Passive intermodulation, or PIM, is often discovered after an RF design appears electrically sound. A receiver desensitizes near a transmit band, a base-station antenna fails a high-power acceptance test, or a production unit exhibits inconsistent spurious performance. Knowing how to reduce passive intermodulation requires more than selecting a low-PIM connector. It requires controlling nonlinear current paths throughout the assembled RF structure.
Key facts: PIM is generated when passive hardware develops nonlinear electrical contacts or materials under RF excitation. The highest-risk locations are usually metal-to-metal interfaces, connector mating surfaces, loose hardware, contaminated joints, dissimilar-metal contacts, and damaged plating. Effective reduction depends on disciplined mechanical design, material selection, assembly control, and measurement that reproduces the actual operating condition.
Key Takeaways
- Odd-order products fall closest to the carriers. With 1940 and 1980 MHz transmit tones, the third-order product lands at 1900 MHz and the fifth-order product at 1860 MHz, both inside a 1850–1910 MHz receive band.
- A PIM figure in dBc becomes dBm by adding the tone power: −153 dBc at 2 × 20 W (43 dBm per tone) is −110 dBm.
- At the receiver, −110 dBm of PIM raises a −104 dBm noise floor (5 MHz, 3 dB noise figure) by about 1.0 dB. A −140 dBc assembly raises it by 7.8 dB.
- PIM grows faster than the carriers. Third-order PIM rises 3 dB per dB of tone power in theory and often about 2.5 dB per dB in practice, so 3 dB more power can add 7.5 dB of PIM.
- At 1.9 GHz the skin depth in gold is 1.8 µm, so a 0.76 µm gold flash leaves about two thirds of the surface current density at the layer below. If that layer is nickel, a ferromagnetic PIM source carries RF current.
What causes passive intermodulation?
Passive intermodulation occurs when nominally passive components produce mixing products from two or more applied RF tones. The mechanism is a nonlinear voltage-current relationship at an interface, junction, or material region. Instead of passing only the original frequencies, the structure generates products such as 2f₁ − f₂ and 2f₂ − f₁. Third-order products are particularly troublesome because they can fall close to desired receive bands.
A perfectly linear conductor with an ideal contact would not create PIM. Real assemblies, however, contain microscopic asperities, oxide films, particulate contamination, plating defects, and mechanically variable pressure. At high RF current density, these features can behave as nonlinear junctions. Magnetically active materials, poor soldering, ferrous fasteners, cracked conductive coatings, and surface corrosion can add further nonlinearity.
The issue is not limited to connectors. A PIM source can reside in an antenna radiator, a feed manifold, a cable braid termination, a waveguide flange, a power-combiner joint, a mounting bracket, or a nearby metallic object carrying induced current. In a high-power system, a mechanically minor detail can dominate the measured intermodulation level. The frequency calculation below shows which of these products reach the receive band.
Which intermodulation frequencies fall in the receive band?
Every PIM product sits at a predictable frequency set by the transmit carriers. The products that matter most are the odd-order ones next to the carriers, because they land close enough to fall inside a nearby receive band:
fIM = m f1 − n f2, N = m + n, |m − n| = 1
fIM,low = f1 − ((N − 1)/2) Δf, fIM,high = f2 + ((N − 1)/2) Δf
BWIM ≈ N × BWcarrier
Where:
- f1, f2 = transmit carrier frequencies, with f1 < f2 (MHz)
- m, n = positive integers; N = product order (3 for 2f1 − f2, 5 for 3f1 − 2f2)
- Δf = f2 − f1, the carrier spacing (MHz)
- fIM,low, fIM,high = the lower and upper products of order N
- BWIM = approximate width of the product when both carriers are modulated with bandwidth BWcarrier
Worked example: two carriers in a 1930–1990 MHz transmit band
Take carriers at f1 = 1940 MHz and f2 = 1980 MHz, so Δf = 40 MHz, with a receive band of 1850–1910 MHz. The third-order product is 2 × 1940 − 1980 = 1900 MHz and the fifth-order product is 3 × 1940 − 2 × 1980 = 1860 MHz. Both fall inside the receive band. The seventh-order product, at 1820 MHz, falls below it. If each carrier carries a 5 MHz signal, the third-order product spreads over roughly 15 MHz (1892.5–1907.5 MHz), entirely inside the receive band.
| Order N | Lower product | Frequency (MHz) | Upper product | Frequency (MHz) | In the receive band? |
|---|---|---|---|---|---|
| 3 | 2f1 − f2 | 1900 | 2f2 − f1 | 2020 | Yes (lower) |
| 5 | 3f1 − 2f2 | 1860 | 3f2 − 2f1 | 2060 | Yes (lower) |
| 7 | 4f1 − 3f2 | 1820 | 4f2 − 3f1 | 2100 | No |
| 9 | 5f1 − 4f2 | 1780 | 5f2 − 4f1 | 2140 | No |
Calculated by SiberSci from fIM = m f1 − n f2.
Run this calculation for every carrier combination before choosing hardware. Moving carriers apart pushes the products further out, but it can also move a higher-order product into the band, and products from three or more carriers multiply quickly in multiband sites.
Where should a PIM investigation begin?
Begin with the complete RF current path and the product frequencies that threaten system performance. The first engineering question is not simply, “Which component has the worst PIM specification?” It is, “Where do large RF currents flow, and where can those currents encounter a nonlinear interface?”
Map the transmit tones, expected intermodulation orders, receiver bands, power levels, modulation conditions, and duty cycle. Then identify every electrically significant junction in the path. This includes intentional interfaces, such as connector and flange contacts, and unintentional ones, such as chassis seams, mounting screws, shield terminations, and conductive debris.
A useful distinction is between internally generated PIM and externally generated PIM. Internal PIM originates within the device under test. External PIM can be produced by adjacent structures illuminated by the antenna field, including loose rooftop hardware, corroded supports, metallic signage, or improperly bonded infrastructure. A low-PIM antenna can still exhibit unacceptable system-level performance if its installation environment contains nonlinear scatterers.
How do design choices reduce passive intermodulation?
The most reliable design approach is to minimize nonlinear interfaces in regions of high current density and make necessary interfaces mechanically stable, clean, and repeatable. This is a field-and-current problem as much as a parts-selection problem. Full-wave electromagnetic simulation helps identify helps identify where surface current concentrates, while practical joint design determines whether that current crosses a stable conductive path.
Use continuous conductive geometries where feasible. A monolithic transition or properly executed permanent joint may be preferable to several removable interfaces, provided serviceability and thermal expansion requirements permit it. The skin-depth section below shows why plating thickness and the layer beneath it matter.
Material compatibility matters. Dissimilar metals can form oxide layers or galvanic corrosion products that degrade contact behavior over time. Plating selection must account for conductivity, corrosion resistance, wear, mating-cycle requirements, and the base material beneath the plating. A finish that performs well in a laboratory sample may not remain stable after vibration, humidity, thermal cycling, or repeated service access. The skin-depth section below shows why plating thickness and the layer beneath it matter.
Avoid ferromagnetic materials near high-current RF paths unless their use has been specifically justified and tested. Fasteners, washers, springs, connector hardware, and enclosure parts should be reviewed rather than assumed benign. A stainless-steel part, for example, is not automatically a low-PIM choice because composition, magnetic response, plating, contact pressure, and location all matter.
Why do plating thickness and skin depth matter for PIM?
RF current flows in a thin surface layer, so the plating, and whatever lies just beneath it, carries the current that can turn nonlinear. The skin depth sets how thick that layer is:
δ = 1 / √(π f μ0 μr σ)
J(z) = Js e−z/δ
Where:
- δ = skin depth (m), the depth at which current density falls to 1/e (about 37%) of its surface value
- f = frequency (Hz)
- μ0 = 4π × 10−7 H/m; μr = relative permeability (1 for copper, silver, gold and aluminum)
- σ = conductivity (S/m)
- J(z) = current density magnitude at depth z; Js = current density at the surface
| Metal (σ, S/m) | 700 MHz | 900 MHz | 1.9 GHz | 2.6 GHz | 3.5 GHz |
|---|---|---|---|---|---|
| Silver (6.17 × 107) | 2.42 | 2.14 | 1.47 | 1.26 | 1.08 |
| Copper (5.80 × 107) | 2.50 | 2.20 | 1.52 | 1.30 | 1.12 |
| Gold (4.10 × 107) | 2.97 | 2.62 | 1.80 | 1.54 | 1.33 |
| Aluminum (3.82 × 107) | 3.08 | 2.72 | 1.87 | 1.60 | 1.38 |
| Brass (2.56 × 107) | 3.76 | 3.32 | 2.28 | 1.95 | 1.68 |
Calculated by SiberSci from δ = 1/√(π f μ0 σ) with μr = 1; conductivities as listed in Pozar (see References). Try other values with the skin depth calculator.
Two practical rules follow. First, a plating layer only isolates the metal beneath it when it is several skin depths thick: at 1.9 GHz, 3δ of silver is about 4.4 µm and 5δ is about 7.3 µm, where the current density at the base has fallen to 5% and under 1% of its surface value. Second, a thin layer does not hide a poor underlayer. A 0.76 µm (30 µin) gold flash at 1.9 GHz is only 0.42δ thick, so about 66% of the surface current density remains at the interface below it. Nickel is a common barrier layer under gold, and nickel is ferromagnetic, which is a well-documented source of PIM (Lui; Henrie et al.). This is why low-PIM connector families commonly use silver or white bronze (copper-tin-zinc) finishes instead of nickel in the current path.
Which assembly controls have the greatest effect?
Controlled cleanliness and torque frequently determine whether a sound design remains low-PIM in production. Finger oils, machining residue, fibers, oxidation, metal particles, and damaged contact surfaces can introduce nonlinear behavior. Cleaning processes therefore need defined solvents, handling methods, drying requirements, inspection criteria, and contamination controls appropriate to the materials involved.
Torque deserves similar discipline. Under-torque can leave micro-motion and insufficient contact area. Over-torque can damage threads, deform interfaces, crack dielectric supports, or disturb plating. Use calibrated tools, documented torque values, specified tightening sequences, and appropriate mating procedures. If a coaxial connector, waveguide flange, or bolted RF joint is sensitive to assembly sequence, that sequence belongs in the manufacturing work instruction.
The following comparison shows why corrective actions should be matched to the failure mechanism rather than applied indiscriminately.
| Suspected source | Typical indication | Preferred corrective action | Common trade-off |
|---|---|---|---|
| Contaminated connector interface | PIM changes after remating | Inspect, clean, and replace damaged mating surfaces | Cleaning cannot restore worn plating |
| Loose flange or fastener | PIM changes with vibration or torque | Apply specified hardware, torque, and tightening pattern | Higher preload may distort thin structures |
| Dissimilar-metal joint | Performance degrades after environmental exposure | Revise material and plating stack; improve sealing | Corrosion-resistant finishes may affect wear or cost |
| High-current seam or braid termination | PIM persists despite good connector tests | Redesign current path and bonding geometry | More permanent joints can reduce serviceability |
| Nearby passive scatterer | PIM changes with antenna orientation or site condition | Remove, bond, relocate, or replace the object | Site remediation may require access coordination |
How should PIM be measured and localized?
PIM testing must reproduce the intended use case closely enough to reveal the relevant mechanism. Two-tone frequencies, tone spacing, power level, connector type, load condition, and measurement bandwidth all affect the result. A component that passes at low power may fail at operating power because contact nonlinearity becomes more active as RF current rises.
Use a measurement configuration with verified instrument residual performance below the required limit. Characterize cables, adapters, loads, and test fixtures before drawing conclusions about the device under test. Test personnel should avoid casual retorquing or cable movement during characterization, because the setup itself can create variable PIM. The residual-error table below shows how far a test setup can shift a reading.
Localization usually requires controlled perturbation. Inspect and remate one interface at a time, apply carefully bounded mechanical stimulus where appropriate, and compare results before and after each change. Thermal stimulus can also be informative, but it must be applied with care because temperature changes alter conductivity, contact pressure, dielectric properties, and mechanical dimensions simultaneously.
For antenna systems and installed equipment, distance-to-PIM techniques, near-field investigation, substitution testing, and systematic removal of suspect external objects can help distinguish internal from external sources. The correct method depends on frequency, available access, antenna architecture, and whether the suspected source is conducted or radiated.
How do you convert a PIM specification to dBm and estimate receiver desense?
PIM specifications are written in dBc, relative to one test carrier, but a receiver responds to absolute power. Convert to dBm, compare with the receiver noise floor, and scale to the real operating power:
PPIM (dBm) = Ptone (dBm) + PIM (dBc)
N (dBm) = −174 + 10 log10(B) + NF
ΔN = 10 log10(1 + 10(PPIM − N)/10)
PPIM,2 = PPIM,1 + s (Ptone,2 − Ptone,1)
Where:
- PPIM = PIM power (dBm); Ptone = power of each test carrier (dBm), 43 dBm for the common 2 × 20 W test
- N = receiver noise floor (dBm); −174 dBm/Hz is thermal noise at 290 K; B = noise bandwidth (Hz); NF = noise figure (dB)
- ΔN = desense, the rise in the receiver noise floor (dB), treating PIM as noise-like and referenced to the same port as N
- s = PIM slope (dB per dB of tone power): 3 for an ideal third-order nonlinearity; measured values differ for each source
Worked example: a −153 dBc assembly in a 5 MHz uplink
The assembly is specified at −153 dBc with 2 × 43 dBm tones, so PPIM = 43 − 153 = −110 dBm. For a 5 MHz receive channel and a 3 dB noise figure, N = −174 + 67.0 + 3 = −104.0 dBm. The PIM is 6 dB below the noise floor and raises it by ΔN = 10 log10(1 + 10−0.6) = 1.0 dB. Now operate at 2 × 46 dBm instead of 2 × 43 dBm. With a measured slope of 2.5 dB/dB, PIM rises 7.5 dB to −102.5 dBm (−148.5 dBc), and the desense grows to 3.8 dB. With the ideal 3 dB/dB slope it reaches −101 dBm and 4.8 dB.
| PIM (dBc) | PIM (dBm) | PIM relative to noise floor (dB) | Noise-floor rise ΔN (dB) |
|---|---|---|---|
| −140 | −97 | +7 | 7.8 |
| −150 | −107 | −3 | 1.8 |
| −153 | −110 | −6 | 1.0 |
| −160 | −117 | −13 | 0.2 |
| −165 | −122 | −18 | < 0.1 |
Calculated by SiberSci. The 2 × 20 W test level is the IEC 62037-1 recommendation for mobile communication systems; the dBc-to-dBm conversion and slope behavior follow Anritsu application guidance (see References).
Two cautions apply. The noise floor and the PIM must be referred to the same point, so include any filter or cable loss between the PIM source and the receiver. And a PIM limit verified at 2 × 20 W does not cover a system that transmits more power per carrier, or more carriers, unless the slope has been measured.
Why do low-PIM specifications sometimes fail in production?
A catalog PIM value is not a guarantee of assembled-system performance. Suppliers may test under conditions that differ from the final product: different tones, lower power, fewer mating cycles, a controlled laboratory environment, or a different fixture. The final assembly can also introduce interfaces that were absent from the component qualification.
Production variation is especially consequential when the design has little PIM margin. If acceptance limits are close to the expected baseline, normal variation in plating thickness, torque, connector wear, cable handling, or environmental exposure can produce costly yield loss. The practical answer is to establish a PIM budget early, allocate margin across components and interfaces, and validate representative production assemblies rather than only engineering prototypes.
Design verification should include environmental and mechanical stresses relevant to the product: temperature cycling, humidity exposure, vibration, shock, connector mating cycles, and transport conditions where applicable. The purpose is not to test every imaginable condition. It is to identify the mechanisms most likely to change contact quality over the product life.
How do you build a PIM budget for an assembly?
A PIM budget adds the contributions of every interface and compares the total with the system limit. Because PIM sources behave like point sources separated by transmission line, their products add with a phase that depends on electrical spacing and frequency (Henrie et al.). Budget for both cases:
Prand = 10 log10(Σ 10Pi/10)
Pworst = 20 log10(Σ 10Pi/20)
e± = 20 log10(1 ± 10−Δ/20)
Where:
- Pi = PIM of each source (dBc or dBm, all at the same test power)
- Prand = total with random relative phase (power sum); Pworst = total when all products arrive in phase (voltage sum)
- Δ = how far the test system's own residual PIM sits below the reading (dB)
- e± = largest possible reading error from that residual, in phase (+) or out of phase (−)
Example: two connector interfaces at −165 dBc each and an antenna at −160 dBc. The power sum is −157.9 dBc, but the in-phase total is −153.5 dBc. Against a −153 dBc limit, the assembly passes with 4.9 dB of margin at one frequency and only 0.5 dB at another. This is one reason IEC 62037-1:2025 notes that PIM is frequency dependent and that swept or multiple fixed frequencies often give more accurate results.
The test system needs the same discipline. IEC 62037-1 calls for the setup's own residual PIM to be at least 10 dB below the specified DUT level, and even then the reading can be off by several decibels:
| Residual below reading Δ (dB) | In phase, e+ (dB) | Out of phase, e− (dB) |
|---|---|---|
| 6 | +3.5 | −6.0 |
| 10 | +2.4 | −3.3 |
| 15 | +1.4 | −1.7 |
| 20 | +0.8 | −0.9 |
Calculated by SiberSci from e± = 20 log10(1 ± 10−Δ/20).
What technical references support a disciplined PIM program?
The IEC 62037 series defines intermodulation measurement for passive RF and microwave devices: IEC 62037-1:2025 sets the general two-tone test method, IEC 62037-3:2025 covers coaxial connectors, and IEC 62037-8:2025 covers PIM generated by objects exposed to RF radiation. For the field, transmission-line, and microwave-network fundamentals that support current-path analysis, D. M. Pozar's Microwave Engineering is a standard reference. C. A. Balanis's Antenna Theory: Analysis and Design is also relevant when antenna currents, mounting structures, and radiated interactions are part of the PIM mechanism.
These references do not replace product-specific engineering. PIM behavior is highly sensitive to geometry, materials, assembly sequence, power level, and environment. A defensible program connects standardized measurement practice to the actual electromagnetic and mechanical conditions of the hardware.
Frequently Asked Questions
Can a low-PIM connector solve a PIM problem by itself?
No. A low-PIM connector can be undermined by contaminated mating surfaces, incorrect torque, damaged cable preparation, poor shield termination, or a nonlinear source elsewhere in the assembly. It is one controlled interface within a larger RF current path.
Is passive intermodulation only a concern for cellular infrastructure?
No. PIM can affect any sensitive, high-power, multi-frequency RF system, including satellite terminals, radar-adjacent systems, distributed antenna equipment, test infrastructure, defense electronics, and certain microwave assemblies. Its severity depends on frequency planning, receiver sensitivity, transmitted power, and system geometry.
What is the practical next step after identifying unstable PIM?
Preserve the as-found condition, document the test setup and stimulus response, and investigate the highest-current interfaces first. A methodical root-cause process protects schedule, avoids unnecessary redesign, and produces hardware that remains stable beyond the laboratory bench.
What PIM level is considered low?
For components tested with two 43 dBm (20 W) tones, third-order PIM below about −150 dBc (−107 dBm) is commonly treated as low PIM, and many antenna-system guidelines fall between −150 and −160 dBc. The right limit for a specific product comes from its receiver: convert the limit to dBm and keep it far enough below the receiver noise floor that the desense is acceptable.
How do you convert PIM from dBc to dBm?
Add the power of one test tone to the dBc value. With 2 × 43 dBm tones, −153 dBc is −110 dBm and −160 dBc is −117 dBm. A dBc figure is only meaningful when the test power is stated, because PIM does not scale one-for-one with carrier power.
Why does PIM change when a connector is tapped or flexed?
Tapping or flexing changes contact pressure, moves loose metal particles and opens or closes micro-gaps at a weak joint, so an unstable reading points to a mechanical interface. This is why IEC 62037 tests apply dynamic stimulus, such as the impact test for coaxial connectors in IEC 62037-3, and why test reports should include the maximum PIM recorded during the test.
Does PIM depend on power and frequency?
Yes. Third-order PIM rises 3 dB per dB of tone power in theory, and measured slopes often fall near 2.5 dB per dB, so a part that passes at 2 × 20 W can fail at higher power. PIM also varies with frequency because products from separate sources add in and out of phase, so tests should sweep the carriers or use several fixed frequencies.
SiberSci provides RF engineering, antenna engineering and microwave engineering support to locate PIM sources, review joint, plating and material choices, set PIM budgets and write test procedures that hold up in production. Contact Dr. Scott Best to discuss a PIM problem.
References
- IEC, IEC 62037-1:2025: Passive RF and microwave devices, intermodulation level measurement – Part 1: General requirements and measuring methods, ed. 3.0. Geneva: IEC, 2025. https://webstore.iec.ch/en/publication/83713
- IEC, IEC 62037-3:2025: Passive RF and microwave devices, intermodulation level measurement – Part 3: Measurement of passive intermodulation in coaxial connectors, ed. 3.0. Geneva: IEC, 2025. https://webstore.iec.ch/en/publication/83733
- IEC, IEC 62037-8:2025: Passive RF and microwave devices, intermodulation level measurement – Part 8: Measurement of passive intermodulation generated by objects exposed to RF radiation, ed. 2.0. Geneva: IEC, 2025. https://webstore.iec.ch/en/publication/83753
- P. L. Lui, “Passive intermodulation interference in communication systems,” Electronics & Communication Engineering Journal, vol. 2, no. 3, pp. 109–118, 1990. https://doi.org/10.1049/ecej:19900029
- J. Henrie, A. Christianson and W. J. Chappell, “Prediction of passive intermodulation from coaxial connectors in microwave networks,” IEEE Transactions on Microwave Theory and Techniques, vol. 56, no. 1, pp. 209–216, 2008. https://doi.org/10.1109/TMTT.2007.912166
- Anritsu, “Passive Intermodulation (PIM).” https://www.anritsu.com/en-us/test-measurement/technologies/pim
- Anritsu, “Passive Intermodulation (PIM) in In-Building Distributed Antenna Systems (DAS),” application note 11410-00871B. https://dl.cdn-anritsu.com/en-us/test-measurement/files/Application-Notes/Application-Note/11410-00871B.pdf
- D. M. Pozar, Microwave Engineering, 4th ed. Hoboken, NJ: Wiley, 2012. https://www.wiley.com/en-us/Microwave+Engineering,+4th+Edition-p-9781118213636
- C. A. Balanis, Antenna Theory: Analysis and Design, 4th ed. Hoboken, NJ: Wiley, 2016. https://www.wiley.com/en-us/Antenna+Theory:+Analysis+and+Design,+4th+Edition-p-9781118642061
Recent Comments