12.5 kHz Apart, 50 km Too Close

Put two high-site 70 cm repeaters less than roughly 50 km apart and separate their output frequencies by only 12.5 kHz. When they interfere, the familiar proposal is: “add better filters.” That proposal misunderstands where the problem lives.

A duplexer can separate a repeater’s own transmitter and receiver when their frequencies are megahertz apart. A band-pass cavity can reject signals well outside its wanted passband. Neither can create unlimited rejection only 12.5 kHz from a wanted narrowband FM carrier while leaving that wanted carrier, its modulation and the receiver sensitivity untouched.

Filters can be part of a properly coordinated system. They cannot repair a fundamentally bad adjacent-channel assignment by themselves.

Scope: this article concerns two analogue FM 70 cm repeaters on different but overlapping sites, with adjacent output or input channels separated by 12.5 kHz. It does not claim that every pair of 12.5 kHz channels is unusable. Professional narrowband systems can reuse adjacent channels when equipment, geography and coordination provide enough isolation.

First Correction: 12.5 kHz Is Channel Spacing, Not the Repeater Shift

The words shift, offset and spacing are often mixed together.

• Duplex shift or offset is the frequency difference between a repeater’s input and output. On 70 cm this is normally measured in megahertz and depends on the regional or national plan.
• Channel spacing is the frequency difference between neighbouring assigned channel centres. Here that difference is 12.5 kHz.

The current IARU Region 1 VHF Handbook includes 12.5 kHz-spaced repeater channel structures within parts of the 430–440 MHz band. Actual assignments remain a national and coordination matter.

A 12.5 kHz channel plan is therefore real and legitimate. But a channel raster is not a promise that any two adjacent high-site repeaters can be placed wherever convenient. Frequency coordination must still account for coverage overlap, wanted-to-unwanted signal ratios, equipment performance and terrain.

Why “Less Than 50 km” Is a Warning, Not a Law

Fifty kilometres is not a magic RF boundary. A hill can isolate two sites only 15 km apart. Two elevated sites with a clear path can interfere well beyond 50 km, especially during tropospheric enhancement.

The useful planning question is not simply distance. It is the unwanted signal level at the victim receiver and throughout the victim repeater’s service area.

Free-space path loss gives a useful lower-complexity reference:

Consider an illustrative 25 W repeater transmitter: 44 dBm at the connector. After feed-line loss and antenna gain, an EIRP around 48–52 dBm is quite plausible. With a reasonably clear 50 km path and gain at the receiving site, the adjacent repeater can arrive near −65 to −60 dBm.

A weak mobile trying to access or hear the wanted repeater may be close to −115 dBm or lower at the receiver input. The adjacent repeater can therefore be 50 dB or more stronger than the wanted signal. At shorter distances or between better sites, the difference can be much larger.

This is why sensitivity alone does not describe a repeater receiver. The critical question becomes whether it can preserve a weak wanted signal while a very strong transmitter operates only one 12.5 kHz channel away.

What the 12.5 kHz Standard Actually Assumes

The European narrowband equipment benchmark is useful because it puts real numbers around the problem. ETSI EN 300 086 V2.1.2 specifies, for equipment intended for 12.5 kHz channel separation:

• a maximum permissible FM deviation of ±2.5 kHz;
• controlled audio-frequency response above 2.55 kHz;
• adjacent-channel transmitter power no greater than 60 dB below conducted carrier power, subject to the standard’s stated floor;
• receiver adjacent-channel selectivity of at least 60 dB under normal test conditions; and
• receiver adjacent-channel selectivity of at least 50 dB under extreme test conditions.

These are equipment limits under defined laboratory conditions. They are not an inter-site coordination guarantee.

If the adjacent repeater arrives 60 dB stronger than a weak wanted signal, a receiver that just meets a 60 dB adjacent-channel figure has no comfortable system margin. Real FM modulation, frequency error, temperature, duplexer loss, antenna patterns, receiver blocking, oscillator phase noise and propagation variation still have to be accommodated.

Why a Duplexer Does Not Solve Adjacent-Channel Interference

A duplexer is designed primarily to let one repeater transmit and receive simultaneously through one antenna. It separates the repeater output from its own input, normally across a duplex offset measured in megahertz.

That is a very different task from separating two carriers only 12.5 kHz apart.

A good duplexer can provide substantial isolation between the repeater’s own TX and RX frequencies while maintaining acceptable insertion loss. Its cavity response does not suddenly become a brick wall between two adjacent narrowband channels.

Adding more ordinary cavities may improve far-out rejection, transmitter noise at larger offsets, or protection against unrelated services. It does not create an arbitrarily narrow transition from “pass my full FM signal” to “reject the next carrier by 60 or 80 dB” within only a few kilohertz.

The Loaded-Q Problem at 435 MHz

The scale of the problem becomes clearer with a simple resonator model.

A single resonator with that loaded Q would provide only modest attenuation at the adjacent channel centre. A simplified resonator response is:

Using that model, roughly 30 dB attenuation at an offset of only 12.5 kHz would require a loaded Q around 550,000. Its corresponding 3 dB bandwidth would be below 1 kHz. A wanted analogue FM repeater signal would not pass through that undamaged.

Multi-pole filters can make skirts steeper than one resonator. Cross-coupled and quasi-elliptic designs can create transmission zeros. But they do not repeal the passband-to-stopband trade-off. Very narrow UHF filters bring:

• insertion loss;
• amplitude and group-delay distortion;
• extreme sensitivity to tuning and temperature;
• limited power handling or physically large resonators;
• tight manufacturing tolerances; and
• a passband that can become narrower than the wanted FM emission.

The problem is not that filter designers lack skill. The problem is that the desired signal and the unwanted adjacent signal occupy neighbouring slices of spectrum with only a small transition region between them.

A Notch Filter Is Not a Free Escape

A natural response is to place a deep notch exactly on the other repeater’s carrier.

At 12.5 kHz separation, that notch must be exceptionally narrow and stable. It must reject the adjacent carrier and its modulation products while leaving the wanted carrier’s modulation sidebands, amplitude response and group delay acceptable. At the transmitter, it must also handle continuous RF power without drifting.

Even a perfect carrier notch would not remove:

• the adjacent transmitter’s modulation sidebands;
• close-in phase noise;
• receiver reciprocal mixing;
• front-end blocking before the narrow filter;
• intermodulation generated in overloaded active stages; or
• wide-deviation user transmitters on the repeater input side.

A laboratory notch can make a spectrum plot look impressive. A reliable, temperature-stable, high-power, field-serviceable solution across two complete repeater systems is a different engineering task.

The Problem Exists at Both Transmitters and Both Receivers

Adjacent-channel compatibility is a four-sided problem:

System element	Failure mechanism	What must be controlled
Repeater A transmitter	Modulation sidebands, over-deviation, hum, noise and close-in phase noise reach B’s channel.	Deviation, audio limiting, adjacent-channel power, oscillator cleanliness and PA linearity.
Repeater B receiver	A’s strong adjacent carrier causes selectivity failure, reciprocal mixing, blocking or intermodulation.	Adjacent-channel selectivity, early filtering, dynamic range, LO phase noise and gain distribution.
Repeater B transmitter	The same mechanisms interfere with A in the opposite direction.	The same transmitter cleanliness requirements.
Repeater A receiver	The same strong-signal mechanisms occur in reverse.	The same receiver performance requirements.

Cleaning only one transmitter or narrowing only one receiver does not make the assignment symmetrical. Unless the path itself is strongly asymmetric, both repeaters must meet the coexistence conditions.

Why “Fixing It” Can Cripple Both Repeaters

If frequency coordination is not changed, enough isolation can sometimes be forced by reducing what each repeater is allowed to do. But that is not a free filter upgrade. It is a system compromise imposed on both sides.

Reduce FM deviation and audio bandwidth

Lowering deviation and aggressively filtering audio reduces occupied bandwidth. Push it too far and the repeater becomes quiet, thin-sounding and incompatible with normally configured narrowband users. Speech intelligibility can suffer, especially in mobile noise.

Narrow both receiver IF passbands

A tighter IF filter can improve adjacent-channel rejection, but it also clips wanted modulation, increases sensitivity to frequency error and can add group-delay distortion. Both repeaters and their user population must remain accurately on frequency.

Reduce receiver sensitivity

Input attenuation, higher squelch thresholds or reduced front-end gain can improve strong-signal survival. They also make weak users disappear. A repeater that no longer hears its intended coverage area has been protected by surrendering the reason it exists.

Reduce transmitter power, antenna gain or antenna height

Lower ERP and controlled antenna patterns can reduce the signal delivered into the other repeater’s area. They also reduce wanted coverage. In some cases this is good coordination; in others it simply turns two regional repeaters into two smaller repeaters.

Add lossy, ultra-narrow filter networks

Insertion loss reduces transmit power reaching the antenna and raises the receive system noise figure by the same amount when placed ahead of the first amplifier. The filter may improve one interference ratio while simultaneously weakening every wanted signal.

CTCSS and DCS Do Not Create RF Isolation

Different CTCSS tones or DCS codes can prevent an unwanted signal from opening a squelch or keying a repeater. They do not remove that signal from the receiver input.

A strong adjacent carrier can still:

• desensitise or block the receiver;
• raise the recovered noise;
• create reciprocal-mixing noise;
• generate intermodulation; or
• mask a weak wanted user.

Tone access manages logic. It does not improve adjacent-channel selectivity or dynamic range.

What Filters Are Actually Good For

This is not an argument against filters. Repeaters need good filtering.

Proper filters can:

• isolate a repeater’s own transmit and receive paths across the duplex offset;
• suppress harmonics and far-out transmitter products;
• reject out-of-band broadcast, paging, mobile or link transmitters;
• reduce noise from transmitters several channels away;
• protect receiver front ends at shared sites; and
• support multicoupler and combiner systems.

Those are all valid tasks because they provide a usable transition band between wanted and unwanted spectrum. The 12.5 kHz adjacent-channel case offers very little transition space.

The Engineering Fixes That Actually Scale

A robust solution starts with coordination, not a shopping list of extra cavities.

1. Choose non-adjacent channels where coverage substantially overlaps. Frequency separation is the cleanest selectivity.
2. Model and measure the inter-site path. Include antenna patterns, feeder loss, terrain, site height and enhanced-propagation conditions.
3. Control service areas deliberately. Use suitable ERP, antenna height, downtilt and directional patterns rather than maximum gain everywhere.
4. Use genuinely narrowband transmitters. Verify deviation, audio response, adjacent-channel power and close-in noise under real modulation.
5. Use receivers with measured close-in performance. Sensitivity is not enough; test adjacent-channel selectivity, blocking, reciprocal mixing and intermodulation at realistic levels.
6. Verify every user path. Wide-deviation mobiles and handhelds can defeat a narrowband repeater plan on the input frequencies.
7. Perform simultaneous on-air testing. Test weak wanted signals while the neighbouring repeater transmits at normal duty cycle.

If the only way to pass those tests is to reduce both repeaters until neither retains its intended coverage and audio performance, the assignment is wrong.

A Practical Coordination Test

For each direction, build a wanted-to-unwanted budget at the victim receiver:

The exact form depends on which mechanism is being tested, but the discipline matters. Use measured values rather than catalogue optimism:

• actual transmitter adjacent-channel power and phase noise;
• actual receiver selectivity, blocking and intermodulation;
• actual duplexer and preselector insertion loss;
• actual antenna discrimination in the relevant directions;
• actual site-to-site and service-area signal levels; and
• margin for temperature, ageing, mistuning and propagation enhancement.

A paper plan with zero margin is already a failed field plan.

Common Claims Versus RF Reality

Claim	RF reality
“They are on different frequencies.”	They are only one 12.5 kHz channel apart. The wanted modulation, transmitter skirts and receiver transition band are immediately adjacent.
“Add a bigger duplexer.”	A duplexer solves TX/RX separation across the duplex offset. It is not a magic adjacent-channel brick wall.
“Use different tones.”	CTCSS/DCS controls access logic, not RF overload or adjacent-channel rejection.
“Both radios meet ETSI, so they must coexist.”	Equipment compliance supplies test limits. It does not provide the geographic isolation or system margin for a particular pair of sites.
“Fifty kilometres is far enough on UHF.”	High 70 cm sites can have strong line-of-sight or enhanced-propagation paths at and beyond that distance.
“Just narrow the filters.”	Narrowing enough to obtain large rejection only 12.5 kHz away also removes wanted deviation, audio bandwidth, sensitivity or power.

In Summary

Two 70 cm repeaters can use 12.5 kHz channel spacing successfully when the complete coordination plan provides enough isolation. That may come from geography, non-overlapping coverage, clean narrowband equipment, directional antennas, controlled ERP and high-performance receivers.

What does not work is placing two strong, overlapping high-site repeaters less than roughly 50 km apart, assigning adjacent channels, and expecting ordinary cavities or duplexers to erase the conflict.

The unwanted carrier is too close to the wanted modulation. The receiver has finite adjacent-channel selectivity. The transmitter has finite adjacent-channel power and phase noise. The path can make the unwanted signal tens of decibels stronger than the weak wanted user.

To force that pair to coexist, both sides may have to sacrifice deviation, audio bandwidth, receiver sensitivity, transmit power, antenna gain or coverage. At that point the filters have not fixed the frequency plan. They have merely made both repeaters smaller and worse.

The durable fix is coordination with margin.

Want more technical RF content? Subscribe for new deep dives and lab notes.

Have a question or field observation? Contact RF.Guru.