Building Low PIM Coax Assemblies with Consistent Dynamic Performance
By Times Microwave
Building coax assemblies with low PIM performance is challenging ,but building assemblies that retain low PIM performance in a dynamic (flexing) environment is even more challenging. Factors for achieving low PIM in a flex environment can be categorized into A) material selection, B) connector attachment, and C) test systems & test methods.
Telecom industry accepted cables for low PIM are generally limited to designs with solid metal outer conductors. Corrugated copper or Times LMR-SW smooth wall cable produce excellent results of up to -122 dBm at 700 to 2100 MHz. To achieve low PIM performance, it is important to eliminate all ferrous metals in the RF path. To achieve optimum PIM performance, Times connectors are manufactured using virgin brass, not recycled brass as is found in some connectors from outside the US. Recycled brass can be contaminated with other metals including iron and steel. A superior (extra smooth) machined finish on all connector surfaces carrying RF and the use of silver or tri-metal (Albaloy) plating on connector bodies is required.
Silver or gold plating is preferred for the center contacts of the connectors. Tri-metal plating is preferred for the connector bodies to eliminate oxidation. Many suppliers have trade names for tri-metal plating, such as Tru-Luster and Suco-Plate, but they all contain a mixture of copper, tin and zinc. Times found connector plating baths must be meticulously free of contaminants and other metals in order to achieve consistently good dynamic PIM performance.
Two methods of attachment result in good PIM performance: Compression clamp and direct solder. Compression clamp connectors have the advantage for field termination but cost more. Solder-on connectors discussed here are typically used in a production environment where processes can be better controlled.
In either case, cable preparation and trimming dimensions are critical, with no ragged edges on either the center conductor or outer conductor. The cable end must be scrupulously cleaned to assure no metal particles are embedded in the core. Before soldering, an alcohol wipe of the solder surface to remove any debris or metal particles is standard practice. Soldering processes must be well defined and closely followed, but different assembly designs require a different mix of time and temperature. Various solder compositions from 100% lead to lead free solder and everything in between have been tried. The final result is the solder composition appears to have little impact on static or dynamic performance. Ultimately it is important to avoid cold solder joints which are a guarantee of poor PIM performance under all conditions.
When soldering, cables and connectors are pre-tinned to improve whetting and solder flow. Traditional resistance soldering tweezers have been replaced with induction methods for more consistent heating. Low heat induction soldering is used for the outer conductor to eliminate the possibility of melting the foam PE core.
Over the last three years Times has introduced several low PIM products. However, the direct solder designs suffered from low yield for dynamic PIM performance. In March of this year an effort was undertaken to determine the cause of this poor performance.
It is generally agreed a given PIM level is more difficult to achieve at lower frequencies than higher frequencies, likely due to RF skin effect. Our experience confirms this is true for coax assemblies. To reduce measurement uncertainty to an industry acceptable level of approximately +/-2 dB to +/- 3 dB, the residual PIM of the test equipment and components should be approximately 10 dB better than the measurement one is attempting to make (actual PIM*). Kaelus model SI-0700LE has superior residual PIM specifications (ours is certified to -135 dBm) and filters that offer excellent rejection out to the edge of the 700 MHz frequency band. This would provide the acceptable margin of error at the worse case (lowest) frequency. The unit was purchased new in January, 2013.
All test adapters and loads are rated at better than -122 dBm from 700 – 3000 MHz. Calibrated adjustable torque wrenches are used at all times. If male connectors are equipped with gaskets the mated pair is torqued to the center of the range for that series specified values. If interface gaskets are not used then the coupling nuts are tightened to approximately 75% of series specified values. This reduces wear and minimizes the generation of metal particles in the interfaces. Full torque tightening without interface gaskets does not measurably improve PIM results under static or dynamic testing and needlessly shortens a test component’s useful life.
PIM analyzer performance is checked daily IAW the manufacturer’s recommendations. Low PIM assembly production test stations are equipped with dry compressed air and alcohol wipes. All test components are thoroughly cleaned and checked for performance at the beginning of a test cycle. All interfaces are then cleaned at least every 3-4 test cycles because microscopic metal particles are created when connecting and disconnecting.
What follows demonstrates the critical nature of maintaining clean interfaces: One of Times’ new low PIM products is a long life test lead (SilverLine-LP). Occasionally customers return their test lead claiming performance failure due to a manufacturing defect. Our first action is to clean the interfaces and retest since this usually solves the problem. Figure 1 represents the debris cleaned out of approximately 10-15 interfaces gathered over a period of time from returns. Even though heavy duty protective caps to cover the interface are provided, the user must still take extreme care not to over-torque, misalign threads or damage interfaces in any way.
PIM testing requires some intuition that only comes with test experience. One can get a “feel” for whether the results are true or influenced by faulty test equipment or components. This is why the importance of having multiple copies of new test components cannot be overstated.
For example, in a recent production run of 250 cable assemblies, defects slowly increased from an average of 10% to 25%. Testing was stopped and all the test components were checked. When one specific adapter was replaced, the PIM trace changed (fell) substantially, dropping an average of 5 db into the pass range. All previously failed cables during that test shift were retested and many passed, returning yield rates to 10%. It should be noted that cleaning the suspect adapter did not improve its performance. It had reached the end of its useful life and was discarded.
In this situation, the only reasonable way to determine if the problem developed in production or in the test system was to stop testing and begin replacing test components until the offending culprit was located. If the entire test system passed, production would have been stopped until the variation that crept into the process was located and corrected.
A 10% yield on assemblies is still unacceptable, so the next step was to look at test conditions. Customers have asked for several different flex or tap tests, however Times defers to IEC62037-2 (2012)E. This rotational flex test is referred to as the “stir the pot” test. In addition, Times uses a wooden handle of perhaps 2-4 ounces (no metal). The mated pair of connectors is rapidly tapped and flexed on both ends out to and including the solder attachment area. The PIM trace must remain stable under all conditions of flexing and tapping of the entire assembly. A trace that jumps above the maximum acceptable PIM value at any time is a failure.
After all materials and assembly processes were controlled, testing methods were evaluated and test hardware with known performance was in place there was still a 10% yield problem on the assemblies. At that point it was determined that the soldering process must be the cause.
This was confirmed when Times discovered re-flowing a solder joint will oftentimes improve PIM performance to a passing level. However, this is not always true and a third reflow almost always resulted in irreparable product. Taking a solder joint apart destroys the evidence so a high power x-ray method for looking at the quality of a solder joint was used.
Figures 2 and 3 are x-rays of solder joints of failed cables. In Figure 2 one can see solder wire that was wrapped around the cable failed to melt. Figure 3 shows a different assembly, but in this case solder was flowed down from the cable entry instead of wrapping around the cable. In both cases the cable and connector were fluxed and pre-tinned. These solder joints looked perfect from the outside, yet in both cases the cables were not metallurgically attached to the connector sufficiently to maintain PIM at acceptable levels during the dynamic, IEC rotation test and/or tap test. It appears that insufficient heat was used in both cases.
Figure 4 is an x-ray image of a passing cable assembly. Clearly the solder joint is far more complete, with solder flowing to the end of the cable. To achieve this result significantly more heat was applied, but for a shorter period of time. Flux was used sparingly. Solder was fed in from the cable entry point and was sucked in rapidly. Upon sectioning, approximately 90+% of the cable was bonded to the connector. The process is hot but still fast enough to prohibit melting a foam PE dielectric. When the process was modified to increase the temperature but shorten the time, solder joints resembling Figure 6 became far more frequent and rejects were halved to approximately 5% on the first pass test. Of the remaining 5%, tap testing revealed problems at a suspected end. Reflowing that single end brought about 80% of them into spec, leaving about 1% of the original lot as irreparable.
In 2011, Times introduced SilverLine-LP (Low PIM), a ruggedized low PIM test leads for cellular testing that has become very popular (Figure 5). The design strategy uses super flexible, 3/8" corrugated cable sheathed inside a very robust steel armor. The armor eliminates or significantly reduces cable kinking and cracking from exceeding the minimum bend radius, thus extending product life. PIM is specified at better than -117 dBm under all conditions.
The 7-16 DINconnectors are terminated with soldered-on center contacts and a soldered outer conductor…..a traditional method for achieving low PIM performance. Because the cable is used constantly in a rugged environment and performance needs to remain at -117 dBm the back nut has been lengthened to completely eliminate motion and side load force at the solder joint (Figure 1). An internal clamping system assures pull forces are removed from the attachment area. A video demonstrating this design will exceed a 500 lb. pull force exists on the Times website. (http://www.timesmicrowave.com/cms/tech-resources/videos/product-demonstration-videos/index.shtml). The oldest cables have been in the field well over two years now.
Also in 2011, Times introduced LMR-SW (Smooth Wall) (Figure 6) for cellular jumpers at the tower top or inside the shelter. This cable features a full, 100% thin wall, seamless aluminum tube outer conductor and foam PE core. Type N and 7-16 connectors use a novel compression clamp method of attachment and result in excellent connector retention. Few special tools are required making on site assembly quick and easy. PIM is better than -122 dBm from 700 to 3000 MHz.
In early 2013, Times introduced its SPP product line (Figure 7). This is a UL910 Plenum rated low PIM jumper for In-Building DAS systems. The product uses a ¼" super flexible corrugated cable but with a PTFE tape wrapped core. This core construction, along with the outer jacket design, achieves the stringent UL Plenum rating and PIM performance of <-112 dBm 700 – 3000 MHz. The cable has low attenuation and is very flexible, making it an ideal telecom jumper that is easy to install in limited space situations.
this article to a friend!