Wireless Systems Design

Automatic Antenna Tuning Saves Power

In short-range applications, a high Q antenna will slash cost and power consumption at the targeted output-power ranges.

In recent years, the world has welcomed an increasing number of short-range, license-free wireless applications that operate in the Industrial Scientific Medical (ISM) band. Among these applications are keyless entry, tire-pressure monitoring, door openers, wireless headphones, wireless mice and keyboards, and wireless local-area networks (WLANs). Of all of these applications, however, only a few very-low-cost solutions can be found with minimal external components. A discrete monopole is usually used instead of on-board printed antennas, which increases the cost and dimension of the final product.1,2

Typically, the ICs in these solutions operate in mobile handheld devices that are powered by small batteries. As a result, the transmitter's power consumption is critical. In most applications, the desired operating range is in the order of 10 m. The required equivalent radiated power (ERP) is lower than −25 dBm even if a relatively noisy receiver (i.e., sensitivity worse than −90 dBm) is applied.

The power that the antenna requires to achieve this ERP depends on the antenna gain and the loss of the impedance matching network (if necessary). In the case of low-impedance antennas, a higher driver current is needed with simultaneous lower-voltage swing. With high-impedance antennas, lower driver current is enough. High-impedance output stages can be realized with open-collector-type circuits. In these circuits, the output voltage on inductive loads is comparable to the supply voltage. The efficiency can therefore be much better.

In the table, a comparison between a small monopole, a quarter-wavelength monopole, and a small loop antenna is shown at 434 MHz. Due to its much higher efficiency (~90%), the monopole antenna has approximately 10 dB higher gain than the small loop (~10%). The tuning inductance that's required by a small monopole increases the cost, however. It also degrades the efficiency dramatically. In the case of an inductance Q of 100 and 400, respectively, it decreases to 5% and 18% (l = 0.05l).3 In addition, the tuned-antenna impedance is close to the driver's output impedance. (Usually, this driver is the emitter follower.) As a result, only half of the driver power goes toward the antenna.

With loop antennas, the chip's capacitive output impedance can be used to tune the antenna. The Q of the chip's output circuitry is usually about half of the Q of the loop antenna. So only 33% of the driver's output current goes to the antenna. (Normally, that driver is the open collector.)

Further improvement can be achieved by applying tapped loop antennas. The antenna Q will then be approximately equal to or less than the Q of the chip output circuitry. As a result, the higher part of the driver's current (~50% or more) flows to the antenna. Due to its higher aperture size, the tapped loop antenna also delivers a higher gain.

The second row of Table 1 shows that the gain of a loop antenna can be even better than the gain of the small monopole in real conditions. The sixth row shows that the necessary current to achieve −25 dBm ERP with a small monopole is an order of magnitude higher. The loop and the one-quarter monopole antenna require about the same current. But a loop antenna is less expensive, smaller, and more insensitive to the so-called hand effect. A loop antenna interacts with the H field, which is less affected by close proximity to the human body compared to the E field.3 With the proper orientation—keeping the antenna plane perpendicular to the body surface—an increase in antenna gain can be observed.

Due to its high Q, the loop antenna requires an automatic-tuning mechanism. This mechanism will keep the power on maximum during various environmental circumstances (temperature, detuning due to the hand effect, etc.). It also will compensate for the technological spreading.

This article describes an automatic-antenna-tuning solution for a transmitter chip working at the 315-, 434-, 868-, and 915-MHz ISM bands. This transmitter's high-speed and high-resolution synthesizer determines the output frequency with the accuracy of the on-chip, crystal-controlled reference (FIG. 1). Thanks to its high resolution, multiple channels can be used in any of the bands. In addition, fast switching allows the transmitter to perform FSK modulation with the phase-locked loop (PLL) by itself. It doesn't require any extra modulation circuitry. The deviation is selectable to accommodate various bandwidth, data-rate, and crystal-tolerance requirements.

The power amplifier with the incorporated antenna-tuning circuit has an open-collector differential output (denoted by RFP and RFN). This output can directly drive a loop antenna with a programmable output.1 The transmitter supports either an EEPROM or microcontroller operation mode. In the EEPROM mode, all of the necessary control words and data that will be transmitted are read out from an EEPROM by pressing a button. That button is connected to any of the wake-up inputs denoted by SW1-4.

FIGURE 2 shows the block diagram of the antenna-tuning loop. The main goal of the operation is to adjust the output block's resonance frequency to the frequency of the transmitted signal. The output block's resonance frequency includes the loop antenna, capacitance bank, package and bonding-pad parasitics, and circuit parasitics. The output block can be modeled by a parallel RLC resonant circuit. At resonance, one can therefore observe a 180° phase shift between the driver's collector current and the differential voltage over the output collectors. The phase shift between the collector current and base voltage is constant. As a result, the base voltage's phase is compared to the phase of the collector current. The two phase shifters are adjusted to have a 90° difference in their phase characteristics in a wide band (300 to 1000 MHz).

A comparator monitors the DC term of the error voltage at the mixer output. To reduce the error voltage, the state of the 4-b control word for the capacitance bank is varied. The whole process takes 6.4 µs. It operates continuously to correct environmental changes (hand effects, etc.). To avoid oscillation, the difference between the up-counting and down-counting comparator level must be higher than the largest error voltage step that exists between any neighborhood-capacitance-bank states. Usually, antenna dimensions are designed to have resonance near the operation frequency with a counter state of 7. The counter is therefore adjusted to state 7 at every turn-on or during reset events. The phase shifter that's connected to the collectors contains a limiter at the input. This limiter eliminates the error-voltage variations that arise from the output-voltage-level changes. Either detuning or an intentional change of RF power causes such changes.

Extremely low output-power levels are reached when power is below the limiter's operation range. Typically, this drop occurs at strong detuning. To avoid chaotic operation in case of such low output-power levels, a level detector switches off the tuning and sets the counter to state 7. The phase shifters and the limiter are integrated within the Gilbert cell mixer.

At this stage, it's essential to eliminate any offset and asymmetry in the parasitic layout capacitances. A symmetrical layout structure with cross-connected transistor pairs is therefore necessary. FIGURE 3 shows the analog parts of the antenna-tuning circuitry together with the driver and bonding pads. The total occupied area is 0.36 mm2. On the left middle side, one can see the 1-2-4-8 ratio of the capacitance bank. The capacitance bank is located within the two output pads.

The tuning loop's operation was measured at all bands at different supply-voltage and output-power levels. The measurements were performed by an HP E4402B spectrum analyzer and an 85024A high-frequency probe. In addition, the internal 10-MHz oscillator was replaced by an external generator in order to perform wideband measurements. The output frequency of the PLL was varied by changing the frequency of the generator.

FIGURE 4 shows the tuning curve measured with an antenna that is 5 3 7.2 mm. It has a strip width of 1 mm on a 0.5-mm-thick FR4 substrate. For the forward and reverse curve, the frequency was started from the lower and upper edge of the investigated band, respectively. The chip successfully found the state of maximum output power except in the extremely detuned cases (above 940 MHz, where resonance cannot be achieved even at state 0). The 16 different state-of-capacitance banks can be recognized, as the automatic tuning always adjusts the resonance to the operation frequency.

This chip easily covers the 868- and 915-MHz ISM bands with the same antenna. It can be assumed that multiband operation also is possible with the 868- and 915-MHz ISM bands. (The same dual-band operation is possible for the 315-/433-MHz ISM bands.)

FIGURE 5 shows a possible 434-MHz application board layout that's controlled by a microcontroller. This layout achieves the minimal bill-of-materials (BOM) requirement allowed by the chip. (The switcher and the surface-mount-diode (SMD) resistor in the antenna are optional, as they are only for test purposes.)

For an EEPROM-mode transmitter to work in either ISM band with a properly sized antenna, it needs several components (FIG. 6). The required external components are an EEPROM, a crystal, switches, LED+resistor (optional), and three SMD capacitors. The result is a low-cost transmitter.

As shown, a fully integrated automatic-antenna-tuning circuit can be successfully achieved. The demonstrated circuit maximizes the radiated power from the connected small loop antenna, thereby improving the power efficiency of the transmitter chip. The circuit can cover a wide range of frequencies. It allows the use of dual-band antennas covering the 868-/915-MHz or the 315-/433-MHz ISM bands.

Acknowledgments:The author expresses his special thanks to all of his colleagues at Integration Hungary Ltd. and Integration Associates, Inc. for their support. In particular, he'd like to thank Péter Onódy, Wayne Holcombe, and Gábor Tóth for their valuable comments. In addition, Gábor Szijártó and Zsolt Benedek deserve special recognition for their efforts to prepare the chip layout.

REFERENCES:

  1. Jerng, Albert et al, "Integrated CMOS Transceivers Using Single-Conversion Standard IF or Low IF RX for Digital Narrowband Cordless System," RFIC 2002, Seattle, WA, June 2002.
  2. Tunheim, Svein Anders, "Implementation of CMOS Low Cost and Low Power RF-ICs," Wireless Systems Design Conference and Expo 2002, San Jose, CA, February 25-28, 2002.
  3. Fujimoto, K. et al, Small Antennas, Research Studies Press Ltd., Letchworth, Hertfordshire, England 1987, ISBN 0863800483.
Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish