Passive intermodulation (PIM) has been recognized as a potential problem in communications systems for nearly 50 years. The phenomenon occurs when two or more signals encounter a nonlinear junction and “child” frequencies are generated that are mathematically related to the “parent” signals. With the advent of cellular communications, PIM began to rise in prominence as a concern, due to the quality of service impact these unintended signals could have by interfering or blocking the uplink (receive) channels of the base station.

As the network technology progressed from 2G to 3G and now 4G, we have gone from wondering what would be the “killer app” that would justify the investments being made in spectrum, infrastructure and technology to now struggling with delivering sufficient bandwidth and capacity to gratify the consumers’ appetite for data intensive applications that are delivered faster and more reliably. And consumers do not just want this, they expect it.

As a result, controlling or eliminating any interfering signals that impair data rates and signal quality is a must in optimizing network performance. Because PIM is a measure of nonlinearity in the system and is highly sensitive in detecting inadequate construction quality, it has become the preferred performance metric at the base station.

A Brief History

In the early days of commercial telecommunications, manufacturers of RF components used in the high power path of the base station were loosely required to verify PIM performance. Some companies “verified by design,” some through first article testing, some through sample testing and some of the more enterprising companies did 100 percent acceptance testing. Those companies doing testing would cobble together signal generators, amplifiers, a spectrum analyzer and a bench full of filters to realize their test bench.

Summitek introduced its first PIM analyzer in September 1996 and brought standardization to the industry and PIM testing. In this same time period, The International Electrotechnical Commission (IEC), Technical Committee 46, formed a working group tasked with defining recommended test methods for PIM. The first edition of IEC 62037 specification was released in 1999. Although this specification covered only “RF connectors, connector cable assemblies and cables,” it was widely adopted as the recommended test method for all components. In its simplest terms, it proposed testing PIM at 2 × 43 dBm (20 W).

The follow-up specification that defines test methods and procedures for a larger variety of components, antennas, filters, cables, connectors and cable assemblies, has been in process for the last ten years and is nearing formal release. This version expands upon the test methods and specifically recommends the need for dynamic testing, where the device under test is mechanically stressed (tapped or flexed, as appropriate).

In 2005, as high data rate networks were becoming a focus, Telstra, in Australia, realized the importance of “clean” infrastructure and engaged Triasx to develop a rugged, field suitable PIM unit. Telstra was the first network operator to require that PIM be measured and confirmed to be within acceptable levels as a requirement in site certification. It is important to note that PIM does not replace the need for return loss verification. It is a complementary test that identifies a set of problems that might go undetected with a return loss measurement. Summitek teamed with Triasx in 2007 to introduce the first 2 × 20 W portable PIM (PPIM) unit in the United States. PIM testing at the base station to verify construction quality is now a widely accepted requirement by nearly every network operator in the US.

The Test Process

Conceptually, PIM testing at the base station is simple. The jumper cable is disconnected at the BTS input, the PPIM unit is attached in its place, the two high power signals are turned on and the resultant IM (typically the third-order product, IM3, because it has the highest signal strength) is monitored.

If the IM level is within performance limits, then dynamically testing of the infrastructure needs to be started. Contrary to the belief of some that a pass in a static test is sufficient for certification, a site cannot be certified unless it can maintain acceptable PIM performance under mechanical stress. This is the standard procedure for every network operator, which is consistent with the emerging specification from the IEC.

Dynamic testing involves tapping all connector interfaces, all filters (diplexers and tower mounted amplifiers) and the back of the antennas. This should not be destructive in force, but sufficient to move any minor mechanical discontinuities that might exist in the component. Additionally, jumper cable assemblies should be gently flexed.

A word of caution about transmitting two high power signals through the antenna: in most cases, if an operator wants to stay within their licensed band, they cannot get enough separation between the two test frequencies to have the IM3 frequency land within the receive (uplink) band. Consequently, they will need to transmit outside their licensed band. Summitek recommends that the carriers be placed at guard band frequencies to minimize the chance that they will interfere with another operator. This should always be done with a good neighbor policy in mind.

It is not recommended to transmit swept frequency signals through the antenna and into the air. A swept frequency test at 2 × 20 W (43 dBm) will almost guarantee that it will interfere with at least one other operator. If a swept frequency measurement is desired, then the antenna should be replaced with a low IM termination to eliminate the possibility of interference.

Locating the Source of PIM
As mentioned above, whether the line being tested passes PIM with a static test or not, it is a requirement to dynamically test all components and interconnects to certify that there are no unacceptably high PIM sources that are hidden under static conditions. It becomes evident in a hurry where the PIM source is if, when tapped or flexed, the PIM goes up or down measurably (more than several dB), as shown in Figure 1.

Several suppliers have developed a range-to-fault (RTF) technology that includes the additional hardware and signal processing software to generate time domain plots similar to those generated in network analysis distance-to-fault. RTF technology is an analysis tool developed to enhance, not replace, standard fixed tone PIM testing.

The most effective way to use RTF analysis is to systematically remove the largest magnitude PIM source identified on the line. Repeat the analysis and continue removing the largest PIM source found until all significant static PIM sources have been removed. Regardless of its location on the line, the distance to the largest PIM source will be predicted most accurately by the algorithm. Each time a PIM source is repaired, the accuracy for locating the next largest PIM source will improve.

As initially stated, RTF analysis is not a replacement for dynamic PIM testing. RTF analysis will enhance site testing and potentially speed the removal of static PIM sources at the cell site. The analysis alone, however, should not be used to certify construction quality because:

• Knowing the range to a fault provides a helpful starting point, but does not ensure there are no other hidden PIM sources within the RF feed system
• The absolute value of the RTF PIM magnitude may not be accurate due to distortion brought about by frequency sensitive group delay in RF devices, such as surge arrestors, filters and TMAs. Furthermore, the windowing function used in the transform can affect the magnitude of the data in the time domain.
  
• “Ghost” PIM sources can be created as a product of the mathematics and/or by impedance mismatches in the system, that reflect PIM generated at different locations on the line

What Transmit Power Should Be Used?
Two, four, 20, 40 W, or more? This is an often asked question. The IEC PIM standard committee confronted this question many years ago and recommended a test power of 20 W (43 dBm) per carrier. Consequently, this has become the industry standard— nearly all manufacturers specify and verify their products based on the third-order IM level, when measured with 2 × 43 dBm carrier powers.

What conditions would have to exist for the benefit of testing at higher power levels?

1. The PIM power level would have to increase in a nonlinear manner, relative to the carrier power.
2. Or, the higher power would have to excite a PIM source not excited at lower power levels.

If either of the above were true, there would be a dilemma as to when to stop increasing the test power, because there would always be a level of uncertainty. Fortunately, based upon more than 15 years of real world experience, as well as theoretical analysis confirmed by experimentation, neither of the above is true.

Theoretically, the IM3 increases by 3 dB for every 1 dB increase in the carrier power. Although it is known from experience that the relationship is rarely 3 dB/dB in passive components, it is indeed linear. The slope will vary depending upon the characteristics of the PIM source, but it is always linear. To demonstrate, three different devices were measured with carrier powers from 33 dBm (2 W) up to 46 dBm (40 W) in 1 dB increments. The devices under test are a jumper cable built on site at a base station, a jumper cable build in the factory and a resistive termination (high PIM source). The test results are plotted in Figure 2. Overlaid on each plot of the raw data is a best fit line. As can be seen, in all cases, the PIM is linear with power and testing at higher power does not result in the illumination of a PIM source that is not already excited at lower power—even all the way down to 2 W.

But what about the second condition postulated above; will testing at higher power cause an intermittent PIM source to show up that might not be seen at lower power? The answer is—possibly. But this is exactly why PIM testing must always be a dynamic test. Each component being manufactured in the factory and every joint and component comprising the base station infrastructure must be subjected to mechanical stress such as a tapping on connectors, tuning screws of filters, backplanes of antennas, etc., and light flexing of cable assemblies. If the component or site is not dynamically tested, the results are invalid.

So, if testing at higher power is not of benefit and PIM is a straight line, why not test at lower powers and adjust the PIM specification to compensate for the lower power? If feasible, this has the added benefit of reducing the potential RF hazard to test personnel. Testing at lower power levels may be acceptable in some cases—particularly for pretesting components—but currently site certification of the base station infrastructure is always done with 2 × 20 W.

Testing Antennas Before Installation
Base station antennas have become increasingly complex as demands for network optimization require greater flexibility and versatility. What began as a simple array of linearly polarized dipoles has evolved into highly complex electrical and mechanical structures that support multiple bands, multiple polarizations and remotely variable radiation patterns. Now the advent of antennas that integrate the radio and the radiating elements is happening.

The complexity of these antennas means that there is a greater opportunity for mechanical failure during the transportation process. Consequently, it has become common practice to test and verify the PIM performance of the antenna prior to installation.

Accurate characterization of the antenna prior to installation requires keen awareness of the surrounding environment and the potential for creating PIM from external sources. Ideally, the antenna should be situated on a non-metallic structure at least one to two feet above the ground pointing straight up into the sky, as shown in Figure 3. The area around the antenna must be clear of any metallic structures (fences, fork lifts and towers) that might cause PIM. The antennas cannot be tested inside a warehouse unless tested inside an anechoic chamber.

CONCLUSION

Passive intermodulation has long been recognized as a potential impairment to the performance of telecommunications systems and was formalized as a quality metric by the IEC in 1999. Modern telecommunications systems are more dependent than ever on a “transparent” RF infrastructure to achieve their performance objectives. Realizing these goals is directly dependent upon the quality of construction of the components used in building the sites, and the quality of the interconnections that join them together. The most sensitive and effective parameter for determining construction quality is PIM. Consequently, PIM testing has become an essential part of the site certification process.

Rick Hartman is President of North American operations at Kaelus Inc. (fomerly Summitek Instruments).