PCB/SMD ~920MHz Antenna for Very Small PCB (Low Ground Area)

[Diozoh]

My board is 35 mm x 20 mm with RF communications aimed at around 920Mhz (915-930 acceptable) and I intend to have the antenna output run to some kind of fixed antenna on the PCB. The signal will travel through a 0402 component which can be moved to divert to a uFL connector to allow for the use of an external antenna if needed.

I'm having a lot of trouble finding an antenna that does not drop abysmally in effectiveness when paired with my PCB dimensions.

Every design for either chip or PCB antennas I have found require a large (relatively) ground area in the non-antenna segments of the board which is several times larger than my entire board.

Documentation about the effect of the ground size:
Antenova (Chip).
Linx (Chip)
Abracon (Chip vs. Trace)

PCB trace antennas also specify ground dimensions for reference designs so I'm assuming that they have the same issue.

The best PCB trace antenna I've found (size-wise) is from TI's DN024, which fits vertically but will require matching (especially as my board is likely to be different thickness than the 0.8 they tested with)

Here is the TI PCB trace antenna next to my (unfinished) board.

Is there any design or method that can avoid dropping from 60% to 10% efficiency without significantly increasing the board size to increase ground area?

This is my first RF design (and first PCB) and I do not have the ability to run simulations, so any help would be appreciated.

[mayor]

Out of curiosity, how are you determining this drop in performance?

I have built 915MHz boards of roughly 40mm x 20mm with that antenna, and their range is excellent in the real world.

If you're curious: https://easyeda.com/martin/cc1352-breakout_copy.

[Diozoh]

Out of curiosity, how are you determining this drop in performance?

At first I was directed to the Linx application note in my original post when going through a datasheet for their 915 MHz "microSplatch" product and went on to find several similar documents from other manufacturers all saying the same thing.

The explanation I found is that the chip is only half of the antenna and the other half is the ground plane, which is why they specify the ground dimensions on the datasheets. Further investigation of PCB trace antennas (especially from the TI antenna guide) yielded the same type of specification with dimensions given for the ground plane in addition to the keepout area.

Your board looks similar to what I'd have aimed for dimensionally, did you use the same board thickness as the reference document for the helical antenna or did you adjust the matching?

Also when you say range is excellent, what is the max. effective distance for your boards assuming 14 dBm TX from the CC1312?

[bookaboo]

It may still be too large for your project but might be worth a look:
https://www.mouser.co.uk/ProductDetail/PulseLarsen-Antennas/W3139?qs=0lSvoLzn4L9wvYcEf4D2gA==

I recently used them on an 868MHz design with great results, for me it wasn't worth tweaking with an F or chip antenna for weeks.

[Diozoh]

It may still be too large for your project but might be worth a look

So from the datasheet we get -1 Gain at 915 (+/- 13) but that measurement comes from a reference board with these dimensions:



I'm confident that this will experience the same issue as covered in the reduced ground area testing in the links I found since it appears to be proportional. I'm starting to suspect that this may be one of those "physical constant" type limitations where it just has to be a certain size because of the frequency.

Might be a good idea to locate the option with the smallest ground plane size in the test reference as I suspect that will retain a value closer to the datasheet when scaled down. The question is whether that would be better than something which drops more but starts at a higher value, no way to know without testing.

[bookaboo]

I had a nice ground plane 60,mm*60mm in my design so was lucky in that respect. The other thing I've noted about helical antennas is that they have are somewhat omni directional and are not as fussy about orientation or copper nearby. Anecdotal so run your own tests.

[mayor]

@Diozoh

I tested my boards and ran out of LOS space at ~600 meters, at 50kbps, PER=0%. That was plenty for my needs, but I suspect that they would go much beyond that range if I could find more space to test.

Thickness: the standard JLCPCB thickness matches the reference design well enough. But, as you suspect, a proper antenna match is necessary. I managed to get < 1.1:1 VSWR, bandwidth at roughly 50MHz, if I remember correctly. Plenty good for my home gamer stuff!