indicate that the RadioAnt performs as well as an earpiece-cord-based antenna for FM radio reception, despite the difference in antenna size. While the active antenna is a nearly optimal design given the small volume, the radiation resistance of about 1 m in combination with a parasitic-loss resistance of at least 1 (such as from the finite conductivity in the radiator metal and interconnect lines) inevitably leads to gains in the range of -30 dBi to -50 dBi. For most RF engineers, it may be difficult to believe that such low gains can be sufficient for any application involving long-range communication, but high ambient noise temperature greatly relaxes the requirement of gain for FM reception.

Most wireless systems operate above 1 GHz, where the ambient noise is close to room temperature and a gain of -10 dBi would translate to a 10 dB reduction of SNR. However, the noise level is about 20 dB higher at FM frequencies (and even higher at AM) in most urban areas due to man-made noise. Therefore, an antenna with poor efficiency will collect less noise as well as less signal compared to a perfect dipole antenna, for example. This is illustrated in Figure 7, where the G/T degradation at different noise temperatures of three antennas with different gains (0 dB, -20 dB and -40 dB) is compared when all antennas are connected to a receiver with a 6 dB NF. For the realistic case of an antenna with -40 dB gain (for a G/T reduction of 46 dB at room temperature), the same configuration has a 19 dB performance improvement at the typical temperature of 23.000 K. With an active antenna, a further 6 dB can be gained from the suppression of the receiver noise figure. Therefore, the SNR performance of a -40 dB gain passive antenna appears to be only 27 dB below that of a perfect dipole.

By co-optimizing the radiating element with a low-noise pre-amplifier, performance similar to that of an earpiece-cord antenna is achieved by the RadioAnt. This enables handsets to support wireless earpieces, as well as transmit and receive FM radio signals, to support many user features.

1. J. J. Lee, “G/T and Noise Figure of Active Array Antennas,” IEEE Transactions On Antennas and Propagation, vol. 41, no. 2, pp. 241-244, February 1993.

2. R. J. Achatz and R. A. Dalke, “Man-made Noise Power Measurements at VHF and UHF Frequencies,” NTIA Report No. 02-390, December 2001.

3. ETSI EN 301 357-1 v1.3.1 (2006-05) “Electromagnetic Compatibility and Radio Spectrum Matters (ERM); Cordless audio devices in the range of 25 MHz to 2000 MHz; Part 1: Technical Characteristics and Test Methods.”

Peter Lindberg is a RF engineer in the research department of Laird Technologies. He obtained his M.S.Eng.Ph., Tech. Lic. and Ph. D. degrees from Uppsala University, Sweden, in 2000, 2005 and 2007. From 2000 to 2002 he worked as an Antenna Engineer at Smarteq Wireless AB before starting his PhD studies in the Microwave Technology group at the department of Signals and Systems, Uppsala University in 2003. He may be reached at peter.lindberg@lairdtech.

Andrei Kaikkonen received the M.S. degree in physics from St. Petersburg State University, Russia, in 1995 and a Ph.D. degree in applied physics from University of Turku, Finland, in 1999. He has been with Laird Technologies, Sweden, since 2004 as RF research engineer working with terminal antenna development. He may be reached at andrei.kaikkonen@lairdtech.com.