Forward and Reflected IP3 Measurements for Switch Components and Systems

RF communication systems using advanced modulation formats rely on the linearity of the signal chain components to meet throughput performance requirements. Nonlinearities result in mixing products which include harmonics and intermodulation (IM) products. Though passive components such as cables, connectors, switches, and antennas are highly linear, they can become significant sources of passive intermodulation (PIM) when transmitting. This article describes the theory, challenges, and methods for testing highly linear passive components for PIM.

A linear device or system produces an output signal that is proportional to its input. The resulting transfer function relating output to input can be plotted as a straight line with slope corresponding to the system gain. Component nonlinearity typically results in the output signal compressing at large input levels. Such a nonlinear transfer function may be approximated by a Taylor series expansion with exponential terms. Frequency mixing occurs in higher order terms (for example 2nd order squaring, 3rd order cubing, etc..) producing both integer harmonics and intermodulation products.

Mixer products for a single input tone of frequency f result in output tones at f, 2·f, 3·f, . . . n·f with decreasing magnitudes for higher order terms. For narrowband communication systems, harmonics typically will be out-of-band and easily be filtered out. The 2nd order mixing of two frequencies f1 and f2 results in sum and difference frequencies (f1+ f2) and |f1- f2| along with harmonics 2·f1 and 2·f2. The 3rd order mixing of two frequencies f1 and f2 result in mixing products that are near the fundamental frequencies f1 and f2 at |2·f1-f2| and |2·f2-f1| with other products falling well above and out-of-band. The in-band IM products are difficult to filter out and appear as spurious frequency components. Advanced modulation formats with many simultaneous frequencies compound IM products into what is referred to as Spectral Regrowth. IM products create sidebands leading to adjacent channel interference and increased occupied bandwidth. 

IM for weakly nonlinear devices such as passive components, switches, and linear amplifiers can be approximated by considering only 3rd order IM effects. Practical characterization is simplified by testing with only two frequencies close together and choosing a figure of merit where fundamental power is equal to the third order intermodulation (IM3) power. This scalar is defined as the Third Order Intercept (TOI) or Intercept Point 3rd Order (IP3). The meaning of IP3 is graphically understood by plotting the device transfer function in log-log format. Plotting fundamental power as an idealized straight line with a slope of 1 (extrapolated beyond compression) will intersect the theoretical IM3 power of slope of 3 (due to cubing in 3rd-order products). The point of intersection corresponds to Input IP3 (IIP3) and Output IP3 (OIP3) as shown in figure 1.

Figure 1. IP3 is the Extrapolated Intersection of the Linear and 3rd Order Curves

IP3 is not measured directly and is well beyond the compression point for the device.  Instead, IP3 can be computed from a single measurement of the fundamental power PO and resulting IM3 in dBc. Knowing the slopes of the fundamental and IM3 curves on the log-log plot, IP3 can be computed by geometry as:

IP3 = |IM3/2| + Fundamental Power

Fundamental power is the power in both fundamentals f1 and f2 with IM3 being the measured IMD in both sidebands |2·f1-f2| and |2·f2-f1|. For practical measurement and filtering considerations, it is easier to measure the single carrier level (SCL) of one of the fundamentals and the IMD in just one of the sidebands. This approach can be taken if the amplitude of the two fundamental tones are equal thereby resulting in sidebands of equal power. The test method presented below provides a Tone Level Calibration procedure intended to balance tone power within +/- 0.05 dB.

IP3 has considerable practical application. Component nonlinear performance can be compared and qualified by the IP3 value itself. The larger the IP3 value, the later the IM products come into play with increasing input power. PIM specifications for connectors and other highly linear components are typically specified in dBc using a standard 2x20 W test method. IP3 is easily computed from PIM by taking care to convert dBW to dBm. From the log-log IP3 plots, it’s evident that for every 1 dB increase in fundamental power, IM3 increases by 3 dB. Alternatively, doubling the fundamental power results in quadrupling IM3. Comparison of representative IP3 levels for various components and test standards include:  high quality linear amplifiers and GaAs RF switches ranging from 40 dBm to 50 dBm, Menlo Micro MM5130 SP4T MEMS RF switch is 95 dBm, typical PIM passing test level is 116 dBm, low PIM connectors greater than 120 dBm.

Passive devices have high IP3 and low levels of PIM better than 140 dBc.  These high ratios of IMD (fundamental to PIM) raise significant challenges for the hardware test setup.  Generation of test tones must be clean and of equal power, properly isolated, and combined for input to the amplifier.  Low PIM attenuation and mixer level must be carefully adjusted to insure IM3 is just above the noise floor of the RF analyzer. Use of low PIM test fixture cables, adapters, and load terminators are required in the transmit path.  Connectors must be of the right type and properly torqued before measurement.  Every path loss must be measured in both forward and reverse directions.  

The automated IP3 test system used to test the Menlo Micro MM5130 SP4T DC to 26 GHz high power MEMS RF switch is shown in figure 2.  Key components include custom LabVIEW program, Kaelus 40 Watt PIM tester, duplexer with high rejection IM filter, and spectrum analyzer for IM measurement.  

Figure 2. Automated IP3 Test System

The custom program shown in figure 3 is written using the National Instruments™ LabView graphical programming environment. It provides DUT specific selection parameters such as device type and the number of switch channels to test. This information is used by the DUT USB driver for device configuration. Resources including spectrum analyzer, programmable power supply, and coaxial switch are accessed by LabView through GPIB. Other selection parameters include test chamber temperature, DUT power supply voltage, forward/reverse path measurement, path losses, and qualification data. This Control Panel is shown below prior to starting the test.

Figure 3. Control Panel GUI is custom LabView Program

Test tone power level calibration is performed using a coupler to measure forward power. The Control Panel includes the Tone Level Calibration procedure which ensures the test tones are within 0.05 dB of each other. Figure 4 below shows the two test tones when observed by a spectrum analyzer. The coupler does not remain in-circuit as the power meter will distort the PIM measurement. The test fixture also includes a thru-path which is included on the MM5130 evaluation kit (EVK) PCB. Dashed lines in the test system diagram illustrate that only one of the three paths is connected at any one time.

Figure 4. Measured Tones of 16.4 dBm, with Coupler Loss of 20.5 dB for Tone Power of  36.9 dBm

The test configures tone frequencies of 869 MHz and 891.5 MHz with IM measurement of the lower sideband at 846.5 MHz. The PIM tester generates two 20 W test tones (2 x 43 dBm) which are reduced by 6 dB in the test fixture by the low PIM, high power attenuator shown in figure 5. The resulting test tone levels of 5 W (2 x 37 dBm) are applied as the input of the DUT. Though the MM5130 is rated to 25 W continuous, testing at 2 x 5 W is well within its operations rating.

Figure 5. High Power, Low PIM 6dB Attenuator

The DUT is the MM5130 mounted on its evaluation board (EVK) and connected using low PIM connectors, see figure 6 below. Each of the four RF switches have separate connectors with the fifth connector being the common RFC. The EVK includes a USB driver interface for switch control and requires the 90 volt gate supply to be sourced externally from the programmable power supply.

Figure 6. Menlo Micro MM5130 EVK Cabled for Testing Switch 1

A close-up view of the MM5130 mounted on the EVK is below in figure 7. The device is a glass package with the MEMS switch structures visible inside.

Figure 7. MM5130 Mounted on EVK PCB

Forward IM is measured through the use of the duplexer and filter shown in figure 8. The 3-port duplexer splits the fundamental tones from the lower sideband IM product. The IM bandpass filter provides 95 dB rejection for measurement of forward IM. Reverse IM normally measured in the PIM tester is output from the PIM Tester for measurement by a spectrum analyzer. The Control Panel selection of Reverse or Forward path configures the coaxial relay accordingly. Measurements from the spectrum analyzer are read by the Control Panel, visually displayed, and IP3 is computed.

Figure 8. 3-port Duplexer with Terminator and Filtered IM Output Ports

An example measurement is shown below in figures 9 and 10 focus in on the Control Panel array of IP3 measurements for each channel switch in the MM5130 and the IM3 sideband plot for the current measurement. Automated IP3 testing can be configured to run voltage and temperature characterization over multiple devices and lots.

Figure 9  Control Panel Display of Measurement Results

Figure 10. MM5130 IP3 Measurement Results for All Four Channel Showing Typical +95dBm

From the measured IM above (single sideband) of -73.928 dBm, the corrected IM at the DUT is found by subtracting net path gain of 6.99 dB with the resulting corrected IM being -80.92 dBm.  Computing IM3 from the SCL of 36.9 dBm, IM3 is 117.82 dBc.  Applying the equation earlier presented, IP3 is computed from IM3/2 + SCL or 117.82/2 + 36.9 dBm and found to be 95.8 dB corresponding to the last switch tested in the figure above.

Summary

IP3 measurement in low PIM devices requires high test tone power levels and sensitive instrumentation to measure the resulting IM sidebands. Successful measurement requires the setup to be constructed using low PIM components and carefully calibrated for tone level and path loss. Automation software facilitates characterization over temperature and voltage and is suitable for lot level testing.

Modern communications systems ranging from 5G infrastructure cells to wideband military radios, demand increasing levels of RF subsystem and component performance. We’ve presented some of the challenges in measuring very high levels of 3rd-order intermodulation products, particularly on ultra low loss and highly linear passive devices. Paying careful attention to the test equipment and accurate characterization of all components will ensure stable IP3 measurements can be achieved on high performance devices such as the Menlo Micro MM5130 switch.