Passive Intermodulation Distortion (PIM) has been a thorn in the side of RF and microwave designers for decades, and no system is immune, from cellular, land mobile, public safety, radio and TV broadcast to satellite and defense systems. Even though many advances have been made to reduce, pinpoint, and measure PIM, it remains a significant challenge that will only increase in severity as 5G provides more places for it to manifest itself. With this mind, it seems an opportune time to revisit PIM once again.
PIM is such a scourge because it’s different from other forms of intermodulation distortion (IMD). It’s not generated by active (non-linear) components but rather by passive, presumably linear components, from antennas to attenuators, cables, duplexers, diplexers, filters, and connectors. PIM is also not confined to systems themselves as everything from drainpipes to roof flanges, tower components, guy wires, and fences can produce it.
Worse yet, PIM can also appear within indoor systems such as in-building wireless and distributed antenna systems (DAS) that create opportunities for PIM because of their proximity to building materials, ductwork, and many other sources. As will be described in more detail below, PIM results in degraded receiver performance in the host system and potentially others nearby, and in worst-case scenarios can render a system useless.
No Power Too Low
PIM has typically been considered a problem for systems generating high RF power levels. This misnomer has remained in its definition primarily because high-power systems were the first to experience it, such as Navy ships and AM, FM, and TV broadcast installations. Given a strong enough transmitted signal and components producing PIM, damaging effects can be experienced even several miles away.
The problem is severe on Navy ships, for example, because more and more discrete transmit and receive systems have been added “topside” over the years that, in combination with the ship’s structure and exposure to the effects of salt spray, present massive opportunities for PIM generation.
That said, PIM has been a problem for lower-power cellular systems since almost their earliest days and increased in severity with the transition from analog to digital modulation techniques. The introduction of higher-order modulation techniques such as OFDM also translates into higher peak power levels and a potentially corresponding increase in PIM.
What’s more, while PIM didn’t present a major challenge when channel spacing was wide in early cellular systems, there were ample guard bands, and their typical signal strength was at least 100 dBc below that of the carrier. This is far from the case today, as even PIM levels of 150 dB below the system’s carrier signal can wreak havoc on system performance. The likelihood of PIM has also increased with the colocation of infrastructure from multiple wireless services and public safety systems along with multiple transmissions using a single antenna.
In the coming years, the infrastructure required for 5G will add to the number of potential PIM sources, as will the greater use of more aggressive carrier aggregation to produce wider channel bandwidths. PIM is also a major headache for testing environments because passive components in the test set such as couplers, attenuators, cables, and connectors between the device under test and the test equipment invariably produce some level of PIM.
Taken together, the problems noted above represent a widespread problem, as does the difficulty in accurately measuring PIM and locating its source. In short, there is no single, unique source of PIM as it encompasses an entire category of intermodulation phenomena and can occur almost anywhere from within a system to structures on the premises. It is arguably a greater challenge than other intermodulation products caused by amplifiers (for example) is obvious and whose intermodulation products (i.e., non-linear distortion) can be eliminated by filtering.
PIM appears when two or more signals are present in a passive non-linear device that mix and produce other signals related to them. Unlike mixers, in which this is the desired result, PIM creates unwanted non-linearities in devices normally considered linear. If these signals are significantly weaker than the desired signal, they may have no effect but if they’re strong enough and appear within the system’s operating frequency, the result can be disastrous because they increase the receiver’s noise floor.
The two main intermodulation products of concern are third- and fifth-order intermodulation, which for two frequencies (f1 and f2) are produced by the mixing of fundamentals and harmonics:
Third-order intermodulation: 2f1 – f2
Fifth-order intermodulation: 3f1 – 2f2
The seventh-order product can also be an issue but less so as PIM signal strength is much lower. This does not rule out its effect entirely but makes it less problematic when the system’s RF output power is low.
To visualize PIM, consider LTE Band 2 (Figure 1) whose downlink frequencies are from 1930 to 1990 MHz and uplink frequencies are 1850 to 1910 MHz. If two transmitter carriers located at 1940 and 1980 MHz are transmitted from the base station and generate even moderate amounts of PIM, a signal component at 1900 MHz will occur in the receive band, raising the noise floor high enough to either degrade or completely disrupt the receiver’s ability to detect the desired signals. Other products can interfere with other systems at 2020 MHz as well.
Another issue is that PIM forces the system to operate at maximum power instead of under power control, which increases power dissipation and causes greater intercell interference. While active components like amplifiers can produce very high levels of intermodulation distortion, it can be removed by filtering, while PIM cannot be because it is generated late in the signal path or outside the system entirely.
In addition, as components age and especially when exposed to the environment, they are subject to loosening of connections, the incursion of dirt, damaged or incorrectly torqued connections, fatigue, cold solder joints, and corrosion, all of which can produce conditions ripe for the production of PIM. Even subtle defects such as a scratch on a conductor, contamination by minute particles during manufacture or after installation, as well as the pressure of a cover on a component such as a coupler or a combiner can increase the level of PIM.
The nonlinearities produced in passive components result from rectification at conductor joints, poor mechanical junctions, or both. It can be caused by many factors, including surface oxidation, loose metal-to-metal contacts, contaminants such as solder splatters, ferromagnetic materials in or near the current path, and contact between dissimilar metals. In addition, insufficiently thick plated metals, improperly torqued connectors, scratches on the connector faces, and contamination in dielectric material can also cause PIM.
Materials with ferromagnetic properties are of particular concern as they are well known sources of PIM, and ferromagnetic content of even 100 ppm in a dielectric material can increase its production. Dissimilar metals in the transmission can cause oxide-metal formation between the surface plating and base material. In addition, nickel can cause non-linear voltage-to-current ratios and some stainless steels can cause a hysteresis effect. As a result, ferromagnetic materials including ferrites, nickel (including nickel plating), and some stainless steels aren’t used in components designated as having very low PIM and are rarely employed in wireless systems.
Fortunately, manufacturers of passive components have made impressive advances in reducing PIM over the years, and there are many designed specifically for applications such as wireless systems in which the lowest possible PIM is not just desirable but essential. These products include almost every type of passive component used in these systems and are specified as meeting low-PIM requirements.
A good example of a connector designed for low-PIM applications is the SC4917 7/16 DIN flange-mount connector from Fairview Microwave (Figure 2), rated for PIM of less than -163 dBc using a standard 20W, two-tone test as defined in the IEC 62037 standard. It operates at up 7.5 GHz with a maximum VSWR of 1.15:1 and uses a silver-plated brass contact with plating rated to meet the QQ-S-365 specification. The connector is one of many designed by Fairview to meet low-PIM applications that include adapters and operation up to 12.4 GHz.
Another example is the FM DV1028 three-way power divider from Fairview that operates from 600 to 2700 MHz and is designed without solder joints to help to reduce PIM to levels of at least -160 dBc. It is well suited for use in DAS applications where many power dividers are used and can handle an RF input power of 300W CW. The power divider can be used outdoors as well as indoors as it meets the moisture requirements of IP67.
Testing to the Rescue
PIM levels are usually evaluated by means of reflective testing. For example, two signals are injected into a device under test (DUT) or into a transmission path, such as leading to an antenna, and the levels of PIM are analyzed from the reflected signals at the input port of the DUT or transmission path. Cables are usually part of the test setup, but they can also be part of the PIM problem, and although they can be a source of PIM at high power levels, for test and measurement applications this is typically not a problem.
Measuring PIM levels has become much more accurate and less difficult thanks to the emergence of portable battery-powered spectrum analyzers that can be used for on-site testing as well as a variety of dedicated PIM testers and accompanying software. When testing is performed with a spectrum analyzer, two signal generators or an arbitrary waveform generator produce two independent test signals to simulate the production of PIM. They typically require a pair of amplifiers to increase the power of the test signals to levels matching those produced by the system transmitter.
One of the most recent approaches is testing PIM over CPRI, which complements RF-based PIM testers and noise rise monitoring. It employs spectral analysis of I/Q data at the optical fiber interface between the baseband unit and remote radio head to visualize the PIM profile. It is an effective approach because the optical CPRI link includes a wealth of baseband data that characterizes both the downlink and uplink signals transmitted and received at RF by the remote radio unit.
The data on the CPRI uplink will show signals received from user equipment as well as interference to the RF signal. Consequently, it can measure all intermodulation products in the uplink as well as harmonic content using a single measurement and monitor traffic remotely in real time without affecting base station’s RF hardware, while also detecting the location of PIM both on and off-site at or below the noise floor and make it possible to identify which solutions will be best suited for mitigating PIM.
The bad news is that PIM will always be present somewhere in a system as any passive component is susceptible to producing it and it is typically impossible to remove through filtering. The good news is that by using components specifically fabricated to produce extremely low levels of PIM, it is possible to reduce it to manageable levels. In addition, armed with knowledge about PIM, system operators and technicians tasked with installing and maintaining wireless systems can take steps to ensure that every component in the signal path is properly attached and free of undesirable conditions.