changed a year ago  
Linked with GitHub

How to (not) ruin your PCB antenna

PCB antenna structures sense their immediate surroundings. Electric permittivity, conductivity and magnetic susceptibility of nearby elements directly impact the resonant frequency. In high speed PCB design, the 20H and 3W rules give an idea how to minimize coupling between planes and traces, respectively,

https://www.autodesk.com/products/eagle/blog/top-10-tips-high-speed-pcb-design/ (please find other source, do not promote Autodesk)

but it's also the primary mission in DC/DC converter design to make very bad antennas, as unintentional radiators are what causes device to fail radiated emissions compliance.

​​​​"The goal of our DC/DC converter design is to make 
​​​​bad antennas out of the loops that carry the high di/dt currents."

https://www.ti.com/lit/an/snva638a/snva638a.pdf?ts=1650145296240

Things to look out for thus include measures that would otherwise be implemented to prevent radiated emissions.

Look out for

  • conductive elements near the antenna, especially extended planes, but also stray wiring or screws,
  • metalized stickers on the enclosure,
  • plastic elements next to or on top of the antenna (keep ~10mm distance, some module datasheets (u.blox?) specify good practices rules),
  • larger circuit components next to the antenna and connectors.

The author has first-hand experience in ruining the range of bluetooth modules mounted centimeters next to steel brackets, and completely eliminating module function with a drop of PU resin seeping into RF shielded sections.

Bad design is everywhere though. On the bright side, some simple methods can help ensure design success, and in turn help build intuition.

Below you may find a few basics covering the characteristics of PCB antennas (mostly for 2.4 GHz), followed by a growing list of examples found out in the wild that are likely to have degraded performance by making individual sets of mistakes when implementing a design containing such an antenna.

Finally, some references are given to help with one's own antenna design and integration, which come down to:

  • understanding factors that lead to detuning of antenna resonance frequencies and altered radiation patterns,
  • design (or use) of planar inverted-F antennas (PIFA) taylored to the final PCB / PCB stack thickness,
  • physically tuning antenna structures in prototypes, and
  • finding appropriate values for the impedance matching netork.

E and H Fields

FEM and FDTD simulations are set up and run routinely to extract characteristic curves (S-parameters and impedance linked to them), but also to obtain the radiation pattern in the far field (via near-field to far-field transform (NF2FF), which is also used experimentally).

E and H fields in time domain however can be extracted at every location in the simulation volume and their plots help understand how an antenna structure is not just a shape on the PCB, but include the space around it.

JTL Engineering B.V. (not affiliated) shares these views of the electric and magnetic fields.

E-Field: as the antenna structure resonates, the highest voltages are observed at the far end of the meandered trace. The animation however shows the electric field as the gradient strength, putting the electric potential in context with the distance to the nearest ground plane.

The take-away from the animated plot is that the antenna structure "sees" (is affected by) a fair bit of its immediate surroundings.

H-Field: Energy in the magnetic field is mostly localized around the inductive loop and meander near the feed line. At 2.4 GHz, the skin depth in Cu is below 1.4 µm, so even a 25µm thick inner layer is not penetrated.

Take-away: even though the shorting path is at ground potential near DC, it is surrounded by an AC magnetic field which cannot be ignored. No components should be placed closer than a few mm away from the loop, and no copper pour should impede the free space shape of the magnetic field lines.

Effect Of Ground Plane Under The Antenna

​​​​Ground Effects 
​​​​The size and shape of the ground plane will affect the radiation pattern. 
​​​​Figure 15 shows an example of how the ground plane affects the radiation pattern. 
​​​​If for example a GND plane is extended, when an antenna board is being plugged 
​​​​onto a base board, this has effects to the antenna match compared to using the 
​​​​antenna board as stand alone. The change in size and shape of the ground plane 
​​​​not only changes the gain but the radiation pattern. Since many SRD applications 
​​​​are mobile, it is not always the peak gain that is most interesting. The TRP 
​​​​and antenna efficiency gives a better indication on power level that is transmitted 
​​​​from the DUT. In Figure 15 one can see that the toroid is flattened 
​​​​in the bottom area, which will result in no power output in that direction.

(source)

A Selection Of Sins

Metal Next To The Antenna

  • metal and mounting hole left of the antenna
  • copper and a connector on the right side of the antenna.


https://aliexpress.com/item/4000437975620.html

  • antenna not on the edge of the PCB assembly
  • antenna surrounded by copper pour

https://hackaday.io/project/184660-9-pisquare-can-connect-multiple-hats-wirelessly

  • antenna surrounded by copper pour

LTT: https://www.youtube.com/watch?v=QoC8R3KHk8E

… which is probably this relay board product:

https://usa.banggood.com/5V-or-7-28V-Power-Supply-8-Channel-ESP8266-WIFI-8-way-Relay-Module-ESP-12F-Development-Board-Secondary-Development-Board-p-1833055.html

A rather interesting antenna design comes under the name "ProAnt Niche", a planar resonant cavity antenna licensed by the Pi foundation, put into its BT / Wifi supporting products, and probably regularly suffering from sub-optimal use in target applications:

  • antenna in the middle of the PCB
  • antenna surrounded by copper and other conductive elements
  • FR4 material likely not removed underneath antenna structure
  • stackable design guides users towards adding a copper plane in 12-15mm distance over the antenna structure, with two of the sides connected to both PCBs via stacking headers to create a partial shielding cage.

Information regarding the design and how mistakes could have been prevented are openly available.

To quote from Hardware Design With RP2040:

​​​​If anything is placed close to the antenna (in any dimension) 
​​​​the effectiveness of the antenna is reduced. Raspberry Pi Pico W 
​​​​should be placed on the edge of a board and not enclosed in metal 
​​​​to avoid creating a Faraday cage."

Obviously, there is an abundance of information on how to design with this antenna-containing module, as the Pico W eval board shows in no uncretain terms: the antenna structure is supposed to be at the edge of the parent PCB, and dielectric material is supposed to be removed underneath:

The Kickstarter images do not expose how the underside of the two products looks, but it would appear that they lack the requisite antenna-shaped cutouts.

Metal Under Antenna

  • groundplane underneath antenna
  • connector right next to antenna
  • PCB traces routed underneath antenna

later released as a Kickstarter project, leaving the groundplane in place.

https://www.kickstarter.com/projects/picobarcodescanner/roundy-round-lcd-board-based-on-pico

Copper Fill Into Antenna Structure

  • non-functional copper fill into meander pattern
  • meander has suspicious aspect ratio
  • no ENIG finish
  • antenna not bounded by groundplane edge but components (original PCB has top and bottom groundplane and via stitching)
  • backside is also solid copper infill


https://hackaday.com/2022/03/26/reverse-engineering-your-own-bluetooth-audio-module/

Dielectric Material On Antenna

  • perforated PCB might be acceptable
  • connectors, cables and probably traces next to, under and over antenna
  • large blob of hot-snot on antenna

https://hackaday.com/2018/06/07/wifi-pool-controller-only-cost-20/

  • module with antenna extends over the edge of the parent PCB, unclear whether the antenna is designed for the resulting PCB thickness
  • heat shrink over antenna, effect unclear

https://de.aliexpress.com/item/32699819094.html?

Anecdotally, here's a report on enclosures influencing GPS reception. With 1-3mm of plastic on top of a dielectric resonator antenna, the number of satellites seen and accuracy achieved indeed dropped in a quick experiment. More distance from the antenna might be needed.

https://meshtastic.discourse.group/t/the-importance-of-gps-antennas-and-request-to-3d-case-documentation-people/1505/3

What Needs To Be Done

Design rules-of-thumb, PCB manufacturing issues and when/when not to worry about RF aspects:
"Practical RF Hardware and PCB Design Tips - Phil's Lab #19"

https://www.youtube.com/watch?v=_Hfzq1QES-Q

If you want the full antenna design experience:

https://lcantennas.com/a-complete-guide-for-pcb-2-4g-antenna-design/

Known good antenna designs are available, e.g.

https://www.silabs.com/documents/public/application-notes/an1088-designing-with-pcb-antenna.pdf

If you fancy the "community edition", inverted-F antennas are now also part of a KiCad library:

https://twitter.com/sad_electronics/status/1517874391803179008
https://github.com/sad-electronics/wch-kicad-lbr

This AliExpress item also looks very suspiciously like someone copy-pasted a known good meandered inverted-F antenna design:

Once implemented, they should be measured or tuned:

"Tektronix - If you’re integrating an off-the-shelf antenna with a wireless module, you need to ensure that the impedance of the antenna matches the impedance of the wireless module. Mismatch can result in reduced data packets, signal range and wasted battery life.
In this video, we show you how to improve the design and verify the performance of the antenna matching circuit placed between the antenna and wireless transceiver module, using the TTR500 Network Vector Analyzer."

take-away is also that footprints for matching network components always need to be in place for proper adjustment capabilities. Real-world performance always has the final say.

https://www.youtube.com/watch?v=recnhI5Uj7w

ESP8266 and kin almost usually have recommendations or unpopulated matching network components in their typical application schematics. A common sight is that the shunt inductors have been omitted and only the series capacitor is placed.

sometimes the antenna structure itself needs mechanical trimming to adjust:

https://twitter.com/maanil_ee/status/1511392806350245891
https://twitter.com/maanil_ee/status/1511398050043998208

Uncertainty about whether to mount on top of a PCB or free-standing never gets old.


https://twitter.com/mikerankin/status/1578787493209866241

ESP32 design guide: https://www.espressif.com/sites/default/files/documentation/esp32_hardware_design_guidelines_en.pdf

Adding u.fl connector footprints

https://www.megiq.com/blog/item/40-ufl-connectors-in-vna-measurements


https://www.gsm-modem.de/M2M/antenna-test/antenna-tests/

Systematic antenna matching with a VNA by trimming:

https://www.lairdconnect.com/resources/white-papers/antenna-matching-within-enclosure

Antenna on a human wrist


https://ieeexplore.ieee.org/document/5615201
https://cora.ucc.ie/bitstream/handle/10468/518/JB_DetuningAV2010.pdf?isAllowed=y&sequence=1

Good-looking Implementations

PCB Chip Antenna Hardware Design - Phil's Lab #130

02:02 Trace vs Chip Antenna
04:55 Pre-Certified Modules
05:58 Chip Antenna Selection
12:54 Matching, Tuning, and Schematic
20:27 Footprint
22:31 PCB Design

https://www.youtube.com/watch?v=tdUZ2IjO9do

tCam Mini (Ver. 5)

  • free-standing PIFA
  • some distance to mounting holes
  • no components left and right of antenna


https://circuitstate.com/featured/tcam-mini-open-source-wireless-thermal-imaging-module-based-on-flir-lepton-3-5-and-esp32/

previous iteration (Ver. 4) shows the antenna still on a piece of FR4, potentially impacting performance:


http://www.danjuliodesigns.com/products/tcam_mini.html

Future Developments: Aperture Tuning

In the future, we may perhaps be seeing modules that integrate a vacator diode (https://shop.richardsonrfpd.com/docs/rfpd/Skyworks Low Cost Antenna Tuning using Skyworks PIN Diodes.pdf) driven via a PWM channel (or other low-cost alternatives to digitally tuned capacitors) to correct for antenna deturning.

If 2 bit are enough for coarse corrections RichWave RTC6603, RTC6607 SPDTs or similar (RTC6617S SP3T, Peregrine Semiconductor / muRata PE613050 SP4T) may see some use.

From https://doi.org/10.1109/APMC46564.2019.9038800 :

While a rare sight in low-cost applications, antenna tuners have long become a relevant component in 4G/5G systems (see https://www.qorvo.com/design-hub/blog/4-things-to-know-about-antenna-tuning-in-4g-5g-smartphones).

Under particular conditions, an alternative antenna can be used with an already certified product. (see https://linxtechnologies.com/wp/blog-post/antenna-fcc-certification/)

"CFR 47 Part 2.1043 defines three types of permissive changes

The summary of these documents is that a Class 1 permissive change is allowed if the new antenna is equivalent to the antenna with which the product was tested. They define an equivalent antenna as well.

Equivalent antennas must be of the same type (e.g., yagi, dish, etc.), must be of equal or less gain than an antenna previously authorized under the same grant of certification (FCC ID), and must have similar in-band and out-of-band characteristics (consult specification sheet for cutoff frequencies).

This gives three conditions that must be met:

  • Same type
  • Equal or lesser gain
  • Similar in-band and out-of-band characteristics "

The general case though is a full qualification process, as referenced below for the example of the ESP12F module.


https://fcc.report/FCC-ID/2AHMR-ESP12F

Other Lab Setups

From the aforementioned Pico W test:

https://fcc.report/FCC-ID/2ABCB-PICOW/5969741

Select a repo