23 cm Under Radar Pressure: Why Belgian ATV Coverage Can Collapse

Is 23 cm becoming unusable for ATV in Belgium?

The concern is that new surveillance radars will make 23 cm ATV unusable, leaving repeaters useful only to nearby stations. There is a sound engineering mechanism behind that concern: interference can remove the margin that previously supported weak, distant links. A strong local signal may continue to decode while a regional path repeatedly fails.

However, a nationwide verdict needs measurements. A radar installation, its nominal operating band and its advertised detection range do not establish the interfering power at a particular Belgian antenna. Neither do they establish a universal distance within which a repeater will work.

My conclusion is that reliable regional ATV coverage can contract severely at affected sites. “Every Belgian 23 cm repeater will only work locally” goes further than the available evidence supports. Some installations may retain useful distant paths; a badly overloaded receiver can also fail on a local one. The distinction matters when deciding whether to change a filter, move an antenna, redesign a repeater or move a link to another band.

Which radars, and will they run 24 hours a day?

The Dutch SMART-L installations are a documented development relevant to this discussion. They must be distinguished by location and operating status. Their existence alone does not identify the source of a particular Belgian interference report.

Nieuw Milligen also needs careful identification: Defence names a temporary Ground Master 400 Alpha there. A replacement radar at one site is not automatically another SMART-L, and a photograph or an unfamiliar radar signal is insufficient to assign a source.

Twenty-four-hour operation does not mean an uninterrupted carrier. A radar can remain operational all day while transmitting pulses, rotating its beam and changing modes. Conversely, a small transmit duty cycle does not guarantee usable television between pulses.

Why a radar can be relevant across a national border

A TNO document places SMART-L operation in the 1.2–1.4 GHz range, which encompasses the 1240–1300 MHz amateur band. That broad equipment range does not mean that every frequency is occupied simultaneously, or that every mode overlaps every ATV channel. TNO supplementary location study, letter page 7/11.

More troublesome reception after a radar change does not, by itself, prove that peak transmitter power increased. Pulse duration, repetition, occupied spectrum, beam scheduling and receiver recovery can change the effect substantially. Comparing only the old and new transmitters' kilowatt figures misses those mechanisms.

The relevant path runs from the radar to the affected receiving antenna. Its strength depends on terrain, antenna heights, both radiation patterns, polarization, obstructions, atmospheric refraction and the radar's instantaneous emissions. A national border contributes no RF attenuation. Elevated repeater sites and clear paths can expose receivers to distant signals; terrain screening can protect other sites.

Tropospheric enhancement can improve an amateur path and the interference path at the same time. Which improves more determines whether reception gets better. A quiet afternoon is therefore a weak basis for promising regional service throughout the year.

Do not turn a radar's aircraft or missile detection range into an ATV interference radius. Detecting a target normally involves a transmitted signal, scattering from that target and a return path. Interference at an amateur receiver can arrive directly over a one-way path, with an entirely different receiving threshold.

The interfering signal does not need a radar echo

For an unobstructed far-field path, the elementary received-power calculation is:

At 1.3 GHz and 100 km, free-space loss is about 134.7 dB. This is an arithmetic reference, not a prediction for a terrestrial Belgian path: Earth curvature, diffraction, clutter and refraction must be addressed before applying it. The usual free-space radar echo power falls approximately as distance to the fourth power; the direct one-way received power falls as distance squared.

The gains must be those in the actual directions concerned. An amateur dish pointing at a repeater may admit the radar through a sidelobe. If both lie in nearly the same direction, a narrower beam may offer little discrimination. Headline radar peak power without directional gain, bandwidth, timing and propagation is insufficient for a compatibility calculation.

Two different ways the picture disappears

Radar energy inside the wanted channel

When appreciable radar energy passes through the ATV channel filter, it directly competes with the wanted modulation. Analogue FM television may show streaks, tearing, loss of synchronization or bursts of noise. Digital ATV may exhibit transport errors, frozen pictures, audio interruptions or loss of lock. None of those symptoms uniquely identifies radar.

A linear filter cannot separate two signals that occupy the same frequencies simply because one is television and the other is radar. Narrowing the accepted bandwidth may reduce the intercepted interference, but only within the bandwidth required by the selected television mode. A notch through the wanted signal can damage its modulation and group-delay response.

The receiver is overloaded before channel selection

A strong signal outside the tuned ATV channel may compress a masthead amplifier, drive a mixer into an unsuitable operating region, trigger gain reduction or clip an analogue-to-digital converter. Several otherwise clear channels may then deteriorate together. A spectrum display can appear to show wideband interference that is partly being created inside the receiving chain.

Filtering protects the stages after the filter. A narrow demodulator filter cannot repair distortion created in an earlier wideband amplifier. A filter after the masthead preamplifier cannot stop that preamplifier overloading. This is why the receiver's complete signal path matters more than the bandwidth shown on its display. Analog Devices: receiver front-end gain, preselection and blockers.

The two mechanisms can occur together. Reciprocal mixing with local-oscillator phase noise and intermodulation can add further impairment. A low noise figure is useful, but it does not specify blocking tolerance, pulse recovery or converter headroom.

Why a low average reading can conceal a serious problem

Defence describes normal SMART-L search operation as one rotation in approximately five seconds, with transmission during about 10% of the operating time. It also describes a stationary mode with different signal behavior. These are public operating descriptions, not a complete waveform specification. Defence: how the radar works.

For equal rectangular pulses with duty fraction D, at a fixed reference plane:

At D = 0.1, the power during a pulse is 10 dB above the time average. Beam rotation, frequency changes and measurement bandwidth can further change the relationship between an instrument's displayed average and the largest input excursions. This simple calculation is not a measurement of either Dutch site.

Modern pulse-compression radar can use modulation within a relatively long transmitted pulse. The short resolved echo in the radar's matched processing does not mean an unrelated ATV receiver experiences only that short duration. Keysight: modulation and pulse compression.

Receiver recovery adds another time scale. A short RF disturbance can produce a longer video outage if gain control, carrier recovery, frame synchronization or decoding must recover. The fraction of corrupted television is consequently not equal to the radar's transmitter duty cycle. Repeating interference every few seconds can make a programme unusable even when many individual RF samples remain clean.

Why nearby stations can survive while regional coverage disappears

Let C be wanted carrier power, N receiver noise in the measurement bandwidth, and I interference in that same bandwidth and at the same reference plane. A useful first calculation is C/(N + I). Add noise and interference as linear powers before converting back to decibels.

Consider an illustrative linear receiver with N = −100 dBm. A distant signal at −86 dBm initially has 14 dB C/N. If in-channel interference during an affected interval is −90 dBm, N + I becomes approximately −89.59 dBm and C/(N + I) becomes just 3.59 dB. A local signal at −64 dBm would have 25.59 dB under the same conditions.

These invented levels explain the mechanism; they are not Belgian field measurements or universal decoder thresholds. Actual pulse interference is non-Gaussian, and loss of lock, coding, interleaving and recovery must be measured. If the receiver saturates, this linear calculation is no longer sufficient.

A simple path-loss model also shows how quickly usable distance can contract. If received wanted power falls as d⁻ⁿ and the required received level increases by Δ dB, then:

This is a sensitivity illustration with the same path model and a fixed penalty, not a coverage map. Actual interference varies with location and direction; receiver overload may produce abrupt failure instead of a smooth smaller footprint.

A repeater can therefore become effectively local for dependable ATV service without its transmitter, antenna gain or licensed power changing. Occasional long-distance contacts during quiet intervals do not restore its former service availability.

A repeater has more than one vulnerable receive path

• User to repeater: interference at the repeater input can prevent distant users accessing it. Increasing repeater output power does nothing for this failure.
• Repeater to viewer: the relay may receive its input correctly while radar affects the viewer's receiver. A clean local test card at the repeater does not establish the remote downlink.
• Inter-repeater link: a microwave backbone receiver can be the failed stage even when both local access links are sound.

Compare a known clean local test source, the repeater's decoded input, its transmitted output and observations at several receiving sites. A clean internet stream alongside a damaged RF picture can help locate a downlink problem, provided the stream and RF output actually carry the same source at that time.

Voice and narrowband data repeaters have different occupied bandwidths and failure criteria. Their survival does not prove wideband ATV remains practical. Nor does ATV failure automatically mean every narrowband mode has failed.

What reduced-bandwidth DATV can achieve

At roughly 290 K, input-referred receiver noise is approximately −174 + 10 log₁₀(B_Hz) + NF dBm. With a 2 dB noise figure, a 2 MHz noise bandwidth gives about −109 dBm; 20 MHz gives about −99 dBm. That 10 dB difference helps explain why narrowband reception and wideband television face different margins.

For raised-cosine single-carrier modulation, nominal spectral width is approximately B = R_s(1 + α). At roll-off α = 0.35, 4 Msymbol/s occupies about 5.4 MHz, while 333 ksymbol/s occupies about 450 kHz. With similarly scaled receiving filters, the latter accepts roughly 10.8 dB less white noise.

At unchanged total wanted RF power, reducing symbol rate increases energy per symbol, at the cost of data capacity. Actual performance still depends on modulation, coding, implementation and accepted video quality. Nominal occupied width and equivalent noise bandwidth are not identical quantities.

Reduced bandwidth can also avoid some interfering frequencies. It cannot guarantee rejection of a radar signal inside the remaining channel, protect an amplifier that overloads ahead of the filter, or make arbitrarily long decoding outages disappear. Forward error correction has finite capability.

What I would change at an affected station

A passive loss ahead of the first amplifier adds approximately its loss in dB to the noise figure of that cascade when referenced at standard temperature. Accepting a modest sensitivity penalty can be worthwhile if it prevents a much larger blocking penalty. That decision needs measurements.

Increasing transmitter power is constrained by authorization and only addresses the wanted signal on the relevant link. It cannot make a saturated receiver linear. An internet feed can preserve access to pictures, but it does not restore the lost RF path.

Measure the failure before buying the cure

1. Record the path and receiver configuration. Include frequency, bandwidth or symbol rate, modulation/FEC, antenna heading, preamplifier, filters, analogue gain and the level reference plane.
2. Record time behavior. Correlate received RF bursts with MER, pre/post-FEC errors, transport continuity errors, loss of lock and video recovery. Save the spectrum, detector, resolution bandwidth and sweep or capture settings.
3. Check measurement overload. A spectrum analyser or SDR can itself produce the apparent interference. Use appropriately rated input protection and attenuation, keep within peak input limits, and verify results at more than one input level.
4. Try a calibrated receive-only attenuation step. Improvement with less input is evidence of a nonlinearity or gain-control problem downstream of that point, not proof of a particular radar. In a linear co-channel-limited receiver, attenuating wanted and unwanted signals equally does not improve their ratio.
5. Compare filter placement. A filter before the masthead amplifier and the same filter after it answer different questions. Keep DC bias paths and equipment ratings correct.
6. Compare sites and days. Distinguish a repeater input failure from a viewer failure. Record the distribution of outages, including propagation changes, before promising a service radius.

A roughly five-second recurrence may be consistent with the published SMART-L search rotation, but it is not a unique identification. Use measurements and competent spectrum investigation to establish attribution. In Belgium, the BIPT and the amateur association's spectrum specialists are the appropriate channels for a documented interference report.

Belgian permissions and the separate Galileo question

The BIPT frequency table currently linked from its licensing FAQ lists 1240–1300 MHz for Class A amateur use on a secondary basis. Secondary stations must not interfere with primary services and have no protection against their interference. The table lists 200 W, with a specific 20 W ERP limit for (D)ATV in 1270–1300 MHz. ERP includes antenna gain relative to a dipole and feeder losses; it is not simply transmitter connector power. An unattended repeater must also comply with its own authorization. BIPT licensing FAQ and linked frequency table.

Separately, ITU-R M.2164 addresses protection of radionavigation-satellite reception from amateur transmissions. The BIPT published a consultation on implementing changes in March 2024. A consultation or an international recommendation should not be presented as though it were, by itself, a new Belgian station licence.

Radar interference concerns unwanted power arriving at an amateur receiver. The RNSS/Galileo issue concerns amateur transmissions reaching navigation receivers. Both constrain use of 23 cm, but their directions of interference and remedies differ. Verify the applicable BIPT decision and station conditions before changing frequency, bandwidth or radiated power.

Plan for availability, not occasional success

I would assess a regional ATV system by whether its users receive a dependable picture throughout the intended operating period. A high signal-strength reading, a successful local demonstration or a distant contact between interference bursts does not establish that.

At an affected Belgian site, radar interference can remove the regional service while leaving a strong local core. The engineering response is to identify the failing receiver, measure the interference in time and frequency, protect the vulnerable stages, and change the path or band when sufficient margin cannot be recovered.

Wier is already a day-and-night operating reality. Herwijnen is planned for continuous service, with a conditional completion timetable. Build the ATV system around that expectation; do not base its future on the radar being switched off at night.

Questions, antenna-factor records or height trials to share? Contact RF.Guru.