LD2410 Sensor: Sees You Even When You Sit Still

A PIR sensor sees heat moving across its field. Sit down to read and, after its timeout, the light goes off. The HLK-LD2410 human presence sensor uses FMCW radar instead: it sweeps its frequency, measures the echo, and picks up movement as small as breathing and posture shifts. It then tells you whether it sees a moving target, a still target, or nobody, and how far away each is.

• Lights that stay on while you read, type or watch a film.
• Heating that holds a room warm while it is occupied and sets back when it empties.
• Screens and fans that switch off when nobody is left at the desk.
• Bed and chair occupancy for night-time and away automations.

The module needs a 5V supply and talks over a 3.3V serial port at 256000 baud, with a simple OUT pin that goes high while someone is present. That makes it an easy partner for an ESP32 running ESPHome, which has a built-in LD2410 component for Home Assistant.

The three most popular LD2410 buys are all the LD2410C, the version most makers start with: Bluetooth tuning from a phone and standard 2.54mm pins. They differ only in quantity. One module is for trying the radar in a single room, the 3-pack covers a few rooms at a lower cost per sensor, and the 4-pack suits a whole-house rollout.

Radar passes through thin walls and sees fans and curtains, so every install needs a few minutes of tuning. Per-gate thresholds, set while you watch live energy readings, handle most of it. The range and gates guide walks through the method.

The first question matters most. Your build path decides whether you want a bare module, a test kit or a finished sensor.

• ESPHome or Home Assistant: a bare LD2410C or B on an ESP32. ESPHome's ld2410 component gives you presence, distance, gate energies and every tuning value as entities.
• Your own ESP32 or Arduino code: a module plus a test kit helps, so you can check the radar on a PC before you debug your sketch.
• No soldering: a finished sensor with a case, a controller and ESPHome already on it.

Phone tuning is the second fork. The LD2410B, B-P and C carry Bluetooth, and Hi-Link's HLKRadarTool app shows live gate energy and lets you set range, sensitivity and timeout from wherever you sit. The original LD2410 and the low-power S have no radio; you configure them over serial or through ESPHome.

Room count is the last question, and it is mostly about value. Multi-packs bring the cost per sensor down, and identical modules mean one YAML file and one tuning routine for every room. If you are unsure, buy one more module than you think you need; a spare on the bench makes tuning faster.

Two quick rules settle most doubts. If you already run Home Assistant, any 5V board on an ESP32 will do, and the C is the least fiddly. If the sensor must sit where there is no USB power, only the S makes sense. The version comparison covers every letter in detail.

• Type: module, test kit or ready-made sensor. A module is a bare board; a test kit adds a carrier and cable; a ready-made sensor is finished and flashed.
• Version: the letter after LD2410. It sets the board size, pin order and pitch, and whether Bluetooth is on board.
• Pack: how many modules come together, from singles to 20-packs.
• Bluetooth: yes means app tuning with HLKRadarTool and the option of Home Assistant's LD2410 BLE integration.
• Cable: a harness in the box saves soldering fine-pitch pads.

The version letter hides the detail that trips your first wiring job: pin order. The original LD2410 and the B count OUT, TX, RX, GND, VCC from pin 1, while the C starts with TX, then RX, OUT, GND, VCC.

Wire a C from a diagram drawn for the original and you put the supply on a data pin. Check the pinout reference for your exact board before you solder.

Pitch is the other practical difference. The original and B use 1.27mm pads on a 7x35mm strip, which is why they often ship with a cable or, as the B-P, with pins already fitted. The C uses standard 2.54mm spacing on a 16x22mm board and takes an ordinary header.

What stays the same across the 5V boards: 24–24.25GHz FMCW, a rated 5m reach, a ±60° field, 0.75m distance gates, about 80mA average draw, and the same serial protocol. Code and YAML written for one work on the others. The LD2410S is the exception on every one of those lines except the band, so treat it as a separate product.

A finished sensor is the shortcut. The AKAMATIS v2.2 pairs an LD2410B with a Seeed ESP32-C3 and an external Wi-Fi antenna in a case, powered over USB-C and pre-flashed with ESPHome. You join its hotspot, enter your Wi-Fi details and adopt it in Home Assistant. It needs Home Assistant already running; without it there is nothing to talk to.

The test kit suits the curious. It puts a Bluetooth B module on a carrier with a cable, so you can see detection in Hi-Link's PC tool or the phone app before you commit to a controller, a case or a line of code.

• Never soldered: finished sensor, or C modules with cables.
• Learning the radar: test kit first, then a bare module.
• Battery power: LD2410S on 3.3V.
• Several rooms, ESPHome: a C multi-pack.

The S is the specialist. Hi-Link rates it at around 0.1mA average when it reports once a second, against roughly 80mA for the 5V boards. That turns a sensor that needs a USB cable into one that can live on batteries for months. The trade is compatibility: it runs at 115200 baud with its own protocol, has no Bluetooth, and is not covered by ESPHome's built-in LD2410 component.

The C with cables is the easy everyday build. Three modules, jumper cables in the box and standard pitch mean an ESP32 on a breadboard and a working sensor within the hour. Power from the ESP32's 5V pin, sensor TX to GPIO16, sensor RX to GPIO17, and the ESPHome YAML takes it from there.

Whichever you pick, tune it in place. Thresholds set on a bench rarely suit the final wall, the furniture and the ceiling fan.

Bare modules are just the radar board. You supply power, a controller and firmware. They cover every version: the original, B, B-P, C, S, D and AA. The same Hi-Link design appears under many brand names, so the version letter tells you far more than the brand on the bag.

• Modules: cheapest per sensor, most flexible, need wiring and firmware.
• Test kits: a module on a carrier with a cable; for PC and app testing, or for code development.
• Ready-made sensors: radar, controller, case and ESPHome firmware in one; plug in and adopt.

Two newer versions widen the family. The LD2410D targets stationary presence for repeat installs, with a 4m-radius cone from a 3m ceiling and 6m on a wall. The LD2410-AA is rated to 9m for moving people and ceilings up to 6m, and comes as a test kit. Both have their own documents, so check pinout and protocol rather than assuming LD2410 compatibility.

Pack size is where cost per room drops. A single module is right for a first experiment. A 3- or 4-pack suits a flat or a small house. The 10- and 20-packs are for installers or for anyone who wants a sensor in every room, hallway and cupboard.

Bare boards also need a home. Radar sees through plastic, so a printed box works well, provided the front wall is thin and sits about a wavelength from the antenna. For case design around a bare board, see the enclosure guide.

1. Version letter. C for most builds, B for slim mounts, S for batteries. An unmarked HLK-LD2410 is usually the original with no Bluetooth.
2. Pin pitch and header. 2.54mm on the C; 1.27mm pads on the original and B unless pins or a cable are included.
3. Bluetooth. Needed for the HLKRadarTool app and the LD2410 BLE integration.
4. Pack size. Count rooms, then add a spare for bench testing.
5. Power. The 5V boards need a supply that can deliver over 200mA; the S needs 3.0–3.6V.

LD2410 price is low for what the radar does, and the per-module cost falls further in multi-packs. Budget for the rest of the build too: an ESP32 or ESP32-C3 board, a USB supply, jumper wires and a printed or bought case. A finished sensor costs more than a bare module, but it includes all of that plus firmware.

• Planning a battery node? The S is the only version built for it.
• Running ESPHome? Any 5V board works with the built-in component.
• Writing your own code? Pick a board with Bluetooth anyway; phone tuning saves hours.

A note on names. Hi-Link LD2410, HLK-LD2410 and LD2410 are the same family, Hi-Link being the Shenzhen maker and HLK its part prefix. Brands on the packaging vary, but the radar board is Hi-Link's design.

If a spec looks odd, such as a range much larger than 5m on a standard C, trust Hi-Link's documents. The datasheet walkthrough lists the official figures for frequency, power, range and resolution.

1. Wire it. Four wires to an ESP32: 5V, GND, and the crossed serial pair. The board-by-board wiring guide covers ESP32, ESP32-C3, D1 Mini, ESP8266 and Arduino.
2. Flash ESPHome. Add a UART at 256000 baud and the ld2410 component; the ESPHome guide has YAML to paste.
3. Adopt it. Home Assistant finds the node and creates presence, distance and tuning entities.
4. Tune it. Turn on engineering mode, read the empty-room energy per gate, and set thresholds above it.
5. House it. Print or buy a case that keeps plastic at the right distance from the antenna.

Most first-day trouble comes from three places. Crossed TX and RX lines give silence. Powering the module from 3.3V instead of 5V gives resets or nothing. A fan, curtain or radiator in one gate gives false presence, which the range and gates guide shows how to tune out.

• No data at all: swap TX and RX, check 256000 baud.
• Random resets: use the 5V pin and a 1A supply.
• Presence that never clears: find the busy gate in engineering mode and raise its threshold.
• Light turns off while you sit: lower the still threshold of your chair's gate.

Bluetooth boards give you a shortcut. Tune the gates in the HLKRadarTool app while standing in the room, then wire the module up; ESPHome reads the stored values back. And if you find you need to know where people are, not just that they are there, the LD2410 vs LD2450 comparison explains when a tracking radar earns its place.

Keep a notebook of your final thresholds per room. When you add the next sensor, you start from numbers that already work.