JavaScript is off: some build content is not available.Parts lists, generated test-plan tables, tabs, copy and downloads are populated by JavaScript. Static explanations and diagrams remain below; enable JavaScript to use the complete build reference.

LoRa mesh · Planned build files · Rev. Oct 22, 2026

Three radios, no towers, and a map of where they stop talking.

Three Meshtastic nodes on 915 MHz pass text messages across campus with no cell or Wi-Fi service, relaying through each other when two are out of direct range. The study measures how far one hop reaches on campus, and what a relay costs in delivery rate and delay.

Companion: Hammerhead speed quad build files ↗.

Nodes3 Heltec V3, ESP32-S3 + SX1262

Link budget152.5 dB, 21 dBm to −131.5 dBm

Airtime0.56 s per 40-byte packet

Parts cost$125.82 all three nodes, Oct 2026

A / Bill of materials

Three nodes for about $42 each

Prices are what I paid on Oct 11, 2026, in USD before tax and shipping. Nothing here needs soldering: the boards come pre-flashed and the batteries plug straight in.

Qty	Part	Why	Source	Each	Line

Heltec now sells a V4 with a stronger 28 dBm transmitter. The V3 is used here because its lower power makes relays easier to force on a small campus.

B / Node & setup

One board, four connections

Each node is a Heltec V3 with a battery, an antenna and a phone or laptop on the other end. The settings table fixes the radio configuration so every test runs the same way.

Block diagram of one node. All three are identical.Fig. B1

Before the first power-up

Screw the antenna on first. Transmitting into an open connector reflects the power back into the radio. If you swap in any battery other than Rokland's Heltec-polarity pack, check the plug with a meter: many 1S packs are wired the other way.

Connection schedule

4 runs

#	Connection	Part	Notes

Meshtastic settings

All three nodes

Setting	Value	Why
Region	US (902–928 MHz)	Legal band and power limits for the US.
Modem preset	LONG_FAST	SF11, 250 kHz bandwidth, coding rate 4/5. The network default, so results compare with other people's.
Hop limit	3	Default. The relay test needs only 1.
Device role	CLIENT on all three	Every node relays the same way, so the relay result isn't a special router setting.
Channel	Private name and key	Keeps the logs to my own traffic. Other nodes on the same preset can still relay, which L10 checks.
Range Test module	Sender on A, every 30 s	Numbered packets make missed ones countable.

Power budget

Per node

Load case	Value	Basis
Transmit burst, SX1262 at +22 dBm	~118 mA	SX1262 datasheet
Average, BLE connected, screen on	60–110 mA	estimate measured in L02
Usable capacity, 80% of 2,000 mAh	1,600 mAh	Stop at about 3.5 V
Runtime	15–27 h	1,600 mAh ÷ average current

C / Radio math

What the radio should do, before it's tested

The link budget sets how much signal can be lost before packets stop arriving. How fast it's lost with distance depends on terrain, which is the number this study measures.

Link budget

LONG_FAST, SF11, 250 kHz

Term	Value	Note
Transmit power	+21 dBm	Heltec V3 rating
Antenna gain, both ends	0 dBi	assumed stock antennas
Noise floor, 250 kHz	−120.0 dBm	−174 + 10·log₁₀(250,000)
Receiver noise figure	+6 dB	Typical SX1262
SNR needed at SF11	−17.5 dB	LoRa can decode below the noise
Sensitivity	−131.5 dBm	Sum of the three rows above
Link budget	152.5 dB	Transmit power minus sensitivity
Fade margin	10 dB	Body blocking, multipath
Allowed path loss	142.5 dB	Used for the range estimates

Path-loss model

Log-distance

PL(d) = 31.7 dB + 10 · n · log₁₀(d / 1 m)

31.7 dB is free-space loss over the first metre at 915 MHz. Free space is n = 2. Real ground-level links run 2.7 to 4, and a handheld node 1 m off the ground loses most of its clearance. The range for each n is where the line crosses 142.5 dB.

Path-loss exponent n	Typical setting	Range at 142.5 dB
2.7	Open, elevated	12.8 km
3.0	Open ground, handheld	5.0 km
3.5	Mixed campus	1.5 km
4.0	Through buildings	0.59 km

Range against path-loss exponent

Model, not data

Each line is one assumed environment. The measured RSSI points get fitted to find which line campus actually follows.Fig. C1

Airtime

8.19 ms per symbol · 1.07 kbit/s

Payload	Symbols	Time on air
20 bytes	28 + 20.25 preamble	0.40 s
40 bytes	48 + 20.25	0.56 s
60 bytes	63 + 20.25	0.68 s
100 bytes	103 + 20.25	1.01 s

Packets include Meshtastic's header, so a short text message lands near 40 bytes. Each relay adds at least one more airtime plus a wait before rebroadcast.

Procedure

Three weeks

Bench check. All three nodes on one desk; 50 messages each way.
Range walk. A stays at a marked spot 1 m off the ground. C walks outward along a line-of-sight route and a through-buildings route. The laptop logger records RSSI, SNR and position for every Range Test packet.
Fit n. Regress RSSI against log₁₀(distance); the slope is −10n. Mark where the rolling delivery rate drops below 90%.
Relay test. Put B near the edge of A's measured range and C past it. Send 100 messages A → C and log delivery, hop count and delay.
Control. Turn B off, move C back into direct range, and repeat. The difference is the cost of one relay.

Relay test layout

Fig. C2

Distances come from the range walk, not a fixed plan.Fig. C2

D / Test log

Each prediction, and how it gets tested

Predictions and pass bands are written before testing, so a result can't be explained away later. A miss gets reported here like any other result.

Planned build document — not measured test results. Every test entry is a prediction or proposed pass band; no measured results are reported here.

ID	Quantity	Predicted	How it's measured	Pass band	When	Measured	Status

Limits of this study

Three nodes and one campus. The fitted path-loss exponent describes these routes on the days they were walked, not LoRa in general. Weather, foot traffic and where the phone sits in a pocket all move RSSI by several dB; repeat runs show how much.