RTK vs PPP: instant centimeters vs patient decimeters

Every precision GNSS product is built on one of two correction approaches. They differ in accuracy and in cost, and — most importantly, and least advertised — in how long they take to get accurate. Almost every disappointed precision-GPS buyer bought the wrong one of these two for the work they actually do.

This guide explains both in plain English, states what we measured with each on the same vehicle, and gives you a way to tell which one your work needs.

What both of them are fixing

A satellite navigation fix is a timing problem. The receiver works out how long each satellite’s signal took to arrive, and turns those travel times into a position. Anything that corrupts the timing corrupts the position, and there are four culprits worth knowing:

• Satellite clock and orbit error. The satellite’s idea of where it is and what time it is are both very good and not perfect.
• The ionosphere. A layer of charged particles that slows the signal down by an amount that varies with time of day, solar activity, and your latitude. This is the largest error source and the hardest one.
• The troposphere. Weather — water vapour in the lower atmosphere, bending and delaying the signal.
• Multipath. Signal bouncing off a shed roof, a grain bin, or a tree canopy and arriving twice.

Together these leave an uncorrected receiver at roughly 3–5 metres — fine for finding a road, useless for finding last year’s row. Both RTK and PPP exist to measure those errors instead of suffering them. They just measure them in completely different ways, and that difference is the whole story.

RTK: instant centimeters, needs a data link

Real-Time Kinematic works by comparison. A reference receiver at a known position sees the same satellites you do, through very nearly the same slice of atmosphere. Because its true position is known, every discrepancy in what it observes is error — and that error is almost identical to yours. Subtract it and most of the error budget disappears at once, without anyone having to model the ionosphere from first principles.

Classically the reference is a base station: your own tripod, surveyed in, with a radio link to the rover. Modern network RTK — including Northing RTK — replaces the single base with a modelled correction stream computed across a whole continent and delivered over the internet, so there is nothing on your side to survey in, power, or maintain.

The property that matters is that RTK does not converge; it fixes. The receiver resolves the whole-cycle ambiguities in the carrier phase — the “how many wavelengths away am I” question — and the moment it does, you are at centimeters. It is a step change, not a slope. You will see it in the app as the fix type going from float to fixed, usually within seconds.

What we measured, on logged dual-receiver drive sessions:

• Fixed within seconds of a cold start. No ten-minute warm-up.
• ±2.5 cm repeatable accuracy while fixed, open sky.
• Fixed for 88% of a 40-minute drive that deliberately included tree cover, a bridge, and buildings.
• 1–2 second recovery after each obstruction.

The price of all this is the data link. RTK cannot work without a live correction stream, and if the stream stops, the accuracy decays. How gracefully it decays is a real engineering difference between products — see what breaks them below.

PPP: cheap or free, but it makes you wait

Precise Point Positioning works by modelling. Instead of comparing you against a nearby reference, it corrects the satellites themselves: precise clock and orbit values, plus atmospheric parameters, computed globally and broadcast to everyone. Your receiver applies them to its own raw observations and solves for position alone — hence “point” positioning.

That independence is PPP’s great advantage. No base station, no per-region network, and in the form Northing Global uses, no internet connection at all: the corrections ride down from the navigation satellites alongside the positioning signals. It works in the middle of an ocean.

The cost is convergence. Without a nearby reference to difference against, the receiver has to separate several unknowns that all look alike — its position, the atmospheric delay above it, and the carrier-phase ambiguities — using only the fact that they change at different rates over time. That takes observation time. Minutes to tens of minutes, depending on how much sky the antenna sees and how settled the atmosphere is, before the accuracy arrives.

Worse for moving work, convergence is not permanent. Drive under a dense canopy and lose lock on enough satellites and part of that clock restarts. A guidance pass that ends under trees can begin the next pass part-way back down the curve.

Northing Global: the free mode built into every receiver

Northing Global is our name for the receiver’s built-in, no-subscription mode. It is PPP physics with PPP’s trade-offs, using free high-accuracy correction broadcasts transmitted by the navigation satellites themselves — which is why it works anywhere on Earth with no account, no data plan, and no coverage zones.

It is in every receiver, it costs nothing, and it never expires. It is also not a substitute for RTK, and we would rather say so here than have you find out in a field. What we measured:

• Parked: after convergence, 50% of positions within 7.7 cm and 95% within 22 cm — genuinely useful for marking points, checking boundaries, and finding a stake again.
• Driving: 24–82 cm median depending on the segment, with convergence resets under obstructions.

That second number is the honest one, and it is the reason Northing Global is described on this site as a fallback rather than a tier. Treat any moving-vehicle precision claim for a convergence-based mode with suspicion, including ours.

The one-table version

	RTK (Northing RTK)	PPP (generally)	Northing Global
Accuracy class	±2–3 cm	3–20 cm	~10–20 cm converged
Time to accuracy	seconds	minutes to tens of minutes	minutes to tens of minutes
After signal loss	re-fix in seconds	partial re-convergence	partial re-convergence
Needs internet	yes (~1 kbps)	varies	no
Coverage	zones	varies	everywhere
Cost	subscription (year one included)	varies	free, forever

Which one your work actually needs

The deciding question is not “how accurate?” but “does the antenna keep moving, and does a wrong answer cost me a pass?”

You want RTK if:

• You are driving a repeated line — guidance passes, planting, spraying, coverage mapping. Overlap and skip are decided by repeatability between passes hours or weeks apart, which is exactly what a converging solution cannot promise.
• You are setting out or checking positions against a plan, where being 20 cm out is a re-do.
• You work in and out of tree lines, yards, or buildings, and need the solution back immediately rather than after a re-convergence.

Northing Global is enough if:

• The antenna can sit still for a few minutes before the number matters — marking a point, recording an asset, checking a boundary or a corner.
• You need decimeter-class truth in a place RTK does not reach, or on a job that does not justify a subscription.
• You are doing rough guidance where a 10–20 cm wander is not costing you input.

A useful sanity check: if you would notice a hand-span of error, you want RTK. If you would not, save the money.

What actually breaks them

Neither approach fails politely by default, and this is where the engineering below the marketing shows up.

Obstructions. Both lose satellites under canopy, beside buildings, and in deep cuttings. RTK recovers in seconds because a fix is a discrete event that can simply happen again; a convergence-based solution has to re-earn part of its accuracy. In our drive testing, obstructions cost RTK 1–2 seconds each and were the main driver of the 12% of the drive that was not fixed.

Correction outages. RTK’s dependence on a live stream is real, but the decay is slower than people expect. In our dropout testing the receiver held centimeter accuracy for about a minute unaided, degraded gradually, and was back to RTK-Fixed 24 seconds after the stream returned.

The failure nobody advertises. GNSS engines have an ugly habit: when a correction stream dies silently, they keep reporting centimeter confidence while the real error drifts. We measured exactly that — quality flags, accuracy estimates and correction-age fields all frozen while true error grew from 7 cm toward 40 cm. A receiver that lies to you about its own accuracy is worse than one that is simply less accurate, because you cannot tell which pass to redo.

That is why every Northing receiver runs an independent watchdog: if corrections stop flowing for 30 seconds it raises “Stale Corrections” to every connected app, whatever the engine underneath claims. More on that in how Northing RTK works and inside the receiver.

What it costs

RTK is a subscription because someone has to keep computing and serving the corrections. What varies between vendors is honesty about the total: a receiver bundled to a display you must buy, corrections metered by the hour, or tiers you outgrow.

Northing One is $999 with the first year of Northing RTK included, and renews at a flat $299 for 12 months — same price every year, a reminder email before every renewal, cancel anytime. If you cancel, the receiver does not brick: it falls back to Northing Global and keeps working everywhere on Earth, for free, forever.

The practical summary: moving work wants RTK; static work tolerates convergence; Northing Global is the remarkable free fallback. That is the logic behind shipping the receiver with a year of RTK included while keeping Global built in underneath — the physics picks the mode, not the marketing.

Next: where Northing RTK coverage reaches and what the hardware does, or see the two services side by side on the home page.