Activate Your LTE Investment with Tunable RF
Activate Your LTE Investment with Tunable RF
As LTE deployments begin to pick up momentum, operators and handset manufacturers should remember that a 4G network alone is not a panacea for their 3G performance-related malaise. In fact, realizing the full promise of LTE requires a collected effort of speed- and reliability-enhancing solutions, an ongoing series of booster shots to fend off congestion caused by excessive network traffic, increased data use, form factor constraints and beyond.
In general, higher data rates utilize complex modulation schemes that impose more stringent signaling requirements. To make matters more complicated, global LTE implementations will utilize more frequency bands than 3G, with 7-band handsets considered a baseline and 13+ bands desirable for true global roaming. And perhaps more than anything else, antenna performance limitations threaten to severely hobble the speed and versatility service providers are counting on for LTE to deliver its promised ROI.
Tunable RF enables better LTE performance by making a physically smaller antenna appear larger to the network. That is, by attaching tunable RF devices to the antenna element itself, engineers can design a physically smaller antenna and have it operate as if it were larger. In this manner, tunable RF addresses the existing space constraints the industry knows so well.
Also, by covering more frequency ranges with a single antenna, tuning behavior automatically reduces the overall number of antennas required for a handset to viably perform. This is significant in light of the trend toward MIMO, where up to four antennas are assigned different functions. Tunable RF allows transmission and reception with maximum efficiency—and with less interference from other sources such as head and hand position.
Anatomy of a high performance Tunable RF Device
Out of the handful of methods entering the market to compensate for the antenna problem, none accomplishes it as effectively as dynamically tunable radio frequency micro-electro-mechanical systems (RF-MEMS) technology.
The leading tunable RF device is a digital capacitor array utilizing RF-MEMS integrated with electronic circuitry on a single silicon die.
RF-MEMS capacitors are mechanical devices, fabricated on the surface of a silicon wafer, consisting of two metal plates that move together electrostatically in response to an applied voltage. A dielectric layer is fabricated between the plates to form the capacitor. In contrast to a solid-state switch, which uses electron flow through a semiconducting substrate, in the RF-MEMS device, electrons flow entirely in metal resulting in very low losses and ultra-linear operation.
Since the RF-MEMS capacitors are integrated onto a CMOS wafer, all control for the MEMS device is physically located on the same die. This saves routing and space as well as minimizes signal coupling to or from control lines. This is particularly important as high voltage (roughly 35V DC) is needed to actuate the devices. Since the RF-MEMS capacitors are located on the same CMOS die, the required voltage is generated on-chip with an integrated charge pump so that the only external supply voltage required is 2.7-3.3V. Additionally, all device drivers can be located on-board; all capacitor settings are selectable from registers written either through an industry-standard SPI or MIPI RFFE serial interface.
Figure 1. RF-MEMS device cross-section.
The mechanical structure of an RF-MEMS device means a relatively low frequency mechanical resonance will be observed—around 60kHz. This is due to the length of the beam resonating at a half-wavelength for the driving signal. When the MEMS device is closed, the resonance is less pronounced, and is shifted up in frequency to several Megahertz. The low mechanical frequencies contribute to the outstanding linearity as the MEMS device cannot respond directly to signal variation in the gigahertz range.
With a variable capacitor array, an important parameter is the “on/off” ratio of each individual capacitor in the array, and thus the array as a whole. When the MEMS device is “up,” or not in contact, the capacitor is in the minimum capacitance state, or “Cmin”. Likewise, when the capacitor is actuated and in the “closed” position, the capacitor is in the maximum capacitance state, or “Cmax”. The capacitance ratio is defined as shown in Equation (1).
( Equation 1)
Each capacitor in an array has a model similar to Figure 2. In this model, C1 and C2 represent parasitic shunt capacitance to ground generally caused by the assembly environment and silicon substrate. Cseries represents the digital capacitor, adjustable between Cmin and Cmax.
Figure 2. MEMS capacitor model
C1 and C2 are not equal as the layout of the MEMS device on-chip affects the value of these parasitic capacitances.
If the device is configured in series, then Cratio is typically 15. Note that there will also be some parasitic shunt capacitance to ground, which will vary depending upon the size of the capacitor, typically 5-15% of Cmax.
If the device is configured as a shunt device, i.e. Port B is connected to ground, then one of the parasitic capacitances, C1 , is in parallel with the shunt digital capacitor, increasing the effective Cmin. In this case, Cratio is typically 7.
With regard to the Quality Factor of the RF-MEMS capacitor, the significantly lower resistance of the metal beam provides a key advantage: lower loss. This lower loss is generally represented in specifications as a “Q” factor. Q is simply the ratio of the reactive impedance to the real impedance shown in Equation (2), where ESR is the “Equivalent Series Resistance” of the capacitor.
Naturally, minimizing ESR for a given C will increase Q. The metal traces on the RF-MEMS beams provide this low ESR, especially in comparison to other technologies. Q measurements for this RF-MEMS technology are typically over 200 at 1GHz measured on-wafer. By comparison, Q values for typical CMOS electrical devices at the same frequency are typically less than 30.
Linearity of devices in the RF Front-End of mobile handsets is typically specified as a Two-Tone Input Third-Order Intercept Point (IIP3). RF-MEMS devices are typically very linear, but are somewhat sensitive to the spacing of the two tones. The two closely spaced tones combine to create a voltage envelope, which peaks at the sum of the voltages of the individual tones and varies at a low beat frequency located at the difference between the two tones. If this beat frequency is below or near the mechanical resonance of the RF-MEMS device, then a higher non-linearity will be observed. As previously mentioned, a mechanical resonance occurs in the 50-100kHz range. For tone spacing in this range, the IIP3 for the MEMS devices is around +70dBm. With wider tone spacing, the linearity improves to over +80dBm.
Also, note that if the die is not properly grounded, modulation may occur between the RF traces on the MEMS devices and the CMOS circuitry under the buried shield. This modulation can increase the non-linearity, so it is important to assure the die is properly grounded.
Figure of Merit
In order to monitor and compare the state of the art for tunable capacitors, a general Figure of Merit (FOM) is used. This FOM provides a quick assessment of the capability of any given tunable capacitor technology examining tradeoffs the areas of loss, capacitance ratio, power handling, and cost (die area).
CR is Capacitance Ratio:
V2 is Maximum RMS voltage across capacitor
Die Area is the die area required for the given capacitance
Ron is the total series resistance in on-state.
In addition to the usual reliability challenges for any semiconductor device, there are two additional categories of reliability issues that apply to contacting MEMS devices:
• Stiction, whereby the two capacitor plates form a bond and will not release, and
• Wear-out, where the parameters of the device change over time due to repeated use.
Stiction is typically a totally random phenomenon, and is controlled by designing the MEMS device to prevent intimate contact of metal-metal areas and/or high electric fields at dielectric surfaces. The best devices on the market have been carefully designed to prevent actuator contact, so the only areas in contact are the capacitor regions. As such, stiction is prevented from occurring.
Wear-out is the normal method of failure and is controlled by properly designing the mechanical MEMS beams and contact regions. Full product-level arrays containing tens of RF-MEMS capacitor devices will survive more than 150 x 106 cycles where a cycle is defined as a state change programmed by a customer through an SPI or RFFE interface.
MEMS devices are actuated by a high-level DC voltage, generated by an integrated charge pump. When this voltage is applied across the actuator terminals, which are connected to the capacitor plates, the plates are drawn together due to the electrostatic force. This is how the capacitor is intended to be switched from Cmin to Cmax.
An RF signal is also composed of a time-varying voltage. This voltage oscillates at the RF frequency, which is typically well above the self-resonant frequency of the MEMS device. Therefore, the RF voltage does not directly modulate the MEMS device. However, the actuation is driven by the square of the voltage which has components at DC as well as at the 2nd harmonic. The measure of this effective voltage at DC is known as the rms (root-mean-square) voltage (see Figure 3). The rms voltage of the RF signal, if too high, can cause a MEMS device to “self-actuate” and thus be in the high capacitance state even if programmed to be in the low capacitance state. Reaching such voltages in cellular front ends requires high powers, generally above 36dBm, and high impedance resonant situations such as may be found in filters or certain phases at high mismatch conditions. Thus a capacitor must be specified in terms of maximum RF rms voltage across the capacitor terminals.
The power-voltage relationship is given by Equation 4, where Z is the characteristic impedance of the system (generally 50Ω), and Vpeak is the peak RF voltage as shown in Figure 3. The rms voltage is given by Equation 5.
Figure 3. Vrms is DC voltage resulting from RF signal.
(if you are going to use this you should also have a zero baseline, and –Vpeak).
and for a 50Ω system, Vrms is
Self actuation does not damage a device. Thus, the absolute maximum conditions may allow operation above self-actuation—again, depending upon the circuit configuration and the tolerance allowed for specification deviation under “absolute maximum” conditions.
While the RF-MEMS device is closed using electrostatic force applied with a high-voltage actuator, the device opens upon removal of the actuation voltage. Once the electrostatic force is removed, the spring force in the beams restores the RF-MEMS device to its open state. This spring force is typically lower than the electrostatic force, for many reasons.
The lower spring-based restoring force means that the device, once closed, will only open again when the actuation voltage is reduced below the “release voltage”. The release voltage for the RF-MEMS capacitors is significantly below the actuation voltage, at approximately 8V. In normal operation, this is not an issue because the integrated capacitor drivers remove the actuation voltage completely to open the capacitor.
If the rms voltage component of the RF signal across a MEMS capacitor exceeds the release voltage, it will prevent actuated MEMS devices from opening. This limits the RF power that can be present when the capacitors are switched to the low state. Again, the power level at which this becomes an issue depends upon the circuit configuration and the load impedance (VSWR); thus, the hot tuning limits must be specified in terms of rms release voltage until the circuit configuration is known.
In normal communications systems, the tuner is typically re-configured during a pause in the transmit data stream. This is referred to either as “compressed mode” for WCDMA, or DTX for general communications. Also, many systems that require hot tuning have lower rms voltage operation. Thus, hot tuning over the full power range will not generally be required.
Feed Point Tuner
Many commercial communications systems can benefit from a high-performance tunable RF device. Mobile phones and portable tablets are two platforms which experience constraints on the antenna. Size constraints limit the ability of the antenna designer to match the antenna over each frequency band of operation to 50Ω. More frequency bands are being added to each mobile platform, which exacerbates this problem. The antenna designer is forced to trade off radiation efficiency of the antenna in order to achieve matching in each band of operation.
Tunable RF devices can be applied to create a Feed Point Tuner, allowing the antenna to be optimized in each band for maximum radiated efficiency rather than 50Ω match. The tuner can then be tuned for each band of operation to match the transceiver to the antenna load. Current WiSpry tuner products can tune over a 19:1 VSWR, and over bands from 824 to 2170 MHz using a proprietary broadband circuit configuration.
Current WiSpry tuner products are controlled in an open-loop fashion. In this configuration, control is provided by one of the processors (typically but not necessarily the baseband processor) in the cellular chip-set using one of the industry standard digital bus formats. For operation in a next generation, closed-loop tuning application, a power sensor and feedback controller would be built into the local loop for the tuner. In this case, the sensor must also detect when the power is below the hot tuning level and change configuration at that time.
Antenna Load Tuner
An antenna load tuner uses the tunable RF-MEMS capacitor elements to directly change the antenna resonance by applying a variable load directly into the antenna structure, causing the antenna response to vary with tuning setting. This is another method of achieving a tradeoff between radiation efficiency and match over a number of frequency bands.
Tunable RF devices can also be used in a resonant circuit configuration to provide either a notch or a pass-through response at specific frequencies. These responses are tunable using the RF-MEMS capacitors, and can provide a very well controlled, digitally tunable RF filter function.
Adaptive PA Matching
A power amplifier (PA) match can be adjusted with RF-MEMS devices, which could allow a PA to be optimized for various modes of operation (linear vs. non-linear), power levels, and frequencies. Most commercial PAs use conventional ladder network matching on the output for efficiency reasons; thus, the RF-MEMS capacitors can provide adjustable capacitor elements while the inductance is realized through conventional, non-adjustable means.
The numerous benefits described above add up to good health all around for critical stakeholders in the mobile ecosystem. Operators get lower infrastructure costs, increased network bandwidth, improved availability, regional platform programmability and potentially decreased churn resulting from higher quality service and increased customer satisfaction. Mobile handset manufacturers can realize up multiple dBs in performance gain, lower BOM costs and complexity, smaller thinner form factors, reduced SKUs and a faster time to market. And subscribers get fewer dropped calls, the possibility of up to 35 percent more battery life, more features for less, and the ability to talk anywhere and anytime. With advantages like these, tunable RF should be a mainstay of LTE’s core diet.
For more WiSpry RF Components informaiton:firstname.lastname@example.org