Figure 1 Biasing network of an SP2T switch controlled with bipolar voltages.
A PIN-diode-based single-pole multi-throw (SPMT) switch with a low power driver for applications where only a single source is available eliminates the need for DC-DC converters and the current required to power them. Further current savings are obtained by powering the enabled branch of an SPMT switch with the currents flowing on the disabled ones. In the example reported, CMOS ports are used to bias an SP4T switch and simultaneously realize a decoding network for the input commands, which are TTL compatible. RF measurements show good performance compared with those obtained with a bipolar supply voltage.
PMT RF switches are widely employed in switching systems, multiband selectors and filter banks.1-3 For a PIN-diode-based SPST switch, the typical configuration of the RF port is a series diode where signal flows in the low attenuation state and a shunt diode that directs the signal to ground in the high attenuation state. Compared to an SPST switch, the SPMT switch must also present a high impedance to the input RF port in order to minimize disturbances on the output caused by the non-activated branches. The currents required to enable one branch while simultaneously disabling the other ones always have opposite signs. This requires a bipolar voltage to properly drive an SPMT switch.
For applications where only a single supply voltage is available, a bipolar voltage can be achieved through the use of a DC-DC converter. This device requires current for its operation in addition to that required to drive the switch. To eliminate the need for a DC-DC converter, an alternative means has been developed to drive an SPMT switch with a single voltage. Since an external user is normally interested in changing an SPMT switch state by using a digital signal without concern for the required internal voltages, a TTL-compatible decoding network based on CMOS components is incorporated as well. The approach is applied to a commercial SP4T switch. RF measurements are compared with those of the same device driven by bipolar voltages, demonstrating comparable performance.
Figure 2 Modified biasing network to drive an SP2T switch with a single voltage.
Concept
The typical configuration of an SP2T switch is shown in Figure 1. V1 and V2 are the voltages that are applied to each on and off path, respectively. Along with the value of the resistance R, they determine the current that flows through a branch. For example, a current of ION = 20 mA may be selected for the branch in the low attenuation state and IOFF = 20 mA may also be selected for the branch in the high attenuation state.
If a commercially available resistance of R = 270 Ω is fixed, the required driving voltages VON and VOFF can then be found. With reference to Figure 1, suppose that one path is operative (V1 = VON and I1 = ION) and the second one is off (V2 = VOFF and I2 = IOFF). Then:
where the voltage across the diode (VD) is considered to be equal to 0.7 V. This result shows that the control of the switch can be realized by two voltages of opposite sign with a total current of 40 mA.
A way to obtain the same working conditions (ION = 20 mA and IOFF = 20 mA) with a single voltage is by modifying the circuit as shown in Figure 2. In this circuit a resistor (RX) is inserted between all diodes and the reference ground. To maintain proper device operation, however, the reference voltage for RF signals must be preserved. This is done by placing a capacitor (CX) in parallel with the voltage source. With this configuration, a positive voltage applied to the second path is sufficient for generating the required currents I1 and I2. In this example, I1 and I2 are the same and the remaining current IX (IX = I2-I1) is zero, but in general this may not be the case. For IX to always be zero, the value of RX must be infinite, i.e. an open circuit.
The voltage applied to disable the second path is V2 = VOFF = 5 V, so the original current values ION and IOFF are maintained if R = 90 Ω, as found from Equations 3 and 4. Since only a single supply voltage is available, the value of V1 = VON, to enable the first path, is zero.
Besides providing the ability to drive a switch with a single voltage, this circuit provides additional current savings. The total current flowing in the circuit is I1 = I2 = 20 mA, which is half the current required for the bipolar circuit (see Figure 1). Table 1 compares SPDT, SP4T and SP8T switches having ION = IOFF = 20 mA, where the total currents required by the two driving methods are shown.
Figure 3 Surface mountable on-chip SP4T switch assembled on a microstrip circuit with split ground plane.
Application
A commercial SP4T switch is used to test the performance of this design.5 The surface-mountable SP4T chip is assembled in a microstrip circuit (see Figure 3). Via holes are used to connect its ground to the bottom layer. The bottom layer has a gap that divides the ground plane into two portions. The outer portion is the DC reference which is separated from the inner part by a gap of 150 µm. To maintain an RF short circuit, a capacitance (CX) is placed across these two portions of the ground plane. For best RF performance, this capacitance should not be concentrated at only one point but should be distributed as uniformly as possible along the perimeter of the gap, as illustrated in Figure 4a.
In practice this is achieved with N evenly spaced lumped components of value Ci connected in parallel. In this case, N = 8 and CX = 8Ci. Each is modeled by an equivalent circuit including a series inductance along with its nominal capacitance. This data was obtained from the datasheets of commercially available SMD capacitors for RF applications.
Figure 4 Split ground plane with a distributed capacitance over the two ports (a) and microstrip transmission line (b).
Figure 5 S-parameters of the microstrip line in Fig. 4b for different values of Ci; |S21| dB (a) and |S11| dB (b).
To determine the proper CX value, a transmission line that models a path of the SP4T switch is simulated as shown in Figure 4b. EM simulations, shown in Figure 5 for different values of Ci, are used to optimize the transmission coefficient in the frequency range 3 to 6 GHz. Good transmission performance (S21 ≈ 0 dB) and good return loss (S11 > 20 dB) are achieved over the entire band for Ci = 4.7 pF (CX = 37.6 pF).
Figure 6 NAND port-based logic circuit employed for driving the SP4T switch.
The actuation voltages are VON = 0 V and VOFF = 5 V, and the currents are ION = IOFF = 20 mA (as in the previous case). These voltages can be supplied by a CMOS device, and therefore NAND logic ports are used to build this function together with a decoding network for the SP4T’s external commands (see Figure 6). In order to use all identical components in this network, two of the NAND ports act as inverters. CMOS technology has several advantages: its high (low) output is close to 5 V (0 V), its static power consumption is virtually zero and its input is compatible with a TTL driver.
Measurements of the SP4T switch are shown in Figure 7 with a bipolar voltage driving (red dashed line) and with the proposed architecture (solid blue line). They show the average transmission coefficient of the four channels in both cases, which includes the biasing network, microstrip lines of the test board and connectors used to carry out the measurements. The isolation between the four channels is comparable with that listed in the datasheet of the SP4T switch, i.e. between 55 and 60 dB in the band considered.
Settling time is shown in Figure 8. The rise time (tr) = 243 ns, between 50 percent of the TTL input command and the 90 percent of the maximum detected RF output. The fall-time (tf) = 15 ns, between 50 percent of the TTL input command and the 10 percent of the maximum detected RF output. These values are compatible with the typical applications where SPMTs switches are employed.
Figure 7 Average transmission coefficient of the SP4T switch driven using the described single supply architecture (solid blue line) and with bipolar voltages (dashed red line).
Figure 8 Settling time measurement showing rise time tr = 243 ns (a) and fall time tf = 15 ns (b).
Conclusion
A PIN diode SPMT switch with a low-power driver uses a single bias voltage. This eliminates the need for a DC-DC converter to convert a single supply into a bipolar supply, as well as the current required to power it. Moreover, with this architecture, the enabled branch is powered by the currents flowing through the deactivated ones, providing further energy savings. A decoding network in CMOS technology is used to provide the required switch voltages, making the driver compatible with TTL input commands. RF measurements show that switch insertion loss is comparable to what is achieved using a bipolar voltage supply.
References
- K. Ma, Q. Sun, F. Cheng and R.M. Jayasuriya, “A 11 to 20 GHz Switched Filter Bank for Software Defined Radio System,” IEEE MTT-S International Microwave Workshop Series on the Art of Miniaturizing RF and Microwave Passive Components, December 2008, pp. 75-78.
- S. Tanaka, S. Horiuchi, T. Kimura and Y. Atsumi, “Design and Fabrication of Multiband P-I-N Diode Switches With Ladder Circuits,” IEEE Transactions on Microwave Theory and Techniques, Vol. 54, No. 4, June 2006, pp. 1561-1568.
- P. Farinelli, E. Chiuppesi, F. Di Maggio, B. Margesin, S. Colpo, A. Ocera, M. Russo and I. Pomona, “Development of Different K-Band MEMS Phase Shifter Designs for Satellite COTM Terminals,” European Microwave Conference Proceedings, September 2009, pp. 1868-1871.
- R. Sorrentino and G. Bianchi, “Microwave and RF Engineering,” Wiley, UK, 2010.
- M/A-COM Technology Solutions, “MASW-004103-1365 Silicon SP4T Surface Mount HMIC PIN Diode Switch.” Online datasheet available at www.macomtech.com/datasheets/MASW-004103-1365.pdf.