How to choose a patch antenna comes down to four decisions you can settle before you compare product photos: which bands the receiver actually uses, how much ground plane the board can give the antenna, how the gain and polarization behave away from zenith, and how much loss sits between the antenna and the receiver input.
Short answer: Lock the band set first (GPS L1 only, multi-constellation L1, or L1 plus L5). Then take the largest ceramic patch the enclosure allows and keep the ground plane close to the one used for the datasheet measurement. Go active when cable loss plus receiver noise figure costs more than roughly 2 dB. Ask for axial ratio and gain away from boresight, not only at zenith. Finally, retune and retest the antenna inside the real enclosure.
Those four decisions matter because the signal is extremely weak before your antenna touches it. The GPS interface specification IS-GPS-200L guarantees a minimum received power of only −158.5 dBW for the L1 C/A signal from a GPS III satellite, measured with a 0 dBi right-hand circularly polarized antenna near the horizon, and −161.5 dBW for the encrypted P(Y) signal (IS-GPS-200L). The civil Standard Positioning Service performance standard describes the same order of magnitude, −160 dBW referenced to a 3 dBi linearly polarized antenna with a 5° elevation mask (GPS performance standards). Antenna selection is simply how you spend or save a few dB against that number, which is why a good patch on a bad ground plane fails while a smaller patch, mounted properly, works.
If your team is searching for how to choose patch antenna options because a tracker, timing receiver, or GNSS module is losing margin, start with the band and the installation rather than the product photo. For a product-first path, review GPS / RHCP patch antenna options in parallel with this checklist.
Quick Selection Checklist
- Confirm the frequency bands: GPS L1 only, multi-constellation L1, or L1/L2/L5.
- Write down the ground plane you can actually offer, in millimetres, before shortlisting parts.
- Add up the loss between antenna and receiver, then decide active or passive.
- Match antenna size to the ground plane and to the bandwidth the band set needs.
- Require RHCP with axial ratio data, not just the word “RHCP”.
- Review the gain pattern at low elevation, not only peak gain at zenith.
- Verify connector, cable, mounting, and operating temperature.
- Retune and retest in the final enclosure before production.
If the shortlist is still unclear after this first pass, separate the two most common RF questions. Use the patch antenna gain guide when the issue is signal margin, pattern, ground plane, or installed orientation. Use the patch antenna bandwidth guide when the issue is GPS L1-only coverage, wider GNSS bands, VSWR window, or enclosure detuning.
Start With the Application
| Application | Typical antenna priority | Watch point |
|---|---|---|
| Asset tracker | Compact size, low power, low cost | Small ground plane and enclosure detuning |
| Telematics device | Stable GNSS with cellular nearby | LTE/4G/5G isolation and filtering |
| Wearable or handheld tracker | Light weight, changing orientation, vibration | Pattern coverage and mechanical fixing |
| RTK rover | RHCP quality, multiband, phase center stability | Axial ratio and ground plane consistency |
| Timing receiver | Stable signal and fixed mounting | Cable loss, active antenna power, outdoor placement |
| Precision agriculture | Multipath resistance and field reliability | Enclosure, radome, and mounting position |
Step 1: Lock the Frequency Bands
If the antenna does not cover the required band, everything else is cosmetic. Start from the receiver’s signal list, not from the marketing phrase “multi-GNSS”. These are the frequencies the receiver may ask the antenna to deliver:
| Signal | Center frequency | Note |
|---|---|---|
| GPS L1 C/A and L1C, Galileo E1, BeiDou B1C, QZSS L1, SBAS L1 | 1575.42 MHz | The interoperable L1 frequency, and the only band many trackers need |
| BeiDou B1I | 1561.098 MHz | 14.3 MHz below L1, outside a narrow L1-only tuning window |
| GLONASS L1 (FDMA) | about 1598.1 to 1605.4 MHz | Channels spaced 0.5625 MHz around 1602 MHz, roughly 27 MHz above L1 |
| GPS L2C | 1227.60 MHz | Lower band, needs a separate resonance |
| GPS L5, Galileo E5a, BeiDou B2a, NavIC L5 | 1176.45 MHz | The second interoperable frequency, used for high accuracy |
| Galileo E5b, BeiDou B2b | 1207.14 MHz | Often covered together with L5 by a wideband or stacked patch |
Frequencies from the ESA Navipedia signal plans for GPS, GLONASS and BeiDou.

Here is the part most selection guides skip. The upper L-band radionavigation-satellite allocation runs from 1559 MHz to 1610 MHz, which is 51 MHz wide, and real multi-constellation L1 reception spans from BeiDou B1I at 1561.098 MHz up to the top GLONASS channel near 1605 MHz, about 45 MHz. A ceramic patch is a resonant structure, so its usable bandwidth is tied to its physical volume. Typical published values for 4 mm thick GPS L1 ceramic patches, measured on the manufacturer’s reference ground plane, look like this:
| Patch size | Typical bandwidth | What it can realistically cover |
|---|---|---|
| 36 x 36 mm | about 40 MHz | Wide L1 coverage including GLONASS and BeiDou B1I |
| 25 x 25 mm | about 20 MHz | GPS L1 plus Galileo E1 comfortably; GLONASS needs a shifted or dual-fed design |
| 18 x 18 mm | about 10 MHz | GPS L1 and Galileo E1 at 1575.42 MHz |
| 15 x 15 mm | about 8 MHz | GPS L1 only, with careful tuning |
| 12 x 12 mm | about 7 MHz | GPS L1 only, little margin for detuning |
| 10 x 10 mm | about 5 MHz | GPS L1 only, tuned for one specific ground plane and enclosure |
Treat those numbers as orders of magnitude rather than specifications, but the conclusion is robust: a 15 mm patch with an 8 MHz window physically cannot serve BeiDou B1I and GLONASS L1 at the same time as GPS L1. When a small single-feed patch is advertised as covering all four constellations, the honest reading is that it resonates around 1575.42 MHz and the other constellations sit on the skirts of the response, where gain and axial ratio degrade. If the receiver really needs all L1 constellations, plan for a larger patch, a dual-fed or stacked design, or accept that GLONASS and BeiDou will contribute fewer usable satellites.
Step 2: Active or Passive, Decided by a Loss Budget

| Choice | Better when | Not recommended when |
|---|---|---|
| Passive GPS patch antenna | Antenna is close to receiver, cable is short, power must be low | Cable loss is high or receiver noise margin is tight |
| Active GPS patch antenna | Cable run is longer, signal loss must be compensated, LNA can sit near the antenna | The receiver cannot provide bias power or nearby transmitters may overload the LNA |
An active antenna puts an LNA, and usually a filter, right behind the radiator, so the noise figure of the chain is set before the cable loss is applied. That is the whole point: loss after the LNA costs you almost nothing, loss before it costs you dB for dB.
You can size the decision arithmetically instead of guessing. Starting from the guaranteed −158.5 dBW at a 0 dBi antenna and thermal noise density of about −204 dBW/Hz at 290 K, the best case carrier-to-noise density before any receiver noise figure or system loss is roughly 45 dB-Hz. That is an engineering estimate derived from published minimum power, not a measured figure, but it frames the problem: you are working with a budget of a few tens of dB in which every 1 dB of connector, cable, mismatch, and polarization loss is directly subtracted. Take the insertion loss per metre at 1.6 GHz from the cable datasheet, add connector and mismatch loss, add the receiver noise figure, and if the total before the receiver exceeds roughly 2 dB, an active antenna is usually the cheaper fix than fighting for antenna gain.
Step 3: Pick the Size, Then Protect the Tuning

Size buys three things at once: gain, bandwidth, and tolerance to your own product. A 25 x 25 mm ceramic patch has noticeably more margin than an 18 x 18 mm or 15 x 15 mm patch, and the extra margin is mostly consumed by the enclosure rather than by the specification sheet.
That last point is where projects fail. A ceramic patch is tuned against a specific ground plane and a specific dielectric environment. Add a plastic housing, adhesive, a battery, or a display and the resonance shifts by megahertz. On a 36 mm patch with 40 MHz of bandwidth a few megahertz is a nuisance; on a 12 mm patch with 7 MHz of bandwidth it can move the passband off the signal, and the GPS L1 C/A main lobe is only about 2 MHz wide null to null. So the smaller the patch, the more the tuning must be done with the real mechanics, and the more you should insist on a supplier who will retune the part for your build rather than ship a catalogue item.
The best patch antenna is therefore not the smallest and not the one with the highest headline gain. It is the one that still meets performance after the enclosure, ground plane, and cable are included.
Step 4: Ground Plane and Placement Decide Installed Performance

A patch antenna is only half an antenna. The ground plane under it acts as the counterpoise and the reflector, so it sets the pattern, the gain, the back lobe, and part of the tuning. At 1575.42 MHz a free-space wavelength is about 19.0 cm, so a quarter wavelength is about 4.8 cm; that is the scale at which a ground plane starts behaving like a ground plane rather than like a small piece of metal near a radiator.
The effect is much stronger than most datasheets suggest. In a peer-reviewed measurement study published in Advances in Radio Science, a multi-band GNSS patch antenna produced no usable satellite reception at all on circular ground planes smaller than about 12 cm in diameter, while larger planes restored normal performance (Punzet and Eibert, 2023). Small ground planes also push energy into a rear lobe, which is exactly the direction ground-reflected multipath arrives from, so a compact tracker can lose accuracy even where it still gets a fix.
Practical placement rules:
- Give the antenna the largest continuous copper area the product allows, and keep it symmetric under the patch where possible.
- Use the supplier’s reference ground plane size as the starting point, then measure on your own PCB.
- Keep metal parts, batteries, and displays out of the top hemisphere and out of the near field.
- Place the antenna where it has the best sky view, not where it is mechanically convenient.
- Avoid routing noisy digital lines near the antenna feed.
- Keep cellular, LTE, 4G, and 5G antennas separated from the GNSS antenna.
- Test with the real enclosure, adhesive, cable, and mounting hardware.
For the deeper design-level treatment of feed position, substrate, and tuning, use the patch antenna design guide.
Step 5: Read the Pattern, Not the Peak Gain

Peak gain is measured at zenith, and zenith is where you need help least. ESA Navipedia notes that a patch typically radiates over roughly 160° of coverage, that gain is maximum at zenith and falls to near unity around 10° to 15° elevation, and that real antennas roll off by 10 to 20 dB between boresight and the horizon (ESA Navipedia: Antennas). Those low-elevation satellites are the ones that fix your position geometry, so a 5 dBi headline number tells you very little on its own.
Ask for two curves instead:
- Gain versus elevation angle, so you can see what is left at 10° to 20° where new satellites appear.
- Axial ratio versus elevation angle, because GNSS signals are right-hand circularly polarized and polarization mismatch is a real loss. A well-behaved GNSS antenna is usually specified within about 3 dB axial ratio over the elevation range of interest; a figure quoted only at boresight hides the degradation off-axis.
If a datasheet does not show axial ratio for a GNSS patch antenna, ask for it before using the antenna in a precision product. For the full RHCP, axial ratio, LHCP, and multipath explanation, read the RHCP vs LHCP patch antenna guide.
Step 6: Cable, Connector, and Mounting
- Choose the connector for the assembly process, not only the footprint: u.FL and MMCX save space but have limited mating cycles, while SMA suits serviceable external antennas.
- Take cable loss per metre at 1.6 GHz from the datasheet; thin flexible coax inside a small device loses far more per metre than the low-loss cable used on a test bench.
- Keep the cable away from moving parts and sharp bends, and fix it mechanically so the connector is not carrying strain.
- Confirm operating and storage temperature, adhesive type, and vibration rating against the real environment.
- For external mounts, check the radome material, IP rating, and UV resistance.
Step 7: Validate in the Final Enclosure
A selection is not finished until it is measured in the product. A workable minimum:
- Measure S11 or VSWR with the antenna in the assembled enclosure, not on a bare board.
- Compare carrier-to-noise ratio per satellite against a known-good reference antenna in the same location.
- Check time to first fix from cold start, outdoors, in the final orientation.
- Repeat with the device in its worst realistic orientation and with a hand or mounting bracket present.
- Confirm behaviour with the cellular modem transmitting at full power.
Supplier Qualification Questions

Ask these before final approval:
- What ground plane size and shape were used for the datasheet measurements?
- Can you provide S11, VSWR, gain pattern, efficiency, and axial ratio, each as a function of elevation angle?
- Is the antenna tuned for free space, a reference board, or a specific ground plane?
- Will you retune the part for our enclosure, and what samples and lead time does that take?
- Which frequencies are actually inside the specified band, and which sit on the skirts?
- What connector and cable options are available, and at what measured loss?
- For active parts: what are the LNA gain, noise figure, current draw, and out-of-band rejection?
- Can you provide samples and test support for our enclosure?
- Are RoHS, REACH, and other required compliance documents available?
For the full topic hub, product paths, and application map, see the Patch Antennas guide.
FAQ
How do I choose the best patch antenna?
Choose the patch antenna that covers your frequency band, fits the ground plane you can actually provide, keeps enough gain and polarization quality at low elevation, and still meets performance after installation in the final device.
Is an active GPS patch antenna better than a passive one?
Active is better when cable and connector loss or receiver noise figure would otherwise cost more than about 2 dB, because the LNA sets the noise figure before that loss. Passive is better when the antenna is close to the receiver and low power matters.
What size GPS patch antenna should I use?
Use the largest size the mechanical design allows. Size buys bandwidth and tolerance to enclosure detuning: roughly 20 MHz for a 25 mm patch versus roughly 8 MHz for a 15 mm patch, against a GPS L1 signal main lobe only about 2 MHz wide.
What ground plane does a GPS patch antenna need?
Start from the supplier’s reference ground plane and verify on your own PCB. As a scale reference, a quarter wavelength at L1 is about 4.8 cm, and published measurements show a multi-band GNSS patch failing to receive at all on ground planes below roughly 12 cm in diameter.
Can one patch antenna work for GPS, GLONASS, Galileo, and BeiDou?
Only if its bandwidth genuinely spans them. GPS L1, Galileo E1, and BeiDou B1C share 1575.42 MHz, but BeiDou B1I sits at 1561.098 MHz and GLONASS L1 runs near 1598 to 1605 MHz, so full L1 coverage needs about 45 MHz of bandwidth, which a small single-feed patch does not have.
Is a bigger ground plane always better?
Bigger helps up to a point: it raises gain toward zenith and suppresses the rear lobe that picks up ground-reflected multipath. Very large planes bring diminishing returns and can change the tuning and axial ratio, so the practical answer is to measure the sizes you can actually build.
Why does my antenna work on the bench but not in the product?
Almost always detuning and placement. The enclosure, adhesive, battery, and cable shift the resonance by megahertz, and the smaller the patch, the less bandwidth there is to absorb that shift. Retune with the real mechanics in place.
Do I need L5 or multiband?
Only if the receiver uses it. L5 and E5a at 1176.45 MHz help with multipath and accuracy, but they require a wider or stacked antenna design and more space, so specify them from the receiver’s signal list rather than as a default upgrade.
Conclusion
To choose a patch antenna, lock the band set, write down the ground plane you can give the antenna, budget the loss from antenna to receiver, then compare size, pattern, and axial ratio at the elevation angles that matter. Every step after that is verification in the final enclosure, which is where installed performance is actually decided.
View GPS/GNSS patch antenna options or send your device layout to our engineering team for selection support.
References
- IS-GPS-200L, GPS Interface Specification (US Coast Guard Navigation Center)
- GPS Performance Standards and Specifications (GPS.gov)
- GPS Space Segment (GPS.gov)
- ESA Navipedia: Antennas
- ESA Navipedia: GPS Signal Plan
- ESA Navipedia: GLONASS Signal Plan
- ESA Navipedia: BeiDou Signal Plan
- Punzet and Eibert (2023), influence of ground plane size on GNSS antenna performance, Advances in Radio Science
Ask our AI antenna assistant. Enter a keyword or question for a free answer based on this article, Global RF Tech content, and general antenna/RF knowledge.
Do not submit confidential or personal information. Privacy Policy


