RTK Without CORS: How Self-Converging PPP Base Stations Work

How a PPP base station delivers centimetre RTK anywhere on Earth with no CORS subscription, no internet and no known point.

What a CORS network actually is — and what it costs

Conventional RTK needs a correction source. For most crews that means one of two things: a subscription to a CORS network (Continuously Operating Reference Stations) delivered over NTRIP, or their own base station set up on a point with known coordinates. Commercial network RTK services such as Trimble VRS Now, Leica SmartNet, and regional state or national CORS networks charge per rover, per year — typically several hundred to a few thousand dollars annually, forever. Every tractor, every survey rover, every RTK drone that consumes corrections is another line on the invoice.

The money is only half of the problem. Network RTK accuracy depends on your distance to the nearest reference station. Beyond roughly 30–40 km baselines, initialization slows down and residual atmospheric errors grow; in many rural, mining, forestry, and developing regions there simply is no station within range at all. Cross-border projects add another layer of friction, because subscriptions rarely roam. And when the network has an outage — or the cellular link that carries NTRIP drops — every rover on the job stops with it.

How crews work around it today

The traditional workaround is running your own base. That solves the coverage question but reopens an older one: where exactly is the base? If you set a receiver such as an Emlid Reach RS3 or RS4 over a known point on a tripod and tribrach, you inherit that point's accuracy. If no known point exists, most receivers offer an averaged single-point position — Emlid Flow, for example, averages for a configurable period — which is fine for relative accuracy within one survey but can sit a metre or more away from truth in a global frame.

Drone-native bases show the same pattern. DJI's own specifications state that a D-RTK 3 Multifunctional Station operating on an uncalibrated single point is accurate to about 1.5 m horizontal and 3.0 m vertical; to get centimetre absolute coordinates you must calibrate it against a network RTK service or a known point — which quietly reintroduces the CORS dependency you were trying to escape. Some crews log static data and submit it to free post-processing services like OPUS (US), AUSPOS (Australia), or NRCan CSRS-PPP (Canada); accurate, but you wait hours for results and need internet to submit.

None of these paths is wrong. They are all workarounds for the same missing capability: a base station that can determine its own absolute position, to centimetres, on its own.

How a self-converging PPP base station works

PPP — Precise Point Positioning — attacks the problem from the satellite side. Instead of differencing observations against a nearby reference station, a PPP receiver applies precise satellite orbit and clock corrections generated by a global tracking network. Those corrections are broadcast from geostationary satellites on the L-Band, received by the same GNSS antenna that tracks GPS, GLONASS, Galileo, and BeiDou. No SIM card, no NTRIP account, no local infrastructure.

A modern multi-frequency PPP engine resolves carrier-phase ambiguities against those corrections and converges to an absolute position of about 1.5 cm horizontal and 3 cm vertical in roughly 3 minutes under open sky. The coordinates come out in ITRF2020 at the current epoch — a globally consistent frame that transforms cleanly to any national datum (read more in our ITRF2020 guide). The receiver then locks that position as its base reference. From this point on, it behaves exactly like a base that was set on a surveyed monument — except nobody surveyed anything.

From PPP to RTK: serving corrections to any rover

Once converged, the base broadcasts standard RTCM 3.x corrections — the same open format used by every RTK rover built in the last decade. Delivery is whatever the site allows: the built-in UHF radio covers a working radius of about 5 km with zero infrastructure; 4G NTRIP serves rovers at any distance where cellular exists; and for drone work, corrections can ride the aircraft's own RC link. A DJI Matrice 350 RTK or Matrice 4E consumes them through Custom Network RTK in DJI Pilot 2 exactly as it would consume a CORS stream (step-by-step in our M350 setup guide).

Because RTCM 3.x is vendor-neutral, one PPP base can simultaneously feed a mapping drone, agricultural auto-steer tractors, machine control excavators, and survey rovers from any manufacturer — Emlid, CHCNAV, South, Trimble, or Leica hardware included. There is no pairing and no per-rover fee.

What actually changes in the field

The practical differences compound. Setup drops to placing the receiver on open ground and pressing power — no tripod levelling over a monument, no averaging session, no NTRIP credentials to type at 6 a.m. Repeatability improves, because the base position is absolute: return to the same site next month and your data lands on top of last month's without re-registration. Cost becomes a one-time hardware purchase plus a single correction-service subscription that feeds unlimited rovers, instead of per-rover network fees.

And coverage anxiety disappears. L-Band corrections rain down identically on a Kansas cornfield, a Chilean mine, and an Indonesian palm plantation. If the sky is open, the base converges — see drone RTK in remote areas for field patterns.

Where conventional workflows still make sense

Honesty matters: a CORS tie is still the right call in a few situations. If your deliverable must be expressed on legacy local control — a boundary survey referenced to existing monuments — you will localize to those monuments regardless of how the base positioned itself. If you already operate inside dense, reliable network coverage with paid-up subscriptions and short baselines, the marginal gain is smaller. And under heavy canopy or in urban canyons, L-Band reception suffers just as GNSS does; plan the base placement in the open and push corrections into the obstructed area over the radio instead.

For everyone working beyond the edge of a network — which is most of the drone mapping, agriculture, and construction world — a self-converging PPP base removes the single most fragile dependency in the RTK chain.

A worked example: one season, two cost models

Put numbers on it. A three-rover operation — one RTK drone, one survey rover, one auto-steer tractor — on a typical network RTK plan pays three subscriptions of roughly $800–1,500 each, every year, and still cannot work outside network coverage. The same operation with a self-converging PPP base pays for the hardware once plus one annual PPP correction service, and feeds all three rovers (and any it adds later) from a single RTCM 3.x broadcast. By year two the network model has usually cost more, and the gap widens with every additional rover and every remote job the network model simply could not do.

There is also a resilience dividend that never shows up on invoices: when a regional network went down for maintenance, or when the crew crossed a border where the subscription did not roam, the self-contained base kept working identically. Owning the correction source turns a monthly operating risk into a piece of equipment in the truck.

Setting one up: what the first ten minutes look like

The workflow is deliberately anticlimactic. Screw on the whip antenna, set the receiver on open ground — no tripod required at 827 g with an IP68 shell — and press power. Join its Wi-Fi hotspot, open the device console at 192.168.2.1:8080, and watch the PPP engine converge: satellites acquired, corrections locked, uncertainty shrinking from metres to centimetres in about three minutes. When the status shows converged, start base mode; the console displays the live RTCM 3.x output, the NTRIP credentials for cellular delivery, and a RINEX logging toggle for PPK backup. From cold start to serving corrections is typically under five minutes — less time than levelling a tribrach over a monument, and with nothing to peg, measure, or type.

Further reading

The convergence mechanics behind the three-minute figure are unpacked in convergence time explained.

Frequently asked questions

Do I still need a CORS or NTRIP subscription with a PPP base station?

No. The base receives its corrections from L-Band satellites and computes its own centimetre position, then broadcasts RTCM 3.x locally. Rovers connect to the base, not to a network, so no per-rover network subscription is required.

How accurate is a self-converging PPP base compared with a base on a known point?

Typical PPP convergence reaches about 1.5 cm horizontal / 3 cm vertical in roughly 3 minutes under open sky — comparable to occupying a surveyed point, and in a globally consistent ITRF2020 frame rather than a possibly outdated local monument.

Will it work with my existing rovers and drones?

Yes, if they accept RTCM 3.x corrections — which covers DJI RTK drones (Matrice 350 RTK, Matrice 4, Mavic 3 Enterprise), Emlid, CHCNAV, South, Trimble, and Leica rovers, plus agricultural auto-steer and machine control systems.

What happens if there is no internet at the site?

Nothing changes. L-Band corrections arrive by satellite and the UHF radio or drone RC link carries RTCM to the rovers, so the entire workflow runs with zero connectivity.