Sell predictive maintenance to small factories with £30 vibration sensors
On the practice data in this mission, an ESP32 sensor costing under £40 flags a failing bearing more than a day before a simple alarm does. Bakery mixers, print-shop compressors and grain dryer fans fail the same way as factory machines, and nobody watches them at all. Build the sensor, the server and the anomaly model, and sell it as a setup-plus-monthly service.
- A practice dataset with a simulated bearing fault, so you can see the whole pipeline work before buying parts
- A business kit: parts list, ROI calculator, install checklist, offer, scripts and a monthly note template
- Per sensor: an ESP32 dev board (around £8 to £12), an ADXL345 breakout (around £4 to £8), a magnet, a small box and a USB power supply
- A server on site: any always-on PC or a Raspberry Pi 5 (around £70 to £100 for a kit)
- Python 3.12 or newer, and the Arduino IDE 2 for flashing
- Comfort with a soldering iron or jumper wires, and the terminal: Terminal survival kit
The business: early warning for the machine that stops everything
Sell a phone alert days before a breakdown, not a sensor.
Every small factory has one machine that stops the whole placeThe bakery's spiral mixer. The print shop's compressor. The joinery's dust extractor. The farm's grain dryer fan in harvest week. When its bearing fails, production stops, the engineer comes out as an emergency, parts are rushed in, and orders go late.
Bearings, imbalance, looseness and misalignment almost always change the vibration before they fail. Large plants pay specialists to measure it. Small businesses don't, because the kit and the analysts cost more than the machine. Cheap sensors and a model that learns each machine's own normal change that maths.
What you sellsensors fitted to the critical machines, a small server on site, phone alerts when a machine's pattern changes, and a monthly one page health note. You're not the repair engineer. You tell the owner early, and their usual engineer fixes it at a planned time.
Careful- What it takes. Capital: about £25 to £40 per sensor and up to £100 for a site server, usually covered by the setup fee.
- Skills: basic electronics (jumper wires, flashing an ESP32), Python, and reading what a model tells you. You don't need to be a vibration analyst: you watch for change and send the owner's engineer to look.
- Hours: half a day to install a site, then 1 to 2 hours a month per site.
- Insurance: public liability for working in other people's premises and professional indemnity for the advice.
- Time to first customer: commonly a few weeks, through a free pilot on one machine.
- Who this suits: someone technical who likes workshops and machines, enjoys building hardware, and wants a recurring-fee business with customers who stay for years.
Everything is in the kit: Download the kit. The firmware, the feature code, the practice data generator, the detector, the live monitor, a Mosquitto config, the parts list, the ROI calculator, the install checklist and the offer pack.