UHF Radio Links for GNSS RTK: Bandwidth, Modulation and Protocols Explained

UHF (403–473 MHz) — the survey standard

Almost every survey-grade RTK radio operates in UHF, typically 410–470 MHz. The wavelength of around 67 cm strikes the sweet spot between long-range propagation and a reasonably small antenna. Signals travel well over rolling terrain, partially diffract around obstacles, and penetrate light vegetation better than higher frequencies.

Practical range: 5–30 km depending on power, antenna height, terrain and modulation. Typical transmit power: 1–5 W for portable rovers; 25–35 W for fixed bases on long projects, or for everyday setups using external base radios paired with integrated rovers. Channels: survey radios operate on pre-programmed UHF channels supplied by your dealer.

900 MHz ISM — shorter range, crowded spectrum

The 902–928 MHz ISM band is historically popular for spread-spectrum survey radios. The trade-offs are shorter range (typically 1–5 km), more crowded spectrum (cordless phones, baby monitors, LoRa devices), and higher path loss through trees and structures. For most professional Canadian surveying work, UHF wins on range and stability. 900 MHz finds its niche in short-baseline UAV operations and small-fleet machine control.

GMSK — best sensitivity, slower

Gaussian Minimum Shift Keying produces a continuous-phase signal with very low out-of-band emission. It is robust at the noise floor, runs cool, and recovers weak signals well. The trade-off: lower throughput per Hz of channel.

Throughput: 4800–9600 bps in 12.5 kHz, up to 19,200 bps in 25 kHz, depending on the protocol implementation.
Choose it when: long-range or marginal-signal projects; range matters more than throughput.

4FSK — modern default

4FSK uses four discrete frequencies, encoding two bits per symbol — roughly twice the data rate of GMSK at the same protocol layer. Shorter transmit bursts mean less heat. The trade-off: slightly lower sensitivity at the noise floor than GMSK.

Throughput: 9600 bps in 12.5 kHz, 19,200 bps in 25 kHz (FEC off; FEC on reduces the rate by about 25%).
Choose it when: default for any link with healthy signal margin and modern multi-constellation corrections.

8FSK and 16FSK — more bits, less range

SATEL's modern transceivers (TR4+, EASy+) support 8FSK (three bits/symbol) and 16FSK (four bits/symbol). Each extra bit per symbol adds throughput but raises the signal-to-noise requirement sharply — moving from 4FSK to 16FSK can reduce practical range by half.

Throughput: up to 14,400 bps in 12.5 kHz, 28,800 bps in 25 kHz on SATEL EASy+ class radios.
Choose it when: short-range, data-heavy applications on a known good link. Not the right choice for marginal long-range work.

TRIMTALK 450S (Trimble)

Trimble's classic survey link protocol, named after the original TRIMTALK 450S radio modem. The protocol outlived the hardware. It runs on GMSK at 4800 bps in 12.5 kHz or 9600 bps in 25 kHz, with built-in FEC. Universally implemented as a compatibility mode by SATEL, Pacific Crest, CHCNAV — the de-facto fallback for cross-brand setups. Slow, but it works almost everywhere.

SATEL 3AS (SATEL Oy)

SATEL's flagship protocol, built around 4FSK. Faster than TRIMTALK at the same channel width and runs cooler thanks to shorter transmit bursts. Throughput: 9600 bps at 12.5 kHz, 19,200 bps at 25 kHz. Newer SATEL variants — 8FSK-1, 8FSK-2, 16FSK-1 — offer higher throughput on TR4+ and EASy+ modules at the cost of range.

Pacific Crest — many flavours

Pacific Crest (now under Trimble) defined a family of protocols: PacCrest 4FSK, PacCrest GMSK, PCC FST (Fast Synchronous Transparent), PCC EOT and EOT4. The variations exist because Pacific Crest radios were sold across many GNSS brands and had to cover every combination of modulation, FEC, and scrambling. Common ground: all are well-supported as compatibility modes by other brands when you need them.

Transparent mode

"Transparent" or "Transparent FST/EOT" passes serial data through with minimal protocol framing — lightweight, but not interoperable across brands unless both ends are configured identically (same FEC, same scrambling). Use it when both ends are the same brand or you want minimum protocol overhead.

CHCNAV — DistLink + compatibility modes

CHCNAV radios — built into the iBase, i93, i89, and i85, plus the standalone DL8 — implement TRIMTALK 450S and SATEL 3AS as compatibility modes for cross-brand work, plus CHCNAV's own DistLink protocol when both ends are CHCNAV. DistLink combines compression with optimized modulation parameters; details in its own section below.

Forward Error Correction (FEC)

FEC adds redundant bits so the receiver can reconstruct corrupted messages. Useful in marginal RF; costly in bandwidth and heat.

• Throughput cost: ~25% reduction. SATEL EASy+ at 12.5 kHz 4FSK runs 9600 bps with FEC off but only 7200 bps with FEC on, since redundant bits replace some of the data.
• Heat cost: longer transmit time = more heat. On healthy line-of-sight links, FEC off lets the radio finish each correction message faster, generating less heat.
• Turn on when: long range, marginal signal, vegetation, urban multipath, or you are seeing dropped packets.
• Turn off when: solid line-of-sight under 10 km on a healthy link, or transmitter heat is becoming a stability issue.

Scrambling

Scrambling combines the data stream with a pseudo-random pattern before transmission, then the receiver runs the same pattern in reverse to recover the original data. The practical effect on a survey radio: only rovers configured with the same scrambling setting can decode your base. Another crew running a base on the same frequency with different scrambling settings will not interfere with your rover — their packets just look like noise. That is its main day-to-day value: keeping each base/rover pair on its own logical channel even when several crews share the same frequency.

Note: this is not real encryption — it is signal separation. Anyone with the right setting (or willing to try a few common ones) can decode your stream. If you need real privacy, that is a separate feature in some radios.

The technical reasons it exists at all:

• Reliable clock recovery — the modulator never emits long runs of identical symbols, so the receiver always has bit transitions to lock onto.
• Smoother RF spectrum — energy spread evenly across the channel instead of concentrated in spurs that leak into adjacent channels.
• Consistent error rate — every part of every message has the same probability of getting through.

The catches:

• Must match on both ends — symmetric. If one end has it on and the other off, the rover decodes gibberish.
• Cross-brand mismatches happen — different vendors sometimes use different scrambler patterns inside their proprietary modes. Both radios set to "scrambling on" might still not decode each other.

What to do in the field: match the setting on both ends and leave it alone. Default is "on" — keep it that way. If a cross-brand link won't lock despite identical protocol, modulation, and FEC, scrambling is the next thing to verify.

Why power is the wrong knob

RF range in free space scales with the square root of power. Doubling power gives you ~1.4× range. Quadrupling gives 2×. To triple your range with power alone you would need 9× the wattage. What you definitely get is heat — power amplifiers are 40–60% efficient, so a 35 W radio dissipates an additional 25–50 W as internal heat. On a sunny day at 1 Hz duty cycle, the radio thermally throttles and your "35 W link" silently becomes a 10 W link.

Battery follows the same curve. A 12 V 20 Ah battery powering a 5 W base for a full workday lasts 3–4 hours at 35 W.

EIRP — the only number that predicts range

Effective Isotropic Radiated Power = transmit power + antenna gain − cable losses. Two bases with very different transmit powers can deliver identical EIRP if antennas and cabling differ.

Example: a 5 W radio with a 5 dBi 5/8-wave whip and 0.5 dB cable loss delivers ~14 W EIRP. A 15 W radio with a 0 dBi stubby and the same cable delivers ~13 W EIRP. Same range — but the first runs cooler, lasts longer on battery, and is easier on the amplifier.

VSWR — protect your radio

If the antenna is poorly matched (wrong tuning, damaged element, broken connector, water in the cable), some transmit power reflects back into the radio. The reflected energy ends up dissipated as heat in the power amplifier. At 5 W this is annoying. At 35 W with a damaged antenna, it can destroy the transmitter in minutes. Modern survey radios have foldback protection that drops power when they detect high VSWR — but if you keep ignoring "low output" warnings, eventually the protection circuit gives up.

The single best thing you can do for radio longevity is check antenna connectors before each project.

Antenna types in survey RTK

• Quarter-wave whip (~17 cm) — the stubby on rover receivers and small handheld radios. Compact, rugged. Needs a ground plane to work properly. Real-world gain ~0–2 dBi. Fine for rover use, underwhelming as a base antenna.
• Half-wave whip (~33 cm) — does not need a ground plane, flexible for tripod and mast mounting. ~2–3 dBi. The default "good base antenna" for most surveying work.
• 5/8-wave whip (~40 cm to 1 m+) — the longer whips you see on professional base stations (GAT24-class, Maxrad, Larsen, Diamond). Concentrates more energy toward the horizon. ~3–5 dBi.
• Yagi (directional) — 8–12+ dBi gain, narrow forward beam. Use on linear projects where the rover stays in a known direction. Requires aiming.

Same wattage, different antenna — what actually changes

Every 3 dB of antenna gain doubles your effective radiated power. Every 6 dB doubles your range in free space. So at the same 5 W transmit power:

A 5 W base radio with a good 5/8-wave whip on a mast can match or outperform a 15 W base with a stubby antenna — without the heat, battery drain, or amplifier strain. This is also why CHCNAV designs the iBase with a 5 W internal radio: paired with a proper external whip on the tripod, real-world range routinely matches or exceeds higher-power radios that ship with built-in stubbies.

Polarization — the easy way to lose 3 dB

UHF survey radios are vertically polarized. Whips on both base and rover need to point straight up. A 30° tilt costs ~1 dB; 60° costs ~6 dB; 90° (one vertical, one horizontal) drops the link entirely. A bent rover whip in transit, a base antenna leaning on a tripod, or a Yagi mounted slightly off-axis are all silent range-killers. Check antennas are plumb-vertical before troubleshooting anything else.

Cable losses — the silent power sink

Coaxial cable between the radio and the antenna eats some of your power before it reaches the air. At 450 MHz, losses vary dramatically by cable type:

• RG-58 (thin): ~30 dB per 100 m → ~1.5 dB lost per 5 m run. Acceptable for short jumpers; brutal on long mast feeds.
• RG-213 (thicker): ~14 dB per 100 m → ~0.7 dB per 5 m. Reasonable balance.
• LMR-400 (low-loss): ~9 dB per 100 m → ~0.45 dB per 5 m. The right choice for any mast install longer than a couple of metres.

For long mast runs, a cable upgrade often pays back more than a power upgrade.

Height beats power, every time

UHF range is dominated by line-of-sight. The horizon distance for an antenna at height h meters is roughly 4.12 × √h kilometers — so an antenna at 2 m sees 5.8 km, at 5 m sees 9.2 km, at 10 m sees 13 km. Both ends elevated combine. Raising your base antenna from a 2 m tripod to a 5 m mast typically extends range more than tripling your transmit power.

Repeaters and the latency cost

A repeater receives the base signal and retransmits it on the same or a different frequency, hopping over terrain or bridging to rovers beyond direct range. Each repeater adds latency — the repeater has to receive the full packet, then retransmit it — typically several hundred milliseconds per hop, more if the over-the-air rate is slow. Two repeaters in series can push latency past the 2-second age-of-corrections budget on slower channels. Prefer one hop with a high antenna over two hops with low ones. Repeaters cannot change protocol or data rate — keep the over-the-air rate consistent through the chain.

What DistLink does

DistLink combines a tighter compression algorithm for GNSS differential corrections with optimized modulation parameters. CHCNAV's published numbers:

• Up to 30 km coverage on a 5 W transmitter — comparable to a 25 W radio running standard protocols
• ~30% improvement in rover radio sensitivity through reduced bandwidth requirements
• 30% lower radio power consumption, enabling 13+ hours of continuous base operation without external batteries

DistLink only works between CHCNAV base and rover. For mixed-brand fleets the iBase falls back to TRIMTALK 450S or SATEL 3AS compatibility modes.

Ultra Base mode — radio + cellular failover

The updated iBase supports "Ultra Base" mode, transmitting corrections simultaneously over UHF radio and over cellular network using the built-in 4G modem. The rover automatically falls over to network corrections when it moves out of radio range and back to radio when it returns. For surveyors working long linear corridors with patchy cell coverage, this eliminates the choice between radio and NTRIP.

One-button startup

The April 2026 firmware added a one-button auto-pairing sequence: power up the iBase, hold the button, and it auto-discovers the matching CHCNAV rover and configures channel, modulation, and protocol with no manual setup. Removes the single biggest source of "I cannot get a fix" frustration: misconfigured radio settings.

Rover stays on float past 5–8 km

• Check antenna height on the base — line-of-sight failure is the #1 cause
• Check antenna, cable and connectors — a damaged whip, bent connector, or corroded coax can cost more signal than doubling your transmit power would gain
• Signal is too weak at distance — try GMSK if you're on 4FSK; GMSK is more robust at the edge of usable range
• Channel may be saturated — switch to a compressed format if both ends support it (DistLink, CMRx, Leica 4G), which roughly halves the bandwidth needed

Connected but no fix

• Verify protocol matches on both ends — TRIMTALK 450S, SATEL 3AS, and Pacific Crest are different protocols and the radios will not decode each other unless both are set to the same protocol or a compatibility mode
• Check channel frequency on both ends (radios sometimes step in 2.5 kHz increments that look identical on the menu but differ in hardware)
• Verify FEC and scrambling settings match exactly on both ends
• Confirm correction data format matches — RTCM 3.2, CMRx, DistLink, Leica 4G are all different and the rover has to be set to decode whatever the base is broadcasting. See our corrections guide for the format compatibility details

Fix oscillating between fixed and float on a "connected" link

• Classic symptom of a saturated channel — corrections piling up in the radio buffer past the rover's age-of-corrections threshold
• Check for poor sky conditions at base or rover — heavy canopy, buildings nearby, or low satellite count means few common satellites between base and rover, which causes the rover to drop in and out of fix even with a healthy radio link
• Switch to a compressed format if both ends support it (DistLink for CHCNAV-to-CHCNAV, CMRx between Trimble units, Leica 4G between Leica units) — roughly half the bandwidth of equivalent RTCM
• Move from GMSK to 4FSK if signal margin allows — doubles channel throughput

Base radio overheating or battery dying fast

• Drop transmit power — 35 W is rarely necessary; 10 W is usually plenty for <10 km in line of sight
• Switch GMSK to 4FSK if signal margin allows
• Turn FEC off if you have a healthy link — meaningfully less transmit time and heat
• Switch to a compressed data format if both ends support it (DistLink, CMRx, Leica 4G) — shorter packets mean shorter transmit bursts and less heat

Cross-brand setup will not link at all

• Before configuring anything, check what protocols, modulations, and data formats each radio actually supports — the base can only transmit what it has, and the rover can only decode what it has. Match both ends to settings that exist on both sides
• Start with the slowest universal setting: TRIMTALK 450S, GMSK, 4800 bps, 12.5 kHz, FEC off, scrambling off. Fewer features active means fewer settings that can mismatch
• Set the correction data format (RTCM 3.x is the most universal) — both base and rover must agree
• Once a fix is locked, optimize upward toward 4FSK and higher rates one variable at a time
• If the rover supports it, enable "Auto detect" or "Auto Rx" mode to let it discover the base configuration