Give away medical masks when you place an order. learn more

Smaller Is Better: Miniaturizing RF - Part 1

In order to offer worldwide coverage and meet marketplace demand, smartphones, tablets, and other mobile devices must support more RF bands than ever and feature a growing number of wireless functions. Despite the fact that their RF circuits are becoming more complex, these devices must remain as compact—if not more compact—with each new model. With that in mind, designers are cognizant that one of the benefits of higher frequency is that components can get smaller. Specifically, antennas and inductors can shrink to surface-mount and PCB sizes. Integrated circuits, too, can contribute and have contributed to this shrink-fest as process geometries allow the integration of complete RF sections that seamlessly blend digital functionality with the analog realm of RF modulation, demodulation, antenna matching, and wave propagation.

This article looks at the miniaturization of some key components and systems that will allow designers of next-generation wireless links (hello, Internet of Things!) to provide smaller and more efficient radios and radio subsystems. In this Part 1 installment, we look at the latest passive components that integrate several discrete parts into smaller, surface-mount versions, saving space and cost and improving performance.

These components are useful with modern, highly integrated standard radio link transceiver chips that already offer small-size solutions. Since these established standards are on their third or fourth generation, integrated passives are highly optimized and include miniature front ends, filters, baluns, and other assorted parts that can serve single-standard solutions like Wi-Fi, or multistandard protocols like Wi-Fi Bluetooth combinations.

In Part 2, we will look at active components such as RF transistors, mixers, modulators, and amplifiers for next-generation radios. These pieces of a radio may be useful for prototyping and even production of special function radio links that are not highly standardized. Here, SoC-style multichip modules or carrier boards can be used to more accurately preserve impedances and antenna characteristics. It should be noted that if high-volume manufacturing is to take place, more integrated custom silicon may be the best solution to reduce cost and size until the IC makers catch up. Even in this case, the components discussed here will still be useful. All parts, data sheets, tutorials, and development kits referenced here are available online at Hotenda’s website.

Let’s be discrete

If you’ve tackled radio projects in the past, you know that discrete and passive components are an integral part of the design; from front-end antenna matching to tight-tolerance mixers, oscillators, transformers, modulators, filters, switches, diplexers, and so on. While many of these functions are available as an integrated part, this may not be desirable, especially during development. For example, filters are always needed to discriminate against out-of-band signals and attenuate them down to very low levels while preserving full power in the bands of interest. Classes of discrete filters include lumped LC, dielectric, and SAW (Table 1).

Lumped L/C Dielectric SAW
Loss Fair (3-5 dB) Best (2-3 dB) Good (2-4 dB)
Attenuation Good Good Best (sharp rolloff)
Size (WxL) Fair (200-400 mm²) Good (20-50 mm²) Best (5-14 mm²)
Power Watts Several Watts 0.01-1 W
Frequency <0.5 GHz 0.7-5 GHz 0.7-5 GHz

Table 1: Monolithic filter type characteristics.

While simple filters can be made of a few very small resistors, capacitors, and inductors, more stages and poles can be achieved using more complex structures.

ICs do not make the best filters. The typical inductor Q that can be fabricated using CMOS process technology is less than 10. With copper and metallization techniques, this can be doubled to around 20. The problem is that to minimize insertion loss, Q values in the hundreds are needed. Presently, only discrete filters can do this.

Filters are used throughout a radio inside and outside the silicon barrier (Figure 1). Here, external filters are needed in the antenna matching and signal paths to allow receive data to be stripped off and transmit data to pass with little or no attenuation. This is especially important when one antenna is used for transmit and receive, or is used with more than one protocol.

Figure 1: Filters are needed extensively inside and outside the RF transceivers. Better performance of the external discretes allows internal IF and base-band filters to share a common path among different protocols and bands.

For both standard “in use today” radios, as well as next-generation radios, the smaller more integrated filter blocks like Johanson’s 2450BP07A0100T single-chip band-pass filters are highly optimized — in this case for the Wi-Fi 2.4 GHz band.

Note that better component matching is achievable with these monolithic assemblies. This is because it is much harder to match component characteristics with parts that are manufactured during different fabrication runs. As a result, these combo parts can exhibit higher tolerances than easily achievable using discrete components.

In this case, the 0402 form-factors provide the 100 MHz bandwidths with a tight 2.45 GHz center frequency with a very small 1 x 0.5 mm area and 0.43 mm height. Another nice factor with combo parts is that the orientations of components can often take advantage of 3-D stackups to reduce noise and interference between stages.

Johanson offers a product training module describing how integrated single-chip filters can provide better performance, reliability and consistency through the manufacturing process. Take notice that several of these parts can be combined to improve performance as well. For example, the addition of the TDK DEA102500LT-6307A1 low-pass filter can help block dual-band higher frequencies like the 5 GHz used with higher-frequency wireless LANs. Cascading filter elements is like adding more poles to a filter, and it can improve performance. However, there will also be signal attenuation so a more sensitive front-end amplifier may be needed, especially for multiband operations.

Tapping off the same antenna point as the 2.4 GHz low- and band-pass filter could be a higher-frequency band-pass filter for 5 GHz as well, such as the TDK DEA105425BT-1293A1. A demo/development kit from TDK, the 445-7845-KIT, allows 900 MHz and 2.4 GHz experimentation and contains several diplexers and filters tailored for common protocols.

Playing in the band

It is not only a single protocol that operates in a specific band. For example, in the 2.4 GHz ISM band, Wi-Fi, Bluetooth, ZigBee, W6LowPan, ANT+, GSM, DECT, and several other protocols may need to coexist. While occupying energy in the spectrum, these can use different modulation techniques and different frequency-hopping algorithms. As a result, you will want to choose filters that allow for the differing band characteristics such as widths and data rates.

Take, for instance, the multilayer ceramic Taiyo Yuden FI105L250014-T filter, which is designed to service W-Lan and Bluetooth. Similarly, the Johanson 1810BP07B200T takes aim at GSM and CDMA cellular applications. When a transmission path must service several interoperable bands and protocols, inductor-based common-mode filters can also be useful to route only desired signals to the filters and demodulators. These are getting smaller as well. Consider, for instance, the Johanson Technology Integrated Passive Components or the Abracon ACMF-03 and ACMF-04 series of thin-film filters good up to 6 GHz. Integrating provides very good temperature ranges (-55º to +150ºC) with impedances at 35 Ω, Abracon’s ACMF-03-350-T, 65 Ω ACMF-03-650-T, and at 90 Ω the supplier’s ACMF-04-900-T. Note that different chip makers’ parts may be optimized for non-50-Ω transmission paths. Note also the characteristics of the parts they cite in their reference designs.

Some functions like GPS may not want to share antennas and traces with other bands. Very specific front-end electronics can make a difference when locking onto a weak satellite signal. This allows specific antennas and independent placement of antennas for best reception. Specialty form-factor reduced circuits for GPS are also becoming more refined. Take a look, for example, at the RF SAW filters from Murata Electronics specifically for GPS receivers.


Similar to filters but serving to interface semiconductor transmitters to PCB and SMT antennas are baluns which basically turn single-ended transmission lines into differential drives for the antenna. Differential connections are key to minimizing common-mode noise and can significantly improve signal quality.

Baluns can integrate half a dozen discrete components into a single, small-sized surface-mounted part (Figure 2), and they are available for standard as well as general-purpose frequencies. They match impedances, have transmission line characteristics, and can require very specific groundplanes and PCB materials.

Figure 2: Monolithic baluns absorb several critical components, and compared to discrete solutions, feature better balanced matching of components.

Because a balun interfaces a transmit path directly to the antenna, it must also be able to handle the power levels associated with the maximum allowable energy in that band. Typically, a balun will be able to handle several watts of power.

Take for instance, the Johanson 2450 Series 2450BL15B100E balun for Wi-Fi and Bluetooth. It shares the same carrier frequencies, but easily passes both modulation techniques for Wi-Fi and Bluetooth and can handle up to 3 W with 50- or 100-Ω impedances.

For data links that require full duplex operations, the modern duplexers and diplexers can let one RF section be transmitting while the receive state is picking a different signal from the same antenna simultaneously (Figure 3). Components like the Avago ACMD-7402-BLK fit into a small 3-pin SMT CSP package or even an 0402 package like the thin-film TDK TFSD10055950-5102A2 or the TDK DPX165850DT-8017A1 diplexer for 2.4 and 5 GHz wireless LANs such as 802.11n.

Figure 3: Well-attenuated high-, low-, and band-pass filters inside monolithic duplexers and diplexers do a pretty good job of passing only the frequencies of interest while attenuating the rest.

Board games

With modern small-wavelength frequency bands in use, the PCB antenna has shown its effectiveness at creating a low-cost and fairly reliable transceiving antenna. While effective, many such designs require very specific groundplanes and keep out areas.

As a result, external antennas have made inroads by providing more flexibility in the antenna placement. This can improve performance by allowing transceiving points to be located further away from noisy digital lines, and even let embedded micrcontrollers select different antennas based on real-time RSSI information.

Several good-performing monolithic surface-mount antennas are readily available at low cost to allow such flexibility, including the TDK ANT016008LCD2442MA1 (Figure 4), which is a multiplayer ceramic dual-band (2.4 and 5 GHz) high-efficiency antenna packaged in an 0603-style surface-mount package. It has a height of only 0.4 mm, making it more than adequate for most compact ISM designs.

Figure 4: Multilayer ceramic antennas are so small and compact they often take up less space than a single discrete passive component.

Another interesting antenna option is the corner-mount unit which can present the antenna away from obstructions that center- or single-edge-mount solutions cannot.

For example, the Antenova A10464 has less than 0.5 mm height and can be corner mounted using a small groundplane on your PCB for GSM applications (Figure 5). Especially in handsets, the head can be a source of interference. Corner mounts allow clearer paths and more chance of line-of-sight connection with towers or satellites.

Figure 5: Corner-mount antennas can provide better line-of-sight characteristics in handheld applications like phones where close proximity to the head can attenuate signals.

In summary

Standard radios used in large numbers are still shrinking to allow more integration with other wired and wireless links. While highly integrated devices handle the active portions of the design quite well for already established and supported protocols, external discretes and passives often do better jobs than on-chip solutions in optimizing performance and cost. Part 2 of this series will discuss active radio elements in discrete and die form.

For more information, use the links provided to access product information pages on the Hotenda website.