Every LoRa project starts with a range figure, and that figure almost always comes from ideal conditions. MeshTRX claims five kilometres: line of sight, clean air, nothing between the antennas. This article is about what the first field measurements showed once the radios reached people who carry them around a city.
What actually happened
The first detailed report from the community read like this: solid voice up to two kilometres, and beyond two, text still gets through. Voice gave up earlier — around a kilometre in dense housing, while text on the same route worked almost out to two.
That is not a disappointment, it is useful knowledge. Five kilometres remains true for line of sight — but only for line of sight. In a city, between buildings, with the radio in a pocket, the numbers are different, and it is more honest to talk about those.
Why voice gives up before text
The difference is not power or antenna. It is how the transmission works.
A text message is one packet. It needs one good window in the air. Did not make it? The app retries a few seconds later, then again, and sooner or later it lands in a moment when the interference stepped back. The person barely notices: the message arrives a few seconds late.
Voice cannot do that. Speech is a stream: twelve packets a second, each of which must arrive on time. Retrying is pointless — by the time the retry lands, the moment is gone. A dozen packets in a row must cross the same air that text was pushing through on its third attempt.
Hence a rule worth remembering: where voice no longer gets through, text usually still works. And the other way round — if text is gone too, the problem is not voice but the link itself.
About comparisons with sensor networks
The same discussion raised a point: neighbouring nodes running different firmware, also on LoRa, reach further. That is true, and the reason is clear.
Sensor networks default to slow modes — a high spreading factor, a narrow band. The slower the transmission, the further it is heard: that is physics, not firmware quality. The price of slowness is throughput. In such a mode a short message squeezes into the air every few seconds, and speech is simply out of the question.
MeshTRX runs a fast mode precisely because it carries voice. Voice needs bandwidth, bandwidth needs speed, speed costs range. That is not an oversight but a deliberate choice, and one worth knowing up front.
A long-range mode for text — a separate, slower transmission that reaches further — is on the roadmap. It will not replace the main mode but complement it: text will travel to places where voice no longer lives.
What a repeater changes
A repeater doubles the path: the radio reaches it, and it reaches further. In practice this was confirmed — voice through the repeater started arriving where it did not arrive directly.
But a repeater has a cost that is rarely mentioned. It works in the same channel and the same half-duplex as everyone else: while it repeats someone's packet, it cannot hear the next one. On the test bench this shows up in numbers — of twenty voice packets, about half make it through the repeater. For speech that is tolerable: an eighty-millisecond gap sounds like a click, not a hole. For a file it would be a disaster, which is why files travel with acknowledgements and re-requests for the missing pieces.
The conclusion is simple: a repeater increases coverage, not the capacity of the air. It helps where there would otherwise be no link at all.
What comes next
Measurements continue, and the next step is obvious: directional antennas and a high vantage point. A repeater on the tenth floor in the city centre, in line of sight, is an entirely different proposition from a repeater on a desk.
Beyond that lies a question the community is already discussing: how to link several repeaters so that each serves its own stations without interfering with the others. That is not one setting — it needs zones, stations bound to their own repeater, and a separate path between repeaters. Such work is on the roadmap, and it makes sense to start it after measurements from a high point show the real distances.
Measurements beat assumptions. That is exactly why this article is written from field numbers rather than calculations.
How MeshTRX works as a whole is in the documentation. What is done and what is deferred is on the project page. Measurements, observations and arguments happen in the Telegram group.