Garmin Edge for Sailing: A Cross-Over Research Note

This is a research note, not a rated review.

A Garmin Edge is a cycling GNSS receiver: it resolves your velocity in the earth-fixed frame — speed and course over the ground — and measures nothing about the water you are sailing through. That makes it a genuinely capable portable logger and a serviceable backup clock, and structurally excludes it from the true-wind and VMG loop a race boat runs. This note is honest about a crossover device at the level of the actual signal chain: how it derives its numbers, why the water frame is the one that matters, and how we would configure and test one. For the full picture of what a race boat needs, start with our race-boat electronics guide.

What an Edge actually measures, and how

Modern Edge units (the 840, 850 and their siblings) run a multi-band GNSS receiver: they track GPS, GLONASS, Galileo, BeiDou and QZSS on two frequencies simultaneously — L1 and L5. The value of the second band is not marketing. L1-only receivers carry ionospheric range error and, worse afloat and in urban canyons, multipath — reflected copies of the signal arriving microseconds late. Because the ionospheric delay scales with the inverse square of carrier frequency, an L1/L5 pair lets the receiver estimate and remove most of that delay, and the L5 signal's higher chipping rate and longer codes sharpen the correlation peak so reflected energy is more easily rejected. In independent cycling testing this collapses horizontal error from the 10–30 m typical of single-band units in hard environments to roughly 1–4 m. Open water is a benign sky by comparison, so on a boat the fix is clean and stable — the one thing an Edge does unarguably well.

The number sailors actually watch, though, is speed, and here the physics is better than most crossover discussions assume. A quality GNSS receiver does not compute SOG by differencing two positions and dividing by time — that method inherits twice the position error over a short interval and is badly noisy. It derives velocity from the Doppler shift of the carrier: the line-of-sight rate to each satellite is read directly from the frequency offset, and the receiver solves the over-determined set for a 3-D velocity vector. In open sky that yields velocity noise on the order of a few centimetres per second; carrier-phase time-differencing (TDCP) can reach millimetres per second. So instantaneous SOG from an Edge is genuinely precise — the limitation is not accuracy, it is what frame the velocity is measured in.

From that solution it exposes three quantities that matter afloat:

• Speed over ground (SOG) in knots, once units are set.
• Course over ground (COG) in degrees — the direction the boat is actually tracking over the seabed, not where the bow points.
• A logged GNSS track for the session, exportable as GPX/FIT afterwards.

Add the built-in lap and timer fields — elapsed time, lap distance, lap speed — and you have the raw material for a training log or a rough start sequence. That is the honest extent of it.

Why "over the ground" is the whole story

Everything an Edge cannot do for racing follows from one fact: a GNSS receiver only observes motion in the earth-fixed (ground) frame. It has no sensor in the water. Racing lives in the water frame, for two reasons. First, the foils generate lift against the water, so it is speed through the water — STW — that sets the operating point of keel and rudder. Second, and decisively, true wind is a water-frame quantity. Apparent wind is the vector sum of true wind and the boat's velocity; to recover true wind you subtract the boat's velocity from apparent wind, and the boat velocity you must subtract is its velocity through the water, because the air mass and the water move together as the tidal stream. In scalar form:

TWS = √(STW² + AWS² − 2·STW·AWS·cos(AWA))

with true wind angle following from the same triangle. The input is STW, not SOG.

Now introduce a tidal stream — the current vector (set and drift). It displaces the ground frame from the water frame by exactly that vector, and it does so in direction as well as magnitude. A 1.5-knot cross-tide on a 6-knot boat swings COG several degrees off heading and changes SOG relative to STW; a fair or foul stream on the nose adds or subtracts drift directly. The practical failure is stark: feed SOG into the true-wind equation instead of STW and you inject the entire current vector into the wind solution. In a 2-knot tideway that is a large, systematic error in both true wind speed and — the number that decides tactics — true wind direction, so any shift the device appears to show is contaminated by tide, not a genuine change in the air. That is not a calibration problem an Edge can be tuned out of; it is missing an input it has no sensor for.

There is a second, subtler limitation even for pure boat-speed work: filter latency. A cycling GNSS solution is smoothed — Kalman-filtered or otherwise low-passed — and typically surfaced at 1 Hz (5 Hz on the fastest Edge profiles). Any smoothing imposes group delay, so displayed SOG lags real acceleration by a fraction of a second and softens the peaks. A racing instrument system runs its water-frame speed and heading at 10 Hz or faster with tight filtering precisely so a trimmer sees the boat load and unload in real time through a wave or off a wind pressure. The Edge's number is precise but damped and late for high-frequency speed work.

This is exactly why a serious one-design carries a paddlewheel (or an electromagnetic/ultrasonic log) for water-frame STW and a rate-compensated fluxgate compass for heading, feeding a processor that resolves true wind and VMG. No GNSS receiver — cycling or marine — substitutes for that chain; see how the proper stack fits together in our race-boat electronics guide.

Where it earns its place

Given all that, an Edge still earns a strap on the rail in three roles:

Training and delivery logging. For a tuning session or a passage, a continuous, precise SOG/COG track you can review afterwards is genuinely valuable — pace across a leg, where the boat bled speed, how a sail change or trim move landed in the numbers. It is a data recorder that happens to fit in a pocket, and its Doppler-derived speed is trustworthy in slack water where SOG and STW nearly coincide.

A backup start clock. The lap button gives a countable count-up and a leg timer. It is not a synchronised race sequence and it computes no distance-to-line, but if a dedicated start timer fails, an Edge on the mast beam is a serviceable fallback for a rolling count.

A free second opinion on SOG. When the boat's instruments are being serviced, or you are sailing a boat without them, an Edge confirms you are moving and how fast over the ground to a fraction of a knot — the same limited-but-real value an Apple Watch offers, in a more rugged package with a far better antenna.

What it does not add is anything tactical: no true wind, no laylines, no line-sight to a committee boat, no depth, no target-versus-actual against a polar. Judge it against a phone in a pocket, not against a processor.

Setting one up so it is actually usable

The default cycling screens are useless afloat. Build one large-font page and strip everything else:

• SOG in knots and COG in degrees as the two hero fields — set units to knots and degrees first, or you will read kilometres per hour and misjudge every number.
• A lap timer and lap distance for start and leg work.
• Disable auto-pause. It is designed to stop the clock when a cyclist halts at lights; slowing head-to-wind before a start, it will freeze the timer at the worst possible moment.
• Force multi-band (L1+L5) GNSS on. The battery cost is trivial next to a clean fix, and the newest silicon holds a stable solution close to structures — including your own rig and body.
• Set the highest recording rate available — 1-second, or 5 Hz on an 850 — so the logged track resolves tacks and gybes instead of straight-lining through them. At 1 Hz a boat covers several metres between fixes, enough to blur a manoeuvre in post-analysis.

On the display: the newer, brighter panels are easier to read in glare but more emissive/glossy, so they can wash out or throw reflections at a low sun angle. The transflective memory-in-pixel (MIP) displays used across the outdoor range behave oppositely — they are reflective, so the brighter the ambient light the more readable they get, at near-zero power, which is ideal for a deck in sun; their weakness is that they dull when your own body shadows the screen and they are poor in low light. Mount for an unshaded sightline either way. We would assess real cockpit readability, at a heeled angle with spray on the lens, before trusting it in a hurry.

The marine-environment reality

An Edge is rated IPX7 — immersion to one metre for thirty minutes, in fresh, still water. Three caveats matter afloat. First, the test uses freshwater; salt is not in the standard, and saltwater is more conductive and markedly more corrosive, so a device that passes IPX7 is not certified against the medium it will actually meet. Second, IPX7 says nothing about high-pressure jets or sustained immersion — that is IPX6/IPX8 territory, built with multiple seal barriers, pressure-compensated glands and corrosion-grade materials such as 316L stainless; a green-water hit over the rail can exceed a static one-metre soak. Third, ingress protection is a depreciating asset: the rating assumes fresh, uncracked gaskets, and it degrades with every thermal cycle, UV hour and knock.

The failure mechanism to design against is dried salt. When saltwater evaporates inside a seal or a button stack it leaves hygroscopic, conductive crystals that draw moisture back, bridge contacts and drive galvanic corrosion at every dissimilar-metal junction — the charging pins first. Practically: mount low and central where the boat throws least water; tether it independently of the quarter-turn mount (there is no marine lanyard point, so improvise a load-rated one); and rinse it in fresh water and dry it after every session, exactly as you would any deck electronics — our salt-corrosion notes apply. Treat a green, crusted charging port as the early-stage failure it is, not a cosmetic one.

What good and bad look like

Good: a multi-band fix acquired in seconds; a Doppler-derived SOG that tracks the boat's own log within a fraction of a knot in slack water; a screen legible at arm's length at a heeled angle; and a full session logged at 1-second or 5 Hz and exported clean. In that state it is a tidy, honest training tool doing exactly what its signal chain permits.

Bad — and mostly not the device's fault: trusting SOG as boat speed in a tideway; reading a "wind" figure the device cannot resolve because it lacks STW; an auto-paused timer at the gun; a shaded-out MIP screen or a glare-blinded emissive one; damped SOG mistaken for real-time speed in a lull; or a salt-seized unit that never got rinsed. Every one of these is a setup or expectation error. The hardware is sound; the temptation to ask it for water-frame answers it structurally cannot give is the risk.

On a Grand Prix one-design

For a campaign like a Melges 40, the conclusion is unambiguous. The boat runs a proper instrument system — water-frame STW from a keel-mounted paddlewheel (sited in the keel specifically to avoid the ventilation and aeration that plague hull transducers on a fast, hard-driven boat), heading from a rate-compensated compass, apparent wind from a masthead unit, and true wind and VMG resolved by a dedicated processor and shown on a masthead-legible display — because winning turns on quantities a GNSS receiver structurally cannot provide. (Treat the specific sensor and processor configuration as needing verification against the class rules and the boat's own instrument documentation.) An Edge has no role in the tactical loop and must never be wired to pretend otherwise.

Where it can sit comfortably is off to the side: a coach or crew member logging a training day, a spare count-up in a pocket, a quick SOG check on a RIB or a delivery leg. If someone on the programme already owns one for cycling, that is a free, rugged, superbly-positioned logger to exploit — nothing more, and nothing to apologise for. Just do not let a €600 cycling computer stand in for the instrument budget a serious campaign has to fund. When you are ready to build that stack properly, our race-boat electronics guide is the place to start.