I finally got around to uploading some videos on Youtube of the NoiseCard doing its thing. I'm pretty satisfied with its performance overall, but long-term a second iteration of the board is certainly needed.

Here's video no. 1 below! In the sun, the card works great. The LEDs are more noticable in person too, I promise.

As I continued with testing, it became apparent that the solar cell needs pretty direct sunlight to produce adequate power for the microcontroller. This comes down to two reasons: the cell feeds directly to a 1.8V switching regulator, so a voltage greater than 1.8V must be continuously present; second, the large capacitors are still only 660uF total, meaning maybe a couple of seconds of runtime once the sun goes away -- if they're fully charged.

It's sunny this time of year though, so more testing:

Measuring at least 80dB on a popular city street (Street No. 2, near Bloor/Yonge in Toronto) is concerning. The microphone's accuracy has checked out too -- during development I made sure performance was consistent with my other noise-monitoring project; both of these were built off of ESP32-I2S-SLM's reliable algorithms.

Even louder than that street though is the city's subway. For this test, I substituted a small coin cell battery in place of the solar cell: a CR1225. Taped to the card, it doesn't protrude much further than the cell. The wires that contacted the cell were a bit loose however, so the LED indication is a bit sporadic:

Overall, the NoiseCard has a lot of potential. To remedy its power supply issues, I'm looking into better energy storage (in the form of SMD supercapacitors) and an energy harvesting circuit to get as much power out of the solar cell as possible. The circuit would boost the cell's voltage over 1.8V so the system can run and charge in lower light conditions. Stay tuned for an update on that...