If you were to overlap everything non-intuitive in electrical engineering you might call that place “antenna design”. Luckily, the 2.4 GHz band has a few players (Wi-Fi, Bluetooth, etc.), so there are resources. You might find along the way that your PCB size needs to be small, but your wireless range needs to be large, in which case you are (1) doing engineering (2) need to make some choices. I want to share with you a few resources and practical tips.
Design guides
PCB trace antennas are a great option if you have the space (think a minimum of 8mm x 15mm with a generous offset from anything else, especially your ground plane). If you don’t have that kind of space, you could either compromise performance with a chip antenna or find a configuration for a wire antenna (which in theory has the best performance of all options).
By far, I like the Cypress Antenna Design and RF Layout Guidelines AN91445 the most. It’s quite practical and easy to follow. While you’re considering antenna type, check out the Texas Instruments Antenna Quick Selection Guide (the top row is all 2.4 GHz).
NXP Compact Integrated Antennas AN2731 gives a little more theory concerning how antennas and different designs actually work. If you’re interested in a particular type of antenna, these guides are more specific:
- Designing with an Inverted-F 2.4 GHz PCB Antenna AN1088
- Low cost PCB antenna for 2.4 GHz radio: meander design for STM32WB Series AN5129
- λ/4 printed monopole antenna for 2.45 GHz Whitepaper
Some practical insights
Can a simple wire antenna outperform a properly designed trace antenna? If it can, it removes a lot of the burden of board design and tuning.
I set up a basic broadcast routine on my nRF5340-PDK and compared the transmission strength of several antennas (by desoldering the SWF connector) using the nRF Connect app on my iPhone at distances of 0, 1, and 2 meters. I report the mean signal strength which is the average of a max and min reading I observed over roughly 15 seconds. The antenna was 1mm copper soldered at 90 degrees and clipped to each length.
Results
The best performance came from 25-30mm length wire (1mm diameter) which outperformed the trace antenna, but only slightly. This is consistent with the fact that this length is one-quarter wavelength of 2.4 GHz. I should also note, using copper of a similar diameter (1.6mm didn’t noticeably affect performance. It might be interesting in the future to characterize the radiation pattern of each, as the wire is almost certainly less sensitive to direction.
Because a stiff, soldered antenna is less than ideal, I tried using some flexible core from a Wi-Fi whip (or “rubber ducky”) antenna that I gutted. Indeed, performance at 27mm length was great, and consistent with the upright copper antenna. As you might expect, twisting and turning the wire significantly impacts its performance.
Designs in the wild
While looking for clever solutions I ripped apart two Bluetooth-enabled devices to see how they solved the problem of performance with physical size constraints. First was a broken Treks Titanium headset that employed an inverted-F style antenna made from a thin piece of copper. This strikes me as a very clever way of achieving performance when your vertical clearance is sufficient and/or dictated by other components (e.g., a micro-USB connector).
The other device was a Google Home which had two planar inverted-F antennas in a circular configuration to fit the device shape (the second one is not shown, near the bottom-right corner of the image). You’ll appreciate the care taken to stitch the ground planes.
Final thoughts
Tuning your antenna can be downright prohibitive if you are on a budget. Even if you are copying a design, it’s recommended that you characterize your own antenna and expect that it might benefit from tuning. Some designs allow you to make your antenna longer than you might need and clip/etch away material until you reach peak performance. The cheapest way is to do something like I did, using broadcast strength as a surrogate for the quality of your antenna configuration. One step up would be to purchase a vector impedance analyzer for $162 to help you in the tuning stage. If you have any insights, ideas, tips, or tricks, please share them in the comments.
Addendum
A wire antenna can work really well but not if the antenna shape is compromised. For an implantable device, the chip antenna ends up being a smart idea. I have been very impressed with the performance of the Johanson Antennas as well as their technical support; their layout guidelines also appear to be ‘applications-based’ and understand that you are working with constrained space. Here, I have a 2450AT42B100E connected through their balun solution, the 2450BM14G001 (using a TI CC2652RB MCU).
The performance is very impressive: at full TX power, I can receive packets up to 35 meters in an unobstructed environment. I used a 4-layer PCB with a top- and second-layer ground plane with as much pour as I could manage on the top plane (and no ground beneath the antenna through all four layers). I have not tuned any of the components; if you have a mission-critical design, I would suggest getting in touch with Johanson, as they provide a simulation service based on your PCB design.