BLE or LoRa? Picking the radio before picking the chip
BLE or LoRa? How range, power budget, gateway cost and who receives the data settle the radio choice before you commit to a microcontroller.

BLE or LoRa is usually the first real architecture question on a connected product, and it is often answered last. Teams pick a microcontroller they like, build a breadboard demo, and only then discover the radio they need does not fit the chip, the battery or the budget for infrastructure. We think the radio decision should come first, because it constrains almost everything downstream.
This post is for founders and product managers who need to make that call early, and for engineers who want a quick framework to check their instinct against. It covers the questions we ask, the figures that matter and the cases where neither radio is the right answer.
Why the radio comes before the chip
The radio sets your range, your power budget, your data model and whether you need infrastructure at all. Once those are fixed, the shortlist of microcontrollers and modules shrinks to a handful, and the choice between them becomes a question of cost, tooling and supply.
Going the other way round causes predictable pain. A team that standardises on a Wi-Fi and BLE part early may later find the device has to report from a field two kilometres from the nearest building. A team that starts with LoRa may find users expect to open a phone app and see live readings. Neither problem is fixed in firmware.
Three questions that settle BLE or LoRa
On most projects, three questions get us to an answer quickly. We ask them in this order, because each one can rule an option out on its own.
- Who or what receives the data? If a phone is the natural receiver, BLE has a huge head start: every modern smartphone has it, and no extra hardware is needed. If the data has to reach a server without a phone nearby, you need a gateway of some kind, and LoRa becomes a serious option.
- How far, and through what? Tens of metres indoors points to BLE. Hundreds of metres to several kilometres, through walls, across a farm or a building site, points to LoRa.
- How much data, how often? BLE can move firmware images, logs and live streams. LoRa is built for small payloads, typically a few to a few dozen bytes, sent minutes or hours apart.
If the answers point in different directions, that is a useful signal too. It often means the product needs both radios, or a different one entirely.
Range and data rate: what to expect in practice
Datasheet range figures assume line of sight and ideal antennas. Real enclosures, bodies, metal and walls take a large bite out of them, so we plan with conservative numbers and verify on the bench and in the field.
| BLE (1M PHY) | LoRa (EU868) | |
|---|---|---|
| Frequency | 2.4 GHz | Sub-GHz, 863–870 MHz |
| Typical indoor range | 10–30 m | Several floors or a large building |
| Typical outdoor range | Tens of metres to ~100 m | A few km urban, 10 km or more with line of sight |
| Usable data rate | Tens to hundreds of kbit/s | Roughly 0.3–5.5 kbit/s |
| Phone can talk to it directly | Yes | No |
BLE 5 adds the Coded PHY, which trades data rate for range. At 125 kbit/s it can reach several times the distance of the standard PHY, but phone support for it is still uneven, so we do not rely on it when a phone is the receiver.
LoRa range comes from its spreading factor. Higher spreading factors (up to SF12) reach further but take much longer on air. A ten-byte payload might take around 50 ms at SF7 and well over a second at SF12. That airtime is the cost you pay in battery, in regulatory duty cycle and in network capacity.
Power budget: the average matters, not the peak
Both radios can run for years on a battery, but they get there in different ways. What matters is the average current over a full duty cycle, not the transmit peak on the datasheet.
- BLE transmits at low power, typically a few milliamps at 0 dBm on parts such as the nRF52840, for very short bursts. The cost comes from how often you advertise or keep a connection alive. A device advertising every second behaves very differently from one advertising every ten.
- LoRa transmitters such as the SX1262 draw tens of milliamps, rising above 100 mA at the highest output power, for tens of milliseconds to over a second per packet. Because transmissions are rare, the average can still be very low.
- Sleep current often dominates both. A microcontroller sleeping at a few microamps is easily undone by a regulator, sensor or pull-up that never sleeps.
We model the budget in a spreadsheet early, then measure it on the first boards with a current probe. The two numbers rarely agree on the first try, and the gap tells you where to look.
Gateways, networks and the cost nobody budgets for
BLE’s hidden infrastructure is the phone, which the user already owns. That is a large advantage, but it means data only moves when the phone is nearby and the app is allowed to run. If the product must report while nobody is around, you need a BLE gateway, and at that point the comparison changes.
LoRa needs a gateway within range. The options are a public or community LoRaWAN network where coverage exists, a commercial network operator, or your own gateways. Owning gateways gives you control and coverage where you need it, but adds hardware, backhaul, installation and maintenance to the bill of materials of the whole system rather than the device.
LoRaWAN also has rules worth knowing before you commit. In the EU most sub-bands are limited to a 1% duty cycle, community networks apply fair-use limits on airtime, and Class A devices can only receive in two short windows after each uplink. That makes commands and configuration changes slow, and firmware updates over LoRa impractical for most products.
When the answer is both, or neither
A common pattern on field devices is LoRa for the data link and BLE for commissioning. The installer configures the device and reads diagnostics with a phone, then it reports over LoRa for years. The cost is a second radio, or a combined module, and a little more certification work.
Sometimes neither radio fits. If the device has mains power and sits near a router, Wi-Fi is simpler. If it moves across a country or must report anywhere without infrastructure you control, LTE-M or NB-IoT is usually the better answer, at the cost of a SIM, a data plan and higher peak current.
Picking the chip once the radio is settled
With the radio fixed, the hardware shortlist is short. For BLE-only products we usually look at the nRF52 family or STM32WB. For LoRa, the STM32WL integrates the sub-GHz radio with the microcontroller, while a separate SX126x transceiver lets you pair it with whichever MCU suits the rest of the design. If Wi-Fi is also needed, an ESP32-family part becomes attractive.
We lean towards pre-certified modules for small and medium volumes. They cost more per unit than a chip-down design, but they reduce RF layout risk and simplify CE RED and FCC Part 15 work. The trade-off is worth revisiting once volumes grow. Our hardware, firmware and testing services cover this whole path, from radio choice to pre-certification.
What we would do before committing
Before any schematic, we would work through this list:
- Write down who receives the data, how far away they are and how often the data has to arrive.
- Estimate payload size and frequency, then work out airtime for LoRa or advertising and connection intervals for BLE.
- Build a first power budget, including sleep currents for every part on the board.
- Decide who owns the infrastructure: the user’s phone, a network operator or your own gateways.
- Test range with dev kits in the real environment, inside a rough enclosure, before you lay out a board.
- Check how the radio choice affects the backend and app, including what a fleet dashboard can show if devices report only a few times an hour.
An afternoon with two dev kits and a car is cheaper than a board respin. Pick the radio first, prove it in the field, and the chip choice becomes the easy part.