DEV Community

Yang Fei
Yang Fei

Posted on

Passive Intermodulation: The RF Spec That Fails After Installation

Passive intermodulation is the specification that most often disappears from a component RFQ — and then reappears as a failed system test. It is called passive because the parts that generate it have no active devices in them. A switch contact, a connector interface or a load can all behave as a weak nonlinear junction when two carriers are present.

What actually generates PIM

Two tones at f₁ and f₂ passing through a nonlinearity produce products at mf₁ ± nf₂. The third-order terms 2f₁−f₂ and 2f₂−f₁ are the dangerous ones because they land close to the original carriers and can fall inside the receiver band. A typical test applies two 20 W tones and measures the third-order product in dBc; a well-built path might be −150 dBc or better, while a dirty or loose interface can be 30–40 dB worse.

The physical causes are mundane:

  • Ferromagnetic materials (nickel underplating, certain stainless steels) in the RF current path.
  • Oxidised, contaminated or loosely mated contact surfaces — the classic metal-oxide-metal junction.
  • Contact pressure that changes with temperature or vibration.
  • Filings, skin oils and plating wear at a connector mating face.
  • Dissimilar metals at a flange or adapter interface.

None of these show up in a small-signal S-parameter measurement. A connector can measure 1.15:1 VSWR and still be a poor PIM performer.

Where it appears in a switched path

An RF switch is a collection of contacts, so each path has its own PIM signature, and it changes as the contacts wear. Hot switching accelerates the problem: arcing roughens the contact surface and raises contact resistance, which makes the junction more nonlinear. A rack that switches under power may start life comfortably within the limit and drift out of it after a fraction of the quoted number of operations.

The coaxial switch range covers electromechanical and terminated configurations; when PIM is a hard requirement, ask for the test conditions (tone power, frequencies, product order) and whether the figure is guaranteed over the rated life, not just at delivery.

Unused ports matter too. A coaxial termination is itself a resistive element in the RF path when leakage reaches it, and its construction determines whether it adds a nonlinear junction. Specify low-PIM loads with the same test conditions as the switch. The same applies to a fixed attenuator pad placed in front of a receiver — the film or chip inside it is another candidate source.

For high-power, narrow-band systems built with waveguide switches, flange flatness, surface finish and the choice of gasket material replace connector torque as the dominant variable. A small gap at a flange can both arc and generate intermodulation. Solid-state PIN switches have a different PIM mechanism: the diode's junction capacitance is voltage dependent, so bias network design and tone spacing matter more than contact cleanliness.

A practical specification checklist

  1. Test condition, not just a number. State two-tone power, tone frequencies, which product order, and the limit in dBc. −150 dBc at 2×20 W is a meaningful figure; "low PIM" is not.
  2. Direction and state. Both forward and reverse, and for every switch position. A matrix has several distinct paths.
  3. After life and after environment. Repeat after the rated mating cycles or switching operations, and after thermal cycling if the assembly sees it.
  4. Assembly-level acceptance. Measure the assembled path including cables, adapters and loads. A low-PIM switch behind a standard adapter does not produce a low-PIM system.
  5. Mechanical controls. Torque wrenches, inspection intervals, dust caps and cleaning procedure. These are the cheapest PIM fixes available.

Cheapest wins on the bench

Before buying new parts: clean and re-torque every interface, replace adapters with low-PIM versions, cap unused ports properly, and re-run the two-tone test. In a surprising number of cases the offending item is a single adapter that was swapped in during debugging. If the limit only fails on one switch position, suspect that path's contact or its termination rather than the whole chassis.

A useful habit is to record the PIM figure alongside VSWR and high-band insertion loss at each calibration interval. All three drift for the same underlying reason — the mechanical interface is ageing — and tracking them together tells you which part to replace.


These notes are part of a reference series on RF component specification from DB Design / MeiXun (Wuxi) Communication Technology Co., Ltd.. For specification review or a two-tone test report, contact Yang Fei at +86-510-85380190 / +86-153-5809-7083, G1-503/703/708, 200 Linghu Avenue, Xinwu District, Wuxi, Jiangsu 214028, China.

Top comments (0)