Engineer by heart and mind • Machine Learning, Embedded Devices, Mobile Robots • Voyage lover •
Born and raised in Poland 🇵🇱
Currently living in Barcelona.
I've developed a GPS-Free Solution for route tracking, which uses machine learning paired with data from motion sensors. This new approach isn't just about counting steps!
https://t.co/54OBVHM4vt
#EnergyEfficiency#Running#MachineLearning#FitnessTracker
BLE allows configuring events intervals - a key factor for optimizing energy use. By tuning these intervals, it’s possible to cut power consumption even further, which is crucial for a charging-free smartwatch.
#BLE#BluetoothLowEnergy#EnergyEfficiency#EmbeddedSystems
I tested the power consumption of BLE connection events on the STM32WB55. Each event draws only 3.29 mA and lasts about 3 ms—a great result for an ultra-low-power smartwatch design. (See attached image for details.)
#BluetoothLowEnergy#STM32#LowPower
TV in the 90s: you turn it on, you watch.
TV 2025:
- turn on, wait for it to load
- popup: TV wants to update, 1.5GB. No.
- scroll sideways, find prime video app or etc
- popup: now app wants to update, 500MB. No!!
- App launching... App loading…
- select account screen
- 🫠
After deeper reconfiguration of the MCU and peripherals before entering STOP mode, I brought it down to under 20 µA. Lesson learned: just enabling a sleep mode isn’t enough - extra tuning and clever hacks are essential for true low power.
#LowPowerDesign#EmbeddedSystems#CortexM
When I first tested STOP mode on the MCU (before adding any features), power consumption matched the documentation. But once I added peripherals and some functionality, it jumped to 370 µA.
#CortexM#STM32#LowPower#Firmware
STM32 also provides a clever setup: two integrated MCUs - one dedicated to BLE, the other for the application. This division helps cut power consumption even further.
The second step is picking the right chip. After research, I found that Nordic and STM32 offer the lowest-power BLE solutions. Nordic is slightly better on paper, but I chose STM32 as I have more experience with it.
#LowPower#STM32#Nordic
I tested a 10×2.7 mm ERM motor to measure its current draw at different PWM duty cycles. The results are shown in the attached graphic. For typical smartwatch use, this translates to ~0.35 mAh per day
#Haptics#ERM#EccentricRotatingMass#EmbeddedSystems
I continue work on my charging-free smartwatch. The next component to tackle is haptic feedback. Two main options: ERM or LRA. LRA uses less power but needs an extra driver, so I decided to go with ERM. #Haptics#LowPower#Smartwatch#ERM
I chose an ultra-low power Cortex-M3 80 MHz from STM. It offers several sleep modes to save energy when idle. In my real-life tests, it showed just 3.35 µA in STOP mode and 10.18 mA in RUN mode—excellent results for this project. #LowPower#EmbeddedSystems#CortexM#MCU#STM32
The processor is another big energy consumer. It must be efficient, yet still strong enough for smartwatch tasks. I realise that w/ good software architecture & optimisation, a powerful CPU isn’t needed. Even a Cortex-M <100 MHz can be enough. #Smartwatch#LowPower#CortexM#MCU
In my own tests with a 1.3" MIP display, I measured just 3.4 mA during refresh. Given how infrequently refresh is needed, this means it can consume up to 10,000× less energy than OLED!😮
#display#EnergyEfficiency#Hardware#MemoryInPixel
One of the most power-hungry parts of a smartwatch is the display. To save energy, I chose efficiency over colours. The options were Memory-in-Pixel (MIP) or e-paper - and I went with MIP for its better refresh rate.
#smartwatch#LowPower#Hardware#TechResearch
I’ve already worked out specific ideas for which technologies to use, which components to select, and how to design the software to minimize power use. The next step is to test these solutions in practice.
#TechDesign#HardwareDesign#SoftwareOptimization
The real challenge in creating a smartwatch that never needs charging isn’t adding an energy source - it’s drastically reducing power consumption. This requires the most effort and true innovation.
#smartwatch#EnergyEfficiency#LowPower
I explored three options for energy harvesting—Peltier cells (body heat), solar cells, and piezoelectric harvesters. The conclusion was clear: only solar cells can deliver enough power to keep a smartwatch running.
#energyHarvesting#smartwatch
To make a charging-free smartwatch possible, two things are essential: drastically lowering power consumption and adding an energy source.
#chargingFree#smartwatch