What you will be able to do
- Work out where intermodulation products fall for any set of carriers
- Explain why third-order products land on working channels and why equal channel spacing makes it worse
- Use the third-order intercept point to predict product levels and decide whether filtering will fix a problem
- Tell transmitter intermodulation from receiver intermodulation, and place the fix where the mixing happens
- Separate intermodulation from blocking, reciprocal mixing, adjacent-channel noise and the image response on site, and place the fix with whoever actually owns it
- Read the intermodulation, blocking and spurious response figures on a radio datasheet against the ETSI standards behind them, one of which Australia’s equipment rules adopt
- Apply the RALI LM 08 Annex D checks, and know what they assume and what they do not check
- Validate a multi-carrier frequency plan and justify the channels you propose
Syllabus
- 1 What intermodulation is Free preview
- 2 Where the products land
- 3 Product arithmetic
- 4 Levels: IP3 and the 3 to 1 rule
- 5 Level calculations
- 6 Dynamic range: SFDR and the second order
- 7 The receive chain: where noise and linearity fight
- 8 Direction: who makes it and who suffers
- 9 Desensitisation: the other ways a receiver goes deaf
- 10 Site architecture: where the fix goes
- 11 Channel sets that cannot make products
- 12 Lab: plan three clean channels
- 13 Coordination judgement
- 14 PIM in the field
- 15 What the ACMA actually requires: RALI LM 08 Annex D
- 16 Measuring intermodulation
- 17 Mixer spurs: the family that gets mistaken for intermodulation
- 18 Check: the rulebook, the bench and the hardware
- 19 The coordination record
- 20 On site: a tower walk and seven deaf receivers
- 21 Final assessment: diagnosis and practice
Read the first unit
Free, no account neededWhat intermodulation is
Intermodulation is the one interference problem a site creates entirely for itself. Nothing is faulty, every licence is in order, every radio meets its specification, and the site still cannot hear. This unit is about why that happens, and why it cannot be fixed at the receiving end.
The mechanism
A perfectly linear device produces an output that is a scaled copy of its input. Real devices are never quite linear. At high signal levels an amplifier compresses, a mixer has curvature by design, and even a metal joint can behave non-linearly. Any of these acts as a weak mixer.
Feed one signal into a non-linear device and you get harmonics, at two, three and more times the original frequency. Those are usually easy to filter, because they sit far from the wanted band. Feed in two or more signals and the device also produces sums and differences of them, and some of those land right next to, or on top of, the original carriers.
For two carriers at f₁ and f₂, products appear at every combination of:
Where the orders come from
The output of a real device can be approximated as a power series of its input:
Each term produces its own family of products. The squared term generates the second-order sums and differences. The cubed term generates the third-order products, and because it multiplies three copies of the input, those products grow on a 3 to 1 slope with drive level. Higher terms carry on up the series, each weaker than the last.
| Order | From | Example | Where it lands | Coordination risk |
|---|---|---|---|---|
| 2nd | a₂·x² | f₁ + f₂, f₂ − f₁ | Far from the band: near DC, or near twice the carrier | Low. A band-pass filter removes it |
| 3rd | a₃·x³ | 2f₁ − f₂, f₁ + f₂ − f₃ | One channel spacing outside the pair, or anywhere for a triple beat | Highest. Lands on working channels and cannot be filtered |
| 5th | a₅·x⁵ | 3f₁ − 2f₂ | Two spacings outside the pair | Real on high-power, many-carrier sites |
| 7th and up | higher terms | 4f₁ − 3f₂ | Further out again | Usually below the noise floor, but check on dense sites |
Why it cannot simply be filtered out
This is what makes intermodulation different from almost every other interference problem, and it is the reason coordination exists.
That leaves exactly three things you can do, and the rest of this module is about choosing between them:
- Stop the mixing. Keep the carriers out of each other with isolators, cavity filters and physical separation, so the non-linear device never sees two strong signals at once.
- Move the product. Change a frequency so the product lands where nobody is listening. This is what frequency planning is, and it is usually the cheapest fix at design time.
- Lower the product. Reduce the drive into the non-linear device, which works three times harder than it sounds because of the 3 to 1 slope.
Active and passive
Intermodulation splits into two families. The frequency arithmetic is identical for both, which is why this module treats them together. The causes and the cures are completely different, which is why they have separate names.
- Active intermodulation
- Mixing in a powered device driven beyond its linear range. Transmitter power amplifiers fed more than one carrier, and receiver front ends hit by strong off-channel signals, are the classic sources.
- Predictable from the third-order intercept point, which is covered later in this module. Cured with filtering, isolation and a more linear front end.
- Passive intermodulation
- Mixing in an unpowered part. A loose, corroded or magnetic joint in a connector, cable or antenna behaves like a tiny diode and mixes anything that passes through it.
- Also called PIM. Not predictable from a datasheet, because it depends on the mechanical condition of the site. Cured by finding and remaking the joint. RALI LM 08 section 6.5.1 recognises mixing in passive parts such as couplers, isolators and mechanical or structural joints.
What it costs when it goes wrong
It is worth being concrete about the consequence, because intermodulation is easy to treat as a theoretical concern until it is not. A product landing on a receive channel raises the effective noise floor on that channel. The receiver does not fail, and it reports no error. It simply stops hearing weak signals.
On a land mobile network that shows up as mobiles dropping out at the edge of coverage, which gets diagnosed as a coverage problem and answered with more sites. On a safety-critical network it shows up as a call that did not get through. The frequency plan was approved, the licence is current, and the equipment meets its specification.
Near the carriers, or far away?
Two transmitters at 450.000 and 450.025 MHz. For each product, decide where it lands. Work out m · f₁ ± n · f₂ if you need to.
2f₁ − f₂
The remaining 20 units cover Where the products land, Product arithmetic, Levels: IP3 and the 3 to 1 rule and more.
How it works
- Read each unit at your own pace, on any device. Progress is saved to your account.
- Answer the graded questions. You can re-sit any unit, and your best attempt counts.
- Do the lab exercises in the real noIM₃ tools, on real scenarios.
- Pass the module and a certificate of completion is issued with a verifiable serial.
This is professional development, not a nationally recognised qualification. noIM₃ is not a registered training organisation.