How to Choose a GNSS Base Station: Receivers Compared

Most people start this decision at the wrong end. They open a few product pages, compare the prices, and try to work out what the extra money buys. The more useful question comes first: are you buying a rover or a base station?

The distinction matters because the same receiver modules appear in both, and the ranking changes completely depending on which job you are buying for.

A rover moves (usually). It is a GNSS receiver that  consumes correction data, sits on a drone or a tractor cab, and lives under real constraints on size, weight and power. What matters most is how quickly it recovers a fixed solution after passing under a tree line or beside a building. If you are working somewhere with unreliable connectivity, it is worth knowing that post-processing the data afterwards with PPK is a legitimate alternative workflow that removes the need for a live correction link entirely.

A base station does the opposite. It is a high precision GNSS or RTK receiver that is firmly installed at a position it already knows, and sits on a roof for years. Because it knows exactly where it is, it can compare what the satellites tell it against what they should be telling it, and the difference between the two is the correction. That is the whole idea: the base station measures the error so the rover does not have to. Size and power draw are irrelevant. What matters is how many signals it can see, how clean those observations are, whether it stays online without attention and how stable the installation is. That's why professionals call it CORS: continuously operating reference station. 

In this article, we’d like to discuss exactly this type of GNSS reference station receiver and how to choose the right one for you.

Four trade-offs decide the outcome, and they pull against each other:

• Accuracy and convergence. Manufacturer specs cluster within a centimetre of each other, so the real difference shows up in how fast a fix is resolved and how well it holds under a partly obstructed sky.

• Bands and constellations. Every receiver here tracks GPS, GLONASS, Galileo and BeiDou. The separation is in frequency bands. Adding L5 and E6 improves ionospheric modelling, shortens initialisation, and gives redundancy when part of the sky is blocked.

• Availability and connectivity. This one is on no datasheet, and for a base station it decides more than most of what is. A reference station that is offline produces nothing, and frequent short outages; a power cut, a router reboot, an ISP fault, erode its value faster than a slightly weaker RF front-end ever will. It is also the only item on this list you can still do something about after you have bought the receiver. Ask not just what a station tracks, but how it stays up: can it run from backup power, and does it have a second route to the internet when the first one fails? 

• Cost. The spread from entry-level to professional-grade is roughly four times. Whether that gap is worth paying depends far more on your environment than on your ambition.

• Ease of setup. Some products ship with a web interface, an Ethernet port and a working NTRIP server. Others need a Raspberry Pi, a serial bridge and an afternoon. This is a real cost, just not a monetary one.

Two caveats before the comparison. The antenna and where you mount it will cap whatever you buy: a mid-range receiver with a good multi-band antenna on a clear rooftop beats a professional-grade module on a compromised site, every time. And plan the power and the connection before you plan the receiver. A UPS and a fallback link cost a fraction of the station and protect everything else you spent. 

The receivers worth comparing

There is a useful distinction to make here. Dozens of base station products are on the market, but they resolve to a much smaller number of receiver modules. Two products from different vendors at different prices may contain exactly the same silicon, differing only in enclosure, connectivity and setup software. Understanding the modules first makes the products much easier to judge.

Five modules cover almost the entire realistic range, listed here from entry-level upward.

Quectel LG69T and LC29H

The LG69T runs an 80-channel STMicroelectronics platform in dual-band configurations, either L1+L2 or L1+L5. Constellation support is variant-dependent and worth checking carefully: Quectel's own specification for several variants lists four concurrent constellations plus optional QZSS, while distributor documentation for others lists up to six including NavIC. Its distinguishing feature is an integrated IMU and dead-reckoning capability that fuses inertial data, wheel ticks and vehicle dynamics with the GNSS solution.

Built for:  automotive, ADAS, robotics and agricultural rovers. Designed to consume RTCM corrections from third-party base stations, not to produce them.

Accuracy:  centimetre-level with RTK, though RTK support is optional and depends on the specific variant.

Price value: $$

The dead-reckoning capability is genuinely valuable on a moving vehicle in an urban canyon and completely irrelevant on a rooftop. With 80 channels and two frequency bands, this is the weakest option for a permanent reference station. Base-station and NTRIP server support also varies considerably between integrations, so verify before buying.

u-blox ZED-F9P 

It is a dual-band receiver running a 184-channel engine, tracking GPS L1C/A and L2C, GLONASS L1OF and L2OF, Galileo E1 and E5b, BeiDou B1I and B2I, plus QZSS. No L5, no E6.

Built for:  hobbyists, DIY builders, drone and robotics integrators, and anyone whose existing rover already speaks F9P.

RTK accuracy:  0.02 m + 1 ppm CEP horizontal (u-blox datasheet). Convergence typically lasts under 10 seconds in open sky.

Price Value:  $$

The honest limitation is signal diversity. Two frequency bands is enough for solid RTK on a short baseline with a clear sky, but performance degrades faster than the alternatives under canopy, in urban multipath, or at longer baselines. As a base station it will serve local rovers well, but it cannot serve the L5 and E6 signals that newer rovers can use.

Bynav M20 

It is built on Bynav's Alice 22nm automotive-grade system-on-chip, certified to ISO 26262 ASIL B functional safety, and runs 1507 channels.

Signal coverage: GPS L1C/A, L1C, L2 and L5; BeiDou B1I, B2I, B3I, B1C, B2a and B2b; GLONASS G1 and G2; Galileo E1, E5a, E5b and E6; QZSS L1, L2, L5 and L6 including CLAS; NavIC L5; SBAS; and L-band. It also includes SAIF interference defence with 60 dB narrowband anti-jamming.

Built for:  autonomous driving, drones and robotics originally, but it performs well as a permanent base and is used as one across the network.

RTK accuracy:  approximately 1 cm according to vendor listings. Bynav does not publish an RMS specification as openly as Unicore or Septentrio, so treat this as less firm than the other figures here.

Price value:  $$

Unicore UM980 and UM982 

Built on Unicore's NebulasIV system-on-chip with 1408 channels, it tracks GPS L1/L2/L5, GLONASS G1/G2/G3, Galileo E1/E5a/E5b/E6, BeiDou B1I/B2I/B3I/B1C/B2a, QZSS L1/L2/L5 and NavIC L5.

Built for:  serious hobbyists through to professionals. Widely used in precision agriculture and autosteer, UAV work, and permanent reference stations.

RTK accuracy:  0.8 cm + 1 ppm horizontal, 1.5 cm + 1 ppm vertical (Unicore datasheet). Initialisation typically under 5 seconds, the fastest published figure in this comparison.

Price Value:  $$$

The UM982 is the same platform with dual-antenna heading. Useful on a vehicle or a rover. Pointless on a fixed base, so do not pay for it unless you need the heading. More station products are built on the UM980 than on any other module, which means setup guides are easy to find.

Septentrio mosaic-X5 

The mosaic-X5 is a professional module in a small package: 448 hardware channels tracking all constellations across all available frequencies. Its distinguishing feature is not the raw specification but AIM+, Septentrio's interference mitigation, which works in the RF front-end before signals reach the digital processing chain rather than filtering afterwards in software.

Built for:  professional surveying, machine control, UAV integration, and any site with a difficult RF environment: near cell towers, airports, radar, or dense urban infrastructure.

RTK accuracy:  0.6 cm + 0.5 ppm horizontal, 1 cm + 1 ppm vertical (Septentrio specification). Carrier phase ambiguities typically resolved in under 10 seconds on triple-frequency signals.

Price tier:  $$$$

. The mosaic-X5 earns its price on observation quality and stability over long periods: it holds a fix where other receivers drop it, and produces cleaner carrier-phase data year after year. If your station will spend years in a noisy RF environment, or if clients depend on your output, that reliability is what you are paying for. Most boards also ship with a built-in web interface, which meaningfully reduces setup effort.

One thing to check before ordering: manufacturers enable feature sets at the point of sale. Septentrio calls them permissions, u-blox calls them fuses. A low-cost board built on a premium chipset may not have base-station mode unlocked. Confirm what is enabled on the specific product, not what the chipset can theoretically do.

Above this range: geodetic equipment

Geodetic receivers such as the Septentrio PolaRx5, Trimble Alloy and Leica GR series cost about ten times more. The extra cost is often not for better GNSS performance. A Septentrio geodetic receiver and a mosaic-X5 use very similar technology. You pay for everything around the receiver: a strong, weatherproof housing, a built-in battery that keeps the station running during power cuts, several internet options including cellular, and a wider operating temperature range. These are the same availability points covered earlier, built into one device. So the right question is not how much accuracy you need, but how difficult your site is. If your power and internet are stable, you do not need this extra protection. If your power or internet is unreliable, a geodetic receiver may be the only way to keep your station online. 

Side by side

Manufacturer specifications where published. Real-world results depend on baseline length, antenna quality, sky visibility and multipath, and prices vary by vendor and region.

Which station products use which module

Once you know the modules, the product landscape resolves quickly. This is roughly how the commonly available base station products map:

Where a vendor does not publish the module inside, judge the product on its published band and constellation coverage instead. A four-band, six-constellation product is in the same performance class regardless of whose silicon is inside. Full specifications for most of these are available from the hardware partners directly.

Which one is right for you

Translating the above into an actual decision comes down to what the station is for and where it will live.

1. Budget-conscious or DIY. The ZED-F9P is a sensible choice for a first station on a budget in a rural area with a clear sky. It is the cheapest credible route to centimetre accuracy, and the community support means you will not get stuck. Spend what you save on a better antenna. However this might not be the best choice for integrating onocoy, as onocoy rewards based on the quality signal.

2. Precision agriculture, UAV, prosumer. The UM980 offers good value for money.  Four frequency bands and six constellations give you fast initialisation and resilience under partial obstruction, at a price that does not require a business case. Precision Agriculture and UAV operators in particular benefit from the sub-5-second initialisation, and the wide product ecosystem means setup is well documented.

3. Professional surveying and machine control. The mosaic-X5 earns its price when your output has consequences: client deliverables, machine control, or a site with real RF interference. If you are working in an urban canyon or near a source of jamming, this is the receiver that keeps working when others do not and earns the highest rewards with onocoy. 

4. Running it as shared infrastructure. Specify at least a UM980 or an M20. Band and constellation coverage is what determines how useful your stream is to anyone else, and dual-band data is markedly less valuable. In an area with existing coverage nearby, go to a mosaic-X5 or reconsider the site entirely.

5. You already own a rover. Match the base to what the rover can actually consume. A four-band base serving a dual-band rover in a rover-base setup is money spent on capability you cannot access, unless you intend to share the data or upgrade the rover later. But if you feed the data to onocoy, the rewards of a four-band station help amortise it much faster. 

What else you need to budget for

The receiver is not the purchase. A working permanent station also needs a multi-band antenna matched to the receiver's frequency bands, which is the component most people underspend on and the one that will cap everything else. Then a rigid outdoor mount that will not shift over seasons of wind and frost, quality coaxial cable kept as short as the installation allows, and a power and network connection that will still be there in many years. Ethernet with PoE is the most reliable option, Wi-Fi is fine for most homes, and an LTE modem is worth it for remote sites.

One detail that sounds obvious and is still worth stating: the antenna must point at the sky. Mounted upside down it will still produce a stream, and a network that does not check observation quality will still reward it, but the data is close to useless. Rigid and correctly oriented, both.

Backup connectivity. A wired connection makes the best primary link and the worst single point of failure. A 4G modem as fallback keeps the stream alive through an ISP outage; some stations have one built in. On remote sites cellular is usually the primary link rather than the backup, in which case the wired connection is the one you do without. 

Budget realistically for all of it. A €300 receiver on a €40 antenna and a drainpipe bracket will underperform a €200 receiver installed properly, and the difference shows up permanently in your observation quality rather than as a one-off inconvenience.

Connecting your station to a wider network

Once a station is installed, it will spend most of its life idle. It tracks satellites continuously, but you only use its corrections during the hours you are actually surveying, flying or driving. The rest of the time it produces data nobody is reading.

You have options for what to do about that, and they are all reasonable. You can run the station in a rover-base setup entirely privately, which is the simplest arrangement and the right one if your data is sensitive or your uptime is unpredictable. You can contribute to a public CORS network or a community caster, which costs nothing and helps the people around you. Or you can connect it to an open marketplace  thereby helping both the GNSS industry and end users, as well as your own bank account.

onocoy operates an open marketplace for GNSS reference data, connecting existing stations so that data users can access corrections where they need them and station owners get more out of hardware they have already paid for. Connecting a station takes minutes.

Every module above is compatible. onocoy is hardware-agnostic and does not manufacture or sell receivers, so any station that outputs RTCM 3.x over NTRIP can connect, whether it is a self-built ZED-F9P setup or a professional mosaic-X5 installation. Most of the products listed in the table above are available from hardware partners as ready-to-run bundles.

What connecting actually gives you:

• You keep using your station. Loop-back access means you keep streaming your own station's corrections for your own work exactly as before. Connecting is not exclusive and does not change your workflow.

• Continuous quality monitoring. Every stream is checked continuously, and owners see how their station is performing. This is useful diagnostic information in its own right: it will tell you if your antenna siting is costing you signal quality, or if a firmware change has degraded your observations.

• Rewards for contributing data. Station owners are rewarded for the data they contribute, with the amount scaling to the constellations and frequency bands delivered, the measured quality of the observations, and uptime. A location factor reduces this in areas already well covered by equivalent stations. Uptime matters more than it first appears. Clients choose stations they can rely on, so a station with frequent outages, even short ones, is much less likely to be picked up by clients, and will therefore gain little from local usage rewards in the future. The direct effect of downtime on base rewards is secondary; lost client usage is the real cost. The same things that make a station good for your own work, clean signals and reliable uptime, are what make it valuable to everyone else. 

• Access for data users in your area. Data users stream corrections on demand and pay per hour of use, which means a station in an area with thin coverage has real value to the people working there.

How much this offsets the cost of your hardware depends on your equipment, your location and your uptime, so there is no single figure worth quoting. What is consistent is the direction. A well-specified station in an under-covered area, kept online, does useful work for other people while doing exactly the job you bought it for. And it helps you to amortize your station much faster.

Worth doing before you order anything: check the opportunity map for station density near your intended site. It will tell you whether you are filling a gap or entering a crowded area, and that single piece of information changes the hardware answer more than any specification comparison.

See where a station would be most useful on the opportunity map, check existing service on the coverage map, or browse ready-to-run bundles from our hardware partners.