Once upon a time, there was a basement with below grade plumbing. Thanks to the wonderful technology of ejector pumps, this posed no immediate problems for the basement in question. One day a terrible thought occurred to the resident of said basement, "How can I be certain that the pump is working? How can I guarantee advanced knowledge of a potential shit flood?" The resident thought about this for a day or so, and finally determined that this was a job well-suited for an accelerometer attached to an Arduino board. And thus the shit pump monitor was born.

1. Connect to a wireless network
2. Begin and continuously monitor the LSM6DS3 sensor's gyroscopic data
3. When a threshold has been reached, make an HTTP call to a web endpoint
4. Healthcheck: Periodically call out to webserver, regardless of gyro data
5. Watchdog Timer: Reboots the device when it gets in an unresponsive state.

1. Purchase a Nano 33 IoT Arduino (Is this strictly necessary? No, but you'll probably need to customize this sketch to support a different board)
2. Create a file arduino_secrets.h in the same directory as the sketch file (See below section)
3. Make some adjustments as needed. (eg. The default gyroscopic threshold is 5, which may or may not be valid for your plumbing situation)
4. Upload the sketch to your arduino board
5. Snugly attach the arduino board to the egress plumbing of your ejector pump
6. Power the board
7. Extra credit: Setup the pump backend for continuous monitoring of your, er, monitor

#define SECRET_SSID "Network SSID"
#define SECRET_PASS "Network password"
#define WEBSERVER "domain.name.of.webserver"
#define AUTH_CODE "some-unique-identifier-for-the-backend-to-recognize"

Note: The final endpoint will be https://<WEBSERVER>/flush.php?authCode=<AUTHCODE> along with some other query params.

• The Nano will crap out after X HTTP requests. A watchdog timer has been put in place to reboot the device once it gets in this weird state.
• The Nano 33 IoT clock runs slow by ~2% (YMMV). It does not have a crystal-based clock, so I've added an adjustment to keep the healthcheck closer to 12 hours.