Fourth-generation (4G) wireless communications systems such as WiMAX and the Third Generation Partnership Project (3GPP) Long Term Evolution (LTE) platform are geared for high performance in terms of data rates for mobile and fixed wireless system users. To deliver that type of performance, they will need enhanced high-frequency components. One such component is presented here: a symmetrical absorptive-mode single-pole, double-throw switch for WiMAX based on PIN diode technology. The switch features a series-shunt topology for broadband operation over the 2300-to-2700- MHz WiMAX frequency band. It exhibits insertion loss of less than 0.75 dB, return loss of 15 dB, and isolation of better than 30 dB. The switch can handle significant levels of RF input power without exhibiting amplitude compression.

Emerging wireless communications systems, including WiMAX, are expected to provide a variety of services, from high-quality voice to high-definition video, through highdata- rate wireless channels. The high data rate requires broad frequency bands. For full functionality, these broadband wireless channels must be connected to broadband fixed networks, such as the Internet and local area networks (LANs).1

Although LTE has garnered a great deal of attention in recent months because of the ongoing deployment of 4G cellular networks in various regions, WiMAX continues to be a viable broadband technology as well for high-speed fixed and mobile Internet access and other wireless services. In fact, demand for wireless services via WiMAX continues to grow around the world, placing great demands on high-frequency component manufacturers to accelerate their product developments with shorter times to market. There is an increasing demand for handling higher-power signals with low insertion loss, and for high-performance switching functions in WiMAX systems, and PIN diodes offer a viable technology for providing the required performance for WiMAX base-station applications.2

An important component in WiMAX base stations is the SPDT antenna switch. Because base-station manufacturers are being asked to make their equipment smaller and less expensive, the component manufacturers feel the pressure to produce products that are smaller and lower in cost. One of the key switching functions in a WiMAX base station is the selection of RF signals between the receiver or transmitter path,3 and this function can be achieved by the use of an SPDT switch. For this application, a transmit/receive (Tx/Rx) switch must exhibit low insertion loss and high isolation in order to maintain high signal quality with minimal performance degradation. Such a switch candidate will be presented in this report of a symmetrical absorptivemodel SPDT WiMAX base-station switch based on a silicon PIN diode structure. This report will also detail a novel structure with inductive- capacitive (LC) matching sections to provide tunable broadband characteristics.4

A positive-intrinsicnegative (PIN) diode is a semiconductor that can be represented as a current-controlled resistance. It can be used to construct an electronic switching element, as it is easily integrated with planar circuitry and capable of high-speed operation.5 With no bias is applied, it behaves like a capacitance. When bias is applied, a PIN diode acts like an inductor.5

Compared to a gallium arsenide (GaAs) PIN diode or metal-epitaxialsemiconductor field-effect transistor (MESFET), a silicon PIN diode is less susceptible to damage from electrostatic discharge (ESD) effects. In addition, a silicon PIN diode has the high breakdown voltage (50 V) and large junction area needed to handle the high power levels of WiMAX base-station requirements. The performance of a PIN diode depends primarily on the chip geometry and the nature of the semiconductor material, particularly in the intrinsic (I) region. A PIN diode switch is more robust than a MESFET switch since it is capable of surviving switching operations in the presence of high levels of RF power, a process known as "hot switching." A MESFET switch can be damaged by the transition through a resistance region where significant power is dissipated.3

This, for high-power base-station applications, a silicon PIN diode switch is to be preferred to a GaAs MESFET switch, which is more suitable for lower-power battery-driven devices such as mobile handsets, which operate at lower RF power levels and require lower power consumption. For frequencies above 5 to 30 MHz, a silicon PIN diode switch will generate less distortion and achieve a higher third-order intercept point than a GaAs MESFET switch.4-6 When lower distortion characteristics are essential to a base-station application; a silicon PIN diode switch is to be preferred to a GaAs MESFET switch.

Conventional SPDT switches can present performance problems for a WiMAX base station application in terms of high insertion loss, inadequate isolation between ports, narrow operating-frequency range, high current consumption, low power-handling capability, and power linearity. But by using a design in which three PIN diodes are placed in series (series- series-series), as shown in Fig. 1, the port-to-port isolation can be improved to more than 25 dB especially when operating with high transmit power levels. Unfortunately, the penalty for achieving such high isolation with multiple PIN diodes is poor insertion loss. During forward bias, the transmit loss equals the sum of the RF resistances (D1, D3, and D5), resulting in relatively high insertion loss of more than 1.5 dB.

Measurements on the experimental PIN diode switch were performed with the S-parameter test setup shown in Fig. 2, which includes a model E5270A eight-slot parametric measurement mainframe and a model E5071B ENA vector network analyzer (VNA) from Agilent Technologies. The E5270A houses as many as four high-power source-monitor units (HPSMUs) or eight medium-power source-monitor units (MPSMUs), forming a test solution with as many as eight discrete test channels. During transmit mode, channel 2 of the E5270A is set +5 VDC bias and channel 4 is set -5 VDC. During receive mode, channel 2 is set to -5 VDC and channel 4 is set to +5 VDC bias.

Model E5071B is a four-port VNA that provides accurate measurements of RF component performance. The key to accurate measurements lies in the calibration of the system at every frequency. This is accomplished by applying precision loads, short, and open circuits to the test ports during a calibration measurement sequence. The magnitude and phase of the incident and reflected waves are recorded by an internal computer, and this data is used to correct for errors incurred during measurements on a device under test (DUT).

Based on the architecture of Fig. 1, an experimental absorptive SPDT PIN diode switch was fabricated on RO4003C laminate from Rogers Corp. It is a non-PTFE, glass-reinforced hydrocarbon/ ceramic laminate with dielectric constant of 3.38 at 10 GHz. The switch is shown in Fig. 3.

The results of measurements for transmit and receive return loss for the experimental PIN diode switch are shown in Table 1. Compared to the switch's return loss (S33) in receive mode, the measured return loss in transmit mode (S11) is closer to the simulated values. The measured S22 and S33 return-loss results in receive mode were better than the performance predicted by means of simulations. The measured return loss in transmit mode was worse than the transmit-mode return loss predicted by the simulations. The S33 transmitmode return-loss performance and the S11 receive-mode return-loss performance levels are both better than 15 dB. Table 1 shows that only the S22 return-loss (transmit-mode) performance at 2.5 and 2.7 GHz is marginal below the target specification of 20 dB. There is, however, a large difference between the measured and simulated results, especially for the S33 receivemode return-loss performance at 2.7 GHz. But on a linear scale, both the measured and simulated S33 receivemode values are close to 52 O.

The lack of correlation between the measured and simulated return-loss values for transmit and receive modes are due to the following:
SMA connector mount quality and connector placement result in reflected signals that cause mismatches;
variations in the lumped-element (inductors and capacitors) devices and coupling signals can cause performance variations; and
the quality of solder connections can be variable.

Continue on page 2

Page Title

Table 2 shows that for a high-power 20-W SPDT switch, an 0.08-dB difference in insertion loss results in a major impact on the total power (in W) transmitted by the base-station antenna. Based on an insertion loss of 0.64 dB at 2.7 GHz, the total power transmitted by the antenna is 17.22 W. For a measured insertion loss of 0.72 dB at 2.7 GHz, the total power transmitted by the antenna is 16.90 W. This is a 0.32-W different in transmitted power for a difference in insertion loss of 0.08 dB between the measured and simulated results. At higher operating frequencies, losses due to printedcircuit- board (PCB) materials and passive components will be higher than at lower operating frequencies.

The resonant circuit shown in Fig. 4 was added to the SPDT switch circuit to improve circuit isolation. Table 3 shows the measured versus simulated isolation performance in both transmit and receive modes. In both modes, the measured S31 isolation is seen to undergo a massive shift of 100 MHz toward the lower frequency direction. Because of this, the value of the inductor used to create a parallel resonant circuit with the series PIN diode must be reduced to shift the resonant frequency from 2.445 GHz to 2.585 GHz (receive-mode S31 performance). As Table 3 shows, the measured S32 ( transmitmode) performance is closely correlated with the simulated results. Among the four sets of measured isolation values, only the receive-mode S31 isolation shows a unique trend with the curve moving toward the higher-frequency direction. This may be caused by some coupling effect or by poor grounding in the receive path circuitry.

As Table 3 shows, all the measured results for isolation, for either transmit or receive mode, are better than the simulated results except for the transmit-mode S31 performance (in dB) at 2.7 GHz, the S32 performance (in dB) at 2.3 and 2.5 GHz, and the receive-mode S31 performance at 2.7 GHz. The difference between the measured and simulated isolation parameters is partially due to the limitations of APLAC model of the HSMP-386x PIN diode from Avago Technologies running in the Advanced Design System (ADS) computer-aided-engineering (CAE) software from Agilent Technologies. All measured isolation values met the design specification of 30 dB.

The results obtained from the experimental SPDT switch demonstrate the potential to use a HSMP-3860 Epi PIN diode in a SPDT switch circuit with low loss and high isolation at WiMAX frequencies. Still, additional effort is needed to find improved performance levels for this switch design. The basic design has good impedance match to other 50-O circuits and components, with minimal insertion loss of better than 0.75 dB worst case for the transmit mode at 2.7 GHz. Although the addition of the resonant LC circuitry narrowed the switch's isolation band, it increased the isolation to more than 30 dB.

ACKNOWLEDGMENT
The author would like to thank Universiti Kebangsaan Malaysia for supporting this research.

REFERENCES
1. J. S. Mandeep, N Misran, H. Abdullah, and T. C. How, "Patch Array Antenna Serves Satcom Needs," Microwaves and RF, Vol. 49, No. 4, April 2010, pp. 90-93.

2. N. A. M. Yunis, R. Wagiran, and V. Postoyalko, "Design of a Microstrip SPDT PIN Diode Switch," Semiconductor Electronics 2002, Proceedings of the 2002 IEEE International Conference on Semiconductor Electronics (ICSE), December 19-21, 2002, pp. 465-469.

3. Widad Ismail, J. S. Mandeep, E. Kadhum, and H. S. Hamid, "Analyze BER Performance of Wireless FSK Networks," Microwaves & RF, Vol. 48, No. 11 November 2009, pp. 80-89.

4. R. H. Caverly and G. Hiller, "Distortion Properties of MESFET and PIN Diode Microwave Switches," IEEE Microwave Symposium Digest, Vol. 2, June 1992, pp. 533-536.

5. David Hoag, James Brogle, Timothy Boles, Daniel Curcio, and David Russell, "Heterojunction PIN Diode Switch" IEEE Microwave Symposium Digest, 2003 IEEE MTT-S International Symposium, Volume 1, June 8-13, 2003, pp. 255 258.

6. R. H. Caverly and G. Hiller, "Distortion Properties of MESFET and PIN Diode Microwave Switches," IEEE Microwave Symposium Digest, Vol. 2, June 1992, pp. 533-536.