RAFT Tuner Design for Mobile Phones

Transcription

1 RAFT Tuner Design for Mobile Phones Paratek Microwave Inc March 2009

2 1 RAFT General Description RAFT Theory of Operation Hardware Interface Software Requirements RAFT Design Process Example of Handset Antenna Measurements Designing to Match to the Transceiver: Using Load-Pull Measurements Vs Designing to 50 Ohms Tuner Design for Handsets Example of RAFT Tuner Performance in WEDGE Handset Example of Handset Radiated Performance Improvements Baseline Performance of Non-RAFT WEDGE Handset RAFT Radiated Performance of Single WEDGE Handset RAFT Radiated Performance Over Multiple Units Conclusions

3 1 RAFT General Description Paratek s Radio Antenna Frequency Tuner (RAFT) utilizes Passive Tunable Integrated Circuits (PTICs) to improve the performance of slim and small multi-band and multi-mode wireless handsets. RAFT optimizes the impedance match and power transfer between transceiver and antenna in mobile handsets. RAFT accomplishes this through the use of ParaScan -based tunable capacitors. RAFT frequency coverage is 800 MHz to 2200 MHz, including WCDMA and GSM bands. The tunable capacitors allow the impedance presented to the transceiver and the antenna to be optimized as a function of frequency, use case (e.g. handset held by the head or on a belt), battery life, linearity, or any other parameter. RAFT comprises an RF tuner that allows mobile handset RF performance to be tailored on a band-byband basis. It also includes control and voltage boost circuitry in the form of a custom ASIC that takes the nominal 3.5 V battery voltage and converts it to the voltage required to tune the capacitors. 1.1 RAFT Theory of Operation Paratek s Radio Antenna Frequency Tuner (RAFT) is a variable impedance matching network incorporating Passive Tunable Integrated Circuits (PTICs) to accomplish a better match to a handset antenna over a wide range of frequency bands and variable use cases (such as head, free space, and pocket). The PTICs in the RAFT circuit are biased to voltages between 2 and 18 volts by a High Voltage ASIC. The Paratek ASIC contains a 32 bit Serial Peripheral Interface (SPI) through which its three high-voltage digital-to-analog converters can be addressed and programmed. The RAFT tuner can have as many different tuning states as the application calls for. For instance, transmit and/or receive frequency bands could be split into sub-bands, and optimal tuning states could be established for each sub-band. Also, different use cases could be addressed by monitoring inputs that determine which physical configuration the handset is currently being used in. As an example, a flip phone has a detector that monitors its closed or open position. This input could be used to adjust the tuner to a different tuning state. 33

4 Similarly, the presence of audio in the earpiece speaker could be used to determine if the handset is positioned next to the users head. Slider phones having similar detectors indicating the physical configuration of the handset would allow the tuning network to be adjusted accordingly. Any of these different tuning states can be defined and utilized by the handset designer in order to meet specific performance objectives. Figure 1 RAFT Circuit Within Handset Block Diagram. Note RAFT can be implemented as a Module or Discrete Components 44

5 1.2 Hardware Interface The 32 bit SPI bus is defined in the specification sheet for Paratek s high-voltage PTIC controller (HV DAC, part number PTHV ), available upon request by visiting Figure 2 RAFT Conceptual Block Diagram 1.3 Software Requirements As described above, designers have great flexibility in determining the variety of tuning states needed to implement in a particular handset. A particular example is outlined here. For a particular handset having quadband GSM and UMTS WCDMA capabilities and only one physical configuration (candy bar shape factor for example), it may be desirable to define 18 different tuning states (see Table 1). In order to utilize these 18 tuning states, the handset software would need to identify which of the above frequency bands the radio was currently operating in, and which physical configuration it was in. Physical configuration could be determined by the presence of earpiece audio: If present, then the phone is assumed to be by the head. If not, it is assumed to be in free space. Frequency band information is available from the handset call processor. In order to use the 18 states optimally, the handset call processor would write an SPI command to the High Voltage ASIC before every transmission or reception. Specifically, in a GSM/GPRS or EDGE system, the appropriate message would be sent to the 55

6 ASIC prior to every transmit burst and prior to every receive time slot. For WCDMA in the UMTS band, the appropriate SPI command would be written to the ASIC prior to any voice or data call. These commands would send the appropriate DAC settings to set the PTICs to the correct DC voltages and thus accomplish the chosen tuning state for that particular band and use case. These correct voltages would be determined during the design process for the particular handset. Optimal tuning states will be defined by the specific DC voltages to be applied to the PTICs in the tuner. The example above was for a specific handset application. The transmit or receive bands could also be further segmented into additional tuner states if the designer chose to do so. In these cases the handset baseband controller would take into account the specific channel the radio was operating on, and set the tuner to the appropriate sub-band of channels (defined by the handset designer). In a different example where WCDMA is intended to operate in additional frequency bands beyond UMTS, RAFT can be designed to have duplex tuning states for each of those bands, which might be different than the simplex states defined in the above example for use by a TDD system such as GSM. Table 1 Example Tuning States for 3 PTIC Tuner in a Candy Bar Form Factor 66

7 It is desirable to have close co-operation between the handset manufacturer, Paratek and the chipset supplier in the development of software that controls the PTICs. The software can be developed in accordance with Paratek s Application Note, RAFT TM Control Overview of Software Requirements For CDMA2000, WCDMA and GSM/GPRS/EDGE. The block diagram in Figure 3 shows an example of a WEDGE handset and the SPI interface used to control the HVASIC that provides the bias to the PTICs. Figure 3 WEDGE Handset with RAFT Block Diagram showing SPI Control 77

8 2 RAFT Design Process The general RAFT design procedure is shown in Figure 4 below and involves measurement of the antenna performance as well as the transceiver performance. The antenna performance can be measured in a number of different use cases since the tuner s impedance state can be varied as a function of operating condition. This provides the handset designer with an unparalleled opportunity to maximize the phone s performance since the limitation of a single fixed matching network between the transceiver and the antenna has been removed. Once these parameters are measured a tuner design can be undertaken which can be realized in the handset as a module, discrete components or a combination of both. Figure 4 RAFT Design Flow Paratek have worked with several handset manufacturers and have consulted on the layout of PCBs to include RAFT circuits using a variety of approaches. These include using Paratek s HVASIC in the Wafer Level Chip Scale Package (WLCSP) and PTIC s in the QFN plastic package, WLCSP as well as modules that include the entire RF tuner. 88

9 3.1 Example of Handset Antenna Measurements Figure 5 shows the performance of the antenna under a range of use cases such as freespace, head (SAM), and hand. This data is the major input into the RAFT design procedure and is measured to the reference plane where the first RAFT component will be placed. This ensures that the impedance presented to the RAFT circuit is correct in terms of both phase and magnitude. It is also vital to have the RAFT circuit tuning elements placed as close to the antenna as possible to reduce the effects of loss and dispersion similar to the way a fixed matching network would be deployed. Free Space Light Grip Heavy Grip Head Hand Phantom Head freq (824.0MHz to 960.0MHz) Figure 5 Example Handset Antenna Impedance Plotted on the Smith Chart Low Bands 99

10 Free Space Light Grip Heavy Grip Head Hand Phantom Head freq (1.710GHz to 2.170GHz) Figure 6 Example Handset Antenna Impedance Plotted on the Smith Chart High Bands 3.2 Designing a Tuner with an Optimal Match to the Transceiver: Using Load-Pull Measurements Vs Designing to 50 Ohms A second set of impedances that need to be considered when designing a tunable antenna matching circuit are those presented by the transceiver. If the antenna impedances are considered as the load of the RAFT network, then the transceiver provides the source impedances. There are two approaches that can be taken with respect to the impedances presented by the transceiver. The first approach is the simplest, and follows the traditional approach of designing to 50 Ohms. This can yield acceptable results for PAs that have outputs close to this desired impedance. In the case of WCDMA PAs this is usually a good target because WCDMA PAs operate at lower output power levels and under more stringent linearity conditions. Designing to 50 Ohms also simplifies the tuner design task since there is only one set of impedances that are varying during the design. The second design approach, while being more complicated can offer significant performance improvements in the case of PA output impedances that are far from 50 Ohms. This approach uses 1010

11 load-pull measurement of the GSM and WCDMA power amplifiers to provide a set of source impedances that are used in the RAFT design. As previously mentioned this can be vital design information since the PAs frequently do not meet their 50 Ohms design target. If the antenna impedance is also far from the 50 Ohm goal there can be considerable mismatch between the transmitter and the antenna resulting in an enormous loss of radiated power. Figure 7 shows the automated load-pull results for an example handset at Channel 128 GSM850 band. The load-pull contours are measured at 0.2 db intervals and the impedances presented by the factory match are also plotted on the same Smith Chart for comparison. It can be clearly seen from this Smith Chart plot that the factory match is severely mismatched to the load-pull of the PA and this will result in a significant reduction in TRP. This reduction in TRP is clearly illustrated by the measurements of a baseline handset shown later in this case study. Figure 8 shows the same chart, but for Channel 124 in the GSM900 band. Here the factory match provides a good match to the load-pull and the resulting TRP is much higher than seen in the GSM850 band. A fixed antenna match is limited to this type of trade-off, but a variable match can overcome common problems such as these. Free Space Light Grip Heavy Grip Head Hand Phantom Head Figure 7 Example Handset Load-Pull Contours for Channel 128 GSM

12 Under WCDMA operating conditions the RAFT circuit has to be designed to provide a duplex match that optimizes both the transmit match and the receive match simultaneously since the transmit and receive switching that occurs in GSM does not take place in a WCDMA call because of full duplex operation. As mentioned earlier this match is usually much closer to 50 Ohms since the WCDMA PA is operated more linearly than a GSM PA and the optimal impedance for the receiver is also close to 50 Ohms. Free Space Light Grip Heavy Grip Head Hand Phantom Head Figure 8 Example Handset Load-Pull Contours for Channel 124 GSM 900 The optimal receive impedance for the Rx state in a GSM call is also set to be 50 Ohms since the optimal Rx impedance is close to this point. It is also worth noting that in operating the RAFT circuit in a GSM call the handset can select from a range of Tx impedances depending on the operating condition of the phone. For example, if the handset is far from the base-station then the RAFT can be programmed to match to a maximum transmit power condition, but if the handset is operating at a reduced power level then the RAFT can be programmed to another state that may have lower Tx 1212

13 power, but increased efficiency. The exact figure-of-merit used to choose the optimal match is limited only by available information from the handset and provides the handset designer with tremendous flexibility. The Load-Pull measurements are used along with the antenna S-Parameter measurements to design a tuner circuit that provides optimal matching with minimal loss and results in a higher performing handset. Additionally the RAFT circuit can be retuned for Tx/Rx operation in GSM, use-cases such as being close to the users head, and also tuned for specific groups of channels (sub-bands) within a given frequency band, giving the designer previously unavailable options. 3.3 Tuner Design for Handsets The RAFT tuner design only employs a small number of components and is similar to many fixed match circuits used in mobile phones. The circuit can be considered in the same way as the universally employed fixed match except Paratek s tunable capacitors are used rather than fixed capacitors. Figure 9 shows two different schematics that could be employed as a tuner (Pi and Tee Configurations) and Figure 10 shows a photograph of a three PTIC tuner circuit as deployed on a handset PCB. RF-IN L1 PTIC1 PTIC3 L3 ANT PTIC2 L2 Figure 9a Example RAFT Tuner Schematics 1313

14 RF-IN PTIC2 ANT L2 PTIC1 L1 PTIC3 L3 Figure 9b Example RAFT Tuner Schematics Figure 10 RAFT Circuit Deployed on a Handset PCB 1414

15 3.4 Example of RAFT Tuner Performance in WEDGE Handset Figure 11 shows the typical performance of the RAFT tuner when embedded in the handset, terminated with the antenna S-Parameters and overlaid on load-pull contours. It can be seen from the plot that the tuner offers good coverage of the optimal power contours providing good TRP. Similar coverage is obtained in other frequency bands. Paratek provides a Large-Signal model of the PTIC, allowing designers the freedom to develop their own tunable matching network, which, when coupled with their antenna design, can result in truly effective handset designs. Figure 11 RAFT Circuit Coverage in a WEDGE Handset at 1980 MHz 4 Example of Handset Radiated Performance Improvements Paratek has made measurements on baseline (Non-RAFT) handsets and RAFT-enabled handsets using our own Satimo Radiation Chamber as well as an external CTIA-approved facility. These measurements are used to allow an accurate determination of the radiated performance of handset with 1515

16 and without the RAFT circuit. This comparison between phones with and without RAFT yields an accurate picture of the benefits employing a Paratek RAFT brings. 4.1 Baseline Performance of Non-RAFT WEDGE Handset Figure 12 shows the baseline TRP performance for two handsets and Figure 13 shows the TIS performance of one baseline unit TRP (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. Figure 12 Baseline Freespace TRP Performance of To Handsets without RAFT Circuit Examining Figure 12, it can be clearly seen that while the phone has good performance in some bands such as GSM900 and WCDMA Band I, the phone is suffering from the mismatch problem illustrated in Figure 7 earlier. Also, the results in Figure 12 show that the performance of the phone in GSM900 Channel 124 is as expected, since the factory fitted antenna matching network does a good job of matching to the PA load-pull contours. While antenna efficiency is an important contributor to the 1616

17 TRP/TIS performance, the mismatch seen here in the GSM850 band is significantly degrading the phone performance by several db TIS (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. Baseline Figure 13 Baseline Freespace TIS Performance of WEDGE Handset without RAFT Circuit 4.2 RAFT Radiated Performance of Single WEDGE Handset Measurements have been made by Paratek at a CTIA approved facility which allowed an independent comparison between a baseline unit, as shipped by a manufacturer, and a RAFT-enabled phone. The results in Figure 14 clearly show the benefit of a tunable antenna matching network with significant improvements being made in the GSM850 band of up to 10 db. In addition to this improvement in the GSM850, band there are also improvements in GSM900, PCS and WCDMA bands with minimal change to the channels that were already performing well from a radiated perspective. 1717

18 TRP (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. Baseline (dbm) RAFT (dbm) Figure 14 Performance of a WEDGE Handset with RAFT Compared to Baseline Handset TIS (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. Baseline (dbm) RAFT (dbm) Figure 15 Performance of WEDGE Handset with RAFT Compared to Baseline Handset 1818

19 4.3 RAFT Radiated Performance Over Multiple Units Paratek has made measurements on multiple handsets of the same type that have all been assembled with the same RAFT schematic to allow comparison against phones that do not have RAFT circuitry installed and also to examine unit to unit variation. Measurements have been made at both an independent facility, as well as at Paratek s own radiation chamber shown in Figure 16 below. Figure 16 Handset Measurement in Paratek s Satimo Chamber 1919

20 The TRP results for four different handsets are shown in Figure 17 and all the phones show similar performance improvements clearly demonstrating the benefit of Paratek s tunable solution TRP (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. RAFT Unit 7 RAFT Unit 4 RAFT Unit 8 RAFT Unit 20 Figure 17 TRP Handset Measurements for Four RAFT Units The measurements also demonstrate good uniformity from handset to handset with only a small variation in TRP between the units. Figure 18 shows the same group of RAFT handsets compared against an average of the two baseline handsets clearly illustrating the improvements that a RAFT handset has over the baseline handset. Finally, in Figure 19, the average TRP of the RAFT units is compared to the average TRP of the baseline handsets, showing excellent agreement with the previous measurements of a single handset (Figure 14). TIS measurements have also been made, but due to the length of time required for TIS characterization and the relatively large number of handsets, the TIS chart shown in Figure 20 is a composite of several measurements on several handsets. The chart represents the performance of typical baseline TIS and typical RAFT handset TIS performance. 2020

21 TRP (dbm) GSM GSM DCS PCS WCDMA I Channel No. Baseline Average RAFT Unit 7 RAFT Unit 4 RAFT Unit 8 RAFT Unit 20 Figure 18 TRP Handset Measurements for Four RAFT Units Compared Against an Average of Two Baselines (No RAFT) TRP (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. Baseline Average RAFT Average (dbm) Figure 19 Average TRP of RAFT Handsets Vs Average TRP of Baseline Units 2121

22 TIS (dbm) DCS GSM850 GSM900 PCS WCDMA I Channel No. Baseline RAFT (dbm) Figure 20 TIS Comparison between Baseline Handsets and RAFT Handsets 5 Conclusions The inclusion of Paratek s PTICs and HVASIC into multi-band and multi-mode mobile phones can bring unparalleled performance enhancements. The complexity of modern handsets and the broad frequency coverage needed means that the traditional fixed match approach is no longer adequate. The use of a PTIC-based tunable solution as demonstrated by Paratek means fewer performance trade-offs, more robust radiated performance, faster design cycles and no-compromise Industrial Design. 2222