Integrating the antenna and radio in mobile phones

Simon Kingsley/Antenova

The designer of internal antennas for modern cellular radio handsets faces many challenges; the increased functionality required of the antenna, the pressure to reduce costs, the reduced space available for the antenna and the problem of expensive customisation during the design life of a handset. These factor are not compatible – increased functionality generally means the antenna must cover more frequency bands and this requires it to be larger, more difficult to customise and quite probably more expensive to manufacture. A new approach is now to address all these issues by considering the antenna and RF front end together as a single unit and making use of balanced RF and antenna structures, thus creating a radio-antenna module.

The need for greater antenna functionality

Global sales of mobile handset are expected to exceed 800m units in 2005, and 1 billion by 2007, see figure 1. There is intense competition between manufacturers to increase their share of this market by offering more services and features to the customer. As a result, the share of smartphones and PDAs is increasing at the expense of more basic phones, see figure 2. The majority of subscribers continue to use 2G networks but there is a steady growth in 2.5G and 3G subscribers and this will be reflected in the handsets and data devices, which will have to be multi-mode to make use of multiple networks. This adds complexity to the design of both the radio and the antenna and the requirement for a small antenna covering five or more bands is now quite common

In handsets at the high end of the market, other radios may also be present such as Bluetooth, GPS receivers, FM radio and, increasingly, WLAN. These too need antennas and finding space for them to work efficiently, and minimising the interaction between them, becomes a major issue for the antenna designer. There is also considerable interest in bringing mobile broadcast services to market, principally mobile TV using the DVB-H or DMB formats. This greatly challenges the antenna designer because the frequencies used are so low (around 175 – 780 MHz) that the radio wavelength is far greater than the size of the handset. Designing antennas that are much smaller than a wavelength, and yet having good bandwidth, is a notoriously difficult engineering problem. All these problems get worse in multi-part phones (such as clamshells and slider phones) because the groundplane does not have a constant length. And if these problems were not enough, waiting patiently in the background is the future requirements for ultra-wide band, RFID and near-field communications, again all requiring new antenna technology.

Cost

Despite the increasing functionality and complexity of small antennas, market forces continuously drive down the cost of handsets and all the components within them. For example, the GSM Association has set a mandate to vendors to make cheaper handsets (less than $30 per unit) and so some models will have to be very low cost. The main costs associated with antennas are the bill of materials, assembly costs and customisation during the design phase. Customisation is not a one-off cost but an expensive process that goes on throughout the design life of the product; see the section on customisation below. The antenna designer must therefore seek not only to make use of lower cost materials and mass production methods, but also find ways to reduce the design costs.

Size

Whilst most handset users are most interested in making calls and sending text messages, style and fashion are still important to them as is the addition of consumer features such as cameras, MP3 players, stereo sound, PC connectivity, location based services, etc. Most of these features take up space in the handset, but the overall space is shrinking as handsets continue to get smaller. This again makes the life of the antenna designer difficult, as the antennas must also be continuously shrunk in size. Generally, with internal antennas, the bandwidth that can be obtained is a function of the volume available. A tri-band antenna might work well with a volume of 2.5 cc, for example, whereas a quad-band might need 3 cc. An interesting fashion requirement at the moment is for ultra-slim handsets; these not only reduce the overall volume available for the antenna, but also apply a restriction on the height available. The bandwidth of internal antennas is also directly related to the available height.

One way to reduce the effective volume taken up by antennas in handsets is for them to share space with other components. Rather than have a separate area of the PCB for the antenna, it can be built like a bridge over the top of other components, or enclose other components inside it. In this way, the antenna still has the volume required to achieve the specified bandwidth, but takes up very little new area on the PCB. If all the RF and transceiver components are located underneath the antenna, this removes the need for lengthy transmission lines beneath the two and effectively creates a radio-antenna module.

Customisation

Despite the fact that the performance of an antenna is related to its size, over the years antennas have continued to get smaller. If one asks the question “can antennas be made indefinitely smaller?” the answer is no, there exists a physical limit, usually called the Chu-Harrington (C-H) limit, that prevents this. The C-H limit is not well defined and is the subject of some academic argument as to its exact representation but it is real enough in the sense that antennas cannot be made vanishingly small. Very roughly, the C-H limit tells us that if the volume of the antenna, expressed in wavelengths, becomes less than unity then the antenna will become increasingly inefficient, lacking bandwidth, or both. Above about 1500 MHz a handset internal handset antenna can be made large enough for it to be an effective radiator. Below this frequency the antenna is not large enough and the whole of the PCB must be made to radiate instead – in effect the whole handset chassis becomes the antenna. This chassis dependence leads to a new type of problem, and to understand why, we need to look at the way antennas work.

The basic principle behind antennas is that accelerating charges produce radiation. A simple way to make an antenna is thus to create two charge reservoirs, separated by a small distance, and then create a mobile charge exchanged mechanism between them in the form of an oscillatory current. The most elementary way of making such a radiating element is called Hertzian dipole and consists of two short, thin linear conductors, driven by a twin wire transmission line with a 180 phase shift between the two feeds. By arranging for the charge flow between the two halves of the antenna to vary sinusoidally, the antenna can be simply analysed and the radiated fields computed.

The Hertzian dipole, indeed any dipole, can be regarded as a balanced antenna. One half of the antenna is driven against the other and there is no need to refer the driving voltages to any other reference level such as ground. The impedance of a dipole is dominated by the inductance of the two wires, the capacitance between them and the radiation resistance (representing the radiated energy lost by the antenna). The ohmic resistance (representing the energy lost in the conductors) is generally much smaller than the radiation resistance if the antenna is designed correctly.

Singled ended or unbalanced antennas are also common, the simplest example being the monopole. Here, one of the conductors is replaced by the groundplane which, if it large enough, creates an image of the missing half such that the field at any point above the groundplane is the same as for a dipole. The radiation resistance of a monopole is half that of a corresponding dipole and an unbalanced feed is needed to drive it. Common unbalanced feeds are co-axial cables, microstrip transmissions lines, coplanar waveguides, etc. In all these feeds there is a centre conductor carrying the signal and at least one outer conductor that is earthed. Electronic circuits can also be balanced or unbalanced and a clarification of the difference between single ended (unbalanced) circuits, fully balanced circuits and push-pull (balanced and referenced ground) is shown in figure 3.

When the groundplane is not large enough to support a complete image of the antenna, such as when an internal PIFA (Planar Inverted F Antenna) is used in a cellular radio handset, then the excitation of radio frequency current on the groundplane becomes much more complex. The whole of the groundplane acts as an antenna to some extent and the design of the handset affects the antenna performance. This is where the problem arises; if there are any significant changes to the handset (PCB length, battery, plastic case, etc.) then this half of the antenna is detuned and has to be corrected by re-designing the PIFA. During the design life of a mobile phone, components are constantly being changed and moved around so the antenna must be constantly re-designed. This antenna customisation work is expensive in both time and money for the antenna designer and the handset designer alike. The ideal situation would be an antenna requiring little or no customisation for any given handset platform and could preferably be used on many different platforms. Such an antenna would help speed up the time-to-market of new handset products.

The ideal antenna

Bringing together the points raised so far, the ideal antenna would be multiband, e.g. quad-band covering all the GSM cellular radio frequencies. It would be balanced above 1500 MHz such that little or no customisation would be needed to use it for different platforms and it would be relatively independent of the PCB and all the components on the PCB. Below 1500 MHz the antenna would necessarily be unbalanced so that the whole PCB/chassis acts as a radiator, but it nonetheless it should be relatively simple to customise it for different platforms, This ideal antenna must not only be simple and cheap to customise, it must also be manufacturable from low-cost materials in the way that conventional PIFAs are. Finally is must either be small, or share PCB space with other components.

A simple balanced antenna such as a dipole could meet many of the requirements above. Unfortunately, a dipole, or similar type of free-space balanced antenna, cannot be used near a conducting surface, such as a mobile phone PCB, because the impedance becomes too low. Another solution must be therefore be found. One way to approach this problem is to write down the antenna equivalents of the circuits given in figure 3, as shown in figure 4. The equivalent of the unbalanced circuit is the monopole or PIFA (fig. 4a) and the equivalent of the balanced antenna is the dipole (fig. 4b), all of which is standard antenna technology. But when we look at how we might make an equivalent of the push-pull circuit then a new type of antenna construct presents itself. The circuit in fig. 3c is in fact a pair of unbalanced filters driven symmetrically about ground, i.e. with a 180-degree phase shift between the inputs. The antenna equivalent of the circuit in fig. 3c can be created by driving a pair of unbalanced antennas disposed back-to-back (fig 4c). By ‘back-to-back’ we mean a pair disposed with mirror symmetry along the axis of charge acceleration. We call this antenna arrangement a complementary pair. The advantage of this arrangement is that each antenna is driven against the ground, but the RF currents flowing in the ground cancel giving a balanced arrangement that does not drive unnecessary currents in the chassis. A further advantage of this push-pull-like antenna arrangement is that if a push-pull power amplifier is used to drive it (i.e. the amplifier is not only balanced but also referenced to ground) odd harmonics are suppressed and if a balanced configuration is maintained between the PA and the antenna there is no possibility of their re-growth.

An alternative complementary pair arrangement is shown in figure 4(d); here the antennas are capacitively-fed short-circuited patches. This arrangement allows for greater control of the resonant loop in the Smith chart than is achieved with PIFAs and offering the possibility of a wider bandwidth solution. An implementation of this idea is shown in figure 5. The resonant frequency of the antenna is managed by careful consideration of the current path between the feed and the ground-pin of the antenna, i.e. creating a longer path between the feed and the ground-pin of each patch will lower the frequency. Ground-pin symmetry is can be used to ensure maximum current cancellation in the ground-plane. Figure 6 shows RF currents flowing on the surface of the PCB. Although there is significant current directly beneath the antenna pair, there is good cancellation everywhere else.

The two-fold symmetry shown in the antenna of figure 5 may be used so that the antenna suffers minimal impedance variation for different locations on the ground-plane. This means the antenna can effectively be located anywhere on the ground-plane in any orientation with minimal tuning and/or matching required to maintain its radiating characteristics. This arrangement is used for all frequencies above about 1500 MHz but because of the C-H limit, the antenna must be unbalanced below this frequency. There are several ways to excite an unbalanced mode in an antenna including creating a separate patch antenna for the low band or by forming a conductive bridge between the two patches, creating a single lower frequency resonance. Separate low band patches have been found to work well in practice.

Radio-antenna modules

Unbalanced antenna architectures have been a major impediment to the development of greater integration within handsets. The customisation needed to develop an unbalanced antenna for each model of handset is expensive enough, but for a complete radio-antenna module it would be so expensive as to be commercial unviable. Nonetheless there are many advantages to creating a single unit containing the radio and the antenna; efficiencies can be improved, size can be reduced, costs are lower and the handset manufacturer does not need to know so much about RF engineering because much of the customisation work will have been carried out within the module design. The antenna of a radio-antenna module could begin to approach the ideal antenna. Now, for the first time, complementary antenna pair architecture makes such radio-antenna modules a commercial proposition by removing many of the customisation disadvantages of unbalanced antennas.

A complementary antenna pair can easily be disposed like a bridge over the top of the RF chip-set by including an electronics bay underneath the antenna; for a quadband or pentaband antenna, this raises the height of the antenna from 5.5 to 7 mm. This radio-antenna module can be split horizontally such that the lower half encompasses all the RF semiconductors plus the balanced drive needed for the antenna, and the upper half contains the complementary antenna pair. This arrangement can be used to simplify testing and enables different manufacturing technologies to be used for the two halves of the module. Combining radios and antennas in this way has many benefits; there are further efficiency gains to be made by removing the transmission line between the two and a reduction in the real estate required on the PCB. However, the most important advantage is that the OEMs and ODMs developing handsets can reduce some of the labour and costs involved in designing the RF front-ends. Less customisation equals a faster time to market and this is important in today’s cellular radio industry. A complete radio antenna module can be used that is simple to install and easily transported between different products. If no unbalanced low band antenna is present then the module can be used anywhere on the product and at any orientation with only minimal changes to the matching components. If an unbalanced low band antenna is present then some customisation may be necessary.

Practical aspects

The antenna shown in figure 5b is driven by an unbalanced 50-ohm transmission line to a point midway between the two antennas. Here a chip balun is used to create a separate 100-ohm feed to each antenna with a 180-degree phase shift between them. A balun is a device for converting unbalanced signals to balanced, and vice versa, and a modern chip balun introduces an insertion loss of about 0.6 dB at cellular radio frequencies. The balun is necessary to test the antenna with conventional 50-ohm test equipment and for connection to existing handsets. Many CDMA handsets also include a diplexer to combine the outputs of the low and high band power amplifiers together. The radio-antenna module then requires a second diplexer to separate them again. These two diplexers introduce a further insertion loss of about 1.4 dB. Figure 7 shows Antenova’s RADIONOVA implementation of the radio-antenna module using a low band flex-circuit PIFA that also incorporates the diplexer and balun to drive a separate high band balanced antenna structure.

Clearly, if the RF front end of the radio were designed to have an unbalanced output to the antenna in the low band and a separate balanced feed to the antenna in the high band then the two diplexers and the balun would be unnecessary. This would reduce the Bill of Materials and save 2 dB of loss. In other words, the functionality and performance of the antenna can be improved through good radio design. Once again, this reinforces the point that the ideal antenna defined earlier is best approached through jointly designing the radio and antenna together. There is still more to be gained however. Most handset power amplifiers have low output impedances (often below 10 ohms) and these must be matched to the 50-ohm unbalanced transmission line and antenna. The matching circuits can introduce loss and the reactive impedance components are not usually completely cancelled out. But there is no need for an antenna to have an impedance of 50-ohm and the reason they are designed like this is probably partly historical and partly because modern test equipment is based on a 50-ohm standard. However, antennas could be designed with a different impedance so as to be a better match to the power amplifiers and to improve efficiency further, especially when the amplifiers are inside the antenna and the transmission line has been eliminated. This joint optimisation of the RF front-end components, instead of designing them as separate components in isolation, is an important aspect of the RADIONOVA concept. RADIONOVA is the marketing name for radio-antenna modules being developed by Antenova, a UK-based antenna company that specialises in novel antennas and RF integration for handsets, laptop computers and other data devices.
The future

With the development of the radio-antenna module, the antenna should no longer be considered as a stand-alone component such as a capacitor. It should be designed to work with other RF components and be integrated with them. This not only saves space and improves efficiency but also enables some long held views on antenna design to be challenged. In future we expect balanced RF components to be developed specifically for the radio-antenna modules and when this happens there will be no necessity to continue using the 50-ohm impedances that are in use today. We expect that impedances will be chosen so as to jointly optimise the efficiency of the power amplifier and antenna. Radio-antenna modules containing balanced complementary pair technology mean the antenna performance begins to approach that of the ideal antenna. This in turn will lead to our original objective of driving down the cost of handsets whilst increasing the functionality and the cellular services offered to the customer.

The first Radio-antenna module to be marketed, Antenova’s RADIONOVA, has at present a rather complex construction involving two flex circuits and two plastic carriers. Recently however, Antenova has developed a single layer/single carrier structure that will be much lower in cost, thereby also helping to meet the original design objectives. Future developments are likely to include multiple antenna modules on a handset (at least in the high bands) to introduce switched diversity and possibly MIMO operation. Multiple antennas can be used to increase data rates and improve the quality of service.