Harnessing Sunlight: How a Tiny Panel Woke Up the Cortex-M — Part 2

February 20, 2024

Blog

This three-part series details a case study on the use of miniature solar panels in powering IoT Devices. Check out the bottom of the article for Part 1, and stay tuned for Part 3, coming soon.

Experiment Preparation: Simulating Real Conditions

• Outdoor Testing: We placed both the Voltaic panels and the Epishine module outdoors where they could receive the maximum amount of sunlight. We precisely measured the voltage and current generated by both devices under various lighting conditions.
• Indoor Testing: We also placed the Voltaic panels and the Epishine module in a controlled indoor environment with artificial lighting. Light meters maintained a consistent level of illumination, simulating indoor lighting conditions. We monitored the energy output of both devices under various lighting levels, providing data on panel performance in conditions close to real indoor usage.

Analysis of PV Panel Performance

Now, let's delve deeper into the results we've obtained in a study of photovoltaic panels, assessing their efficiency under various lighting conditions for powering the NRF52832 and ATM3202 development kits.

NRF52832 Energy Consumption

This DK exhibits varying power consumption depending on its operational state, which plays a key role in evaluating its compatibility with photovoltaic panels and modules.

• Startup Current: During the experiment, an oscilloscope with a 10-ohm resistor in the circuit was used to measure the NRF52832's startup current. The measurements showed that the voltage across the resistor was about 700 mV, corresponding to a startup current of approximately 70 mA. This is crucial as the startup current determines how well a PV panel can power up the device.

• Operating States: Like its counterparts, the NRF52832 has three main operating states - idle, advertising, and connecting. In idle mode, the device consumes only a few nanoamperes, making it exceptionally energy-efficient in this state. In advertising mode, consumption can vary from 1 to 10 mA, depending on the intensity and frequency of signal transmission. In the connecting state, current consumption depends on the specific application and can vary significantly. At this point, we were very hopeful that the capacitors we chose for the experiment would help ensure an uninterrupted power supply to the device.
• Real Scenarios: With all onboard LEDs of the NRF52832 turned on, the power consumption is about 13 mA. During startup, which lasts about 50 ms, current consumption increases to 70 mA. These data helped us understand the conditions that solar panels must meet to effectively power the device in its various operating modes.

ATM3202 Energy Consumption

For our study, the ATM3202 DK was programmed with two example applications: a BLE advertising and a hibernation mode. Initial tests showed distinct power usage patterns in both modes. Despite our focus on testing the previously described PV panels, the Atmosic ATM3202 Development Kit includes its own integrated energy harvesting module with a PV cell, which we also evaluated for completeness.

• Power Profiling: The ATM3202 DK's power consumption was tested under BLE advertising and hibernation modes. Using a 220 Ω shunt resistor, we observed the voltage drop, so it was difficult to estimate power usage.

For a more detailed analysis, the Nordic Power Profiler Kit was used, showing that the peak consumption was around 1.3 mA during BLE advertising starts.

• Operating States and Server-Client Interactions: Further testing involved BLE server-client interactions, focusing on connection, notification, and disconnection phases.

Our observations, backed by power profiling, showed that the device maintained low and efficient power usage across these states. For example, during BLE advertising, the average power consumption was around 18 μA, with a peak of about 70 μA, which seemed pretty impressive

• Real-World Application: The PV cell that comes along with the ATM3202 Evaluation & Development Kit and is tailored for indoor use, was capable of charging the onboard capacitor in about 10 seconds. The board started advertising within 30 seconds after the energy harvesting initiation.

During our tests with the ATM3202 board, we noticed it sometimes lost its connection with the mobile app. This could be due to changes in the light indoors, which can affect how much power the board gets. Also, BLE technology can react to things around it, and the board's software configuration might add to this issue. We found that when we increased the light inside, these problems went away. This suggests that for the ATM PV cell, the right amount of light is important for keeping the board connected.

Voltaic Panels Performance

• Voltaic P121 R1L: In outdoor experiments under sunny conditions, the P121 R1L panel, with a maximum power of 0.3 W and a voltage of 5.9 V, showed impressive results. It generated a current of approximately 120 mA, significantly exceeding the NRF52832's startup current of 70 mA.

The panel also maintained a sufficient energy output level during indoor testing with artificial lighting.

For the ATM3202 DK, the P121 R1L's output of 120 mA in sunny conditions is more than adequate, considering the ATM3202's lower startup current. The panel's performance under artificial indoor lighting also effectively powers the ATM3202 in a variety of light conditions.

• Voltaic P122 R1J: Outdoors, with a maximum power of 0.32 W and a voltage of 2.3 V, the P122 R1J panel generated about 43 mA of current, insufficient to start the NRF52832.

However, when used with the Voltaic C116 lithium-ion capacitor, the panel showed increased efficiency, with enough peak power to overcome the device's startup and then continuous power to the NRF52832, even in non-ideal conditions.

Given the ATM3202's lower power consumption, the P122 R1J's output of 43 mA is sufficient for its operation, even more so with the aid of the C116 capacitor. This setup could provide a stable power source for the ATM3202, accommodating its energy requirements in different environmental conditions.

Additional Information on the C116 Solar Charger

The Voltaic C116 capacitor, with a capacity of 250 F and a maximum operating voltage of 3.8 V, played a key role in the experiments. It stored energy up to 1805 J (approximately 0.5 watt-hours), enabling the NRF52832 to start up in unstable or insufficient sunlight conditions. With the P122's maximum power output of 0.3 W, the capacitor could be fully charged in 100 minutes under clear weather.

With the ATM3202's efficient power management, the energy stored in the C116 could allow for longer operation times or more reliable performance in varying light conditions, making it a suitable pairing for the ATM3202 as well.

Epishine Module Performance

A feature of the module is its ability to store electrical energy ranging from 1.9 Ws to 3.4 Ws depending on the output voltage, as confirmed in our study.

Even without a solar output, and considering the energy required for the NRF52832 startup (0.07 A x 3.3 V x 0.05 s = 0.01155 J), it was determined that approximately 1.88845 Ws of energy remained in the supercapacitor after startup (1.9 Ws - 0.01155 Ws). For the ATM3202 DK, considering its lower startup energy (0.05 A x 3.3 V x 0.05 s = 0.00825 J), approximately 1.89175 Ws would remain in the supercapacitor after startup (1.9 Ws - 0.00825 Ws).

With the NRF52832 consuming 0.0429 W (0.013 A x 3.3 V), its estimated operating time at 3.3 V was about 35 seconds (1.5015 Ws of energy) when the supercapacitor was fully charged again.

Additionally, we investigated the module's performance at an output voltage of 1.8 V. In this setting, with the fully charged supercapacitor, the NRF52832 could operate for about 145 seconds under the peak load, and At 1.8 V, the ATM3202 could operate for around 574 seconds (about 9.5 minutes) under peak load, which is never needed in real usage cases.

It demonstrates the effectiveness of the Epishine module in various lighting conditions and, most importantly, confirms its ability to handle peak power demands during startup and connection of the IoT device based on the boards we chose for the experiment.

---

Editor's Note: For Part 1 of this three-part case study, visit Harnessing Sunlight: How a Tiny Panel Woke Up the Cortex-M — Part 1.