Table of Contents
LoRa Antennas
Designs to consider
- 1/4 wave ground plane
- 5/8 λ
- Dipole
- Sleeve dipole
- EFHW
- Collinear
Design centre frequency for UK / EU: 869.525 MHz
PCB Antennas
Inverted F
TI design note: https://www.ti.com/lit/an/swra228c/swra228c.pdf
Polarisation
There was a suggestion on mastodon to use circular polarisation to mitigate multipath losses from sea reflections.
Supplied with dev boards
Tested small antenna that came with Heltec Wireless Tracker with a LiteVNA.
Short black plastic rod. SMA male.
It has a peak at around 985MHz. At 868MHz it is pretty much useless.
Tested again on end of SMA test lead supplied with VNA. Peak closer to 1GHz. It lowers when I touch the connector. Highly sensitive to nearby hand, touching, nearby objects. In general very poor, nearly useless.
Antenna that came with Waveshare module for Pi pico, with rigth angle connector, is much better, but still sensitive to position / touch.
Antenna with Lilygo T-beam looks the same as heltec ones except it is marked 868MHz. It is much better, somewhat usable, but sensitive as the others.
Performance on all was better connected directly to VNA. With test lead, resonance, if any, went up to near 1GHz.
Performance with supplied u.fl → SMA pigtail and SMA u.FL adapter on the VNA was quite good. I guess the length of the pigtail was suitable to from the ground side of the dipole which is missing from the (probably) monopoles of the antennas.
The separately purchased Heltec paddle antenna is sort of ok at 868MHz. It has a bunch of peaks not quite where you want them.
ebay one, grey, "8dBi"
Quick test with LiteVNA, mounted directly on prototype nyan boxes, both metal and plastic. Results about the same on both. Return loss from about 15dB to 19dB, depending on position relative to my body and surrounding stuff, and whether I touched the metal base or not. Touching made more difference on the plastic box.
It seems tuned to about 854MHz, at which point it does much much better than at the tested 869.24MHz. At one point 30dB RL.
These measurements were taken without considering the added uFL → N pigtail, so should be taken with a pinch of solt.
Cables
Tested pigtails that came with devboards. Losses at 882MHz (a point used by the VNA): 100mm cable 0.1dB, 150mm 0.2dB.
Some random products / linkdump
LoRa Fiberglass Antenna, 858-878MHz, 8dBi, 1300mm £59.00 https://uk.robotshop.com/products/lora-fiberglass-antenna-858-878mhz-8dbi-1300mm