Just finished my first full run of the custom tank controller

I ran into somthing similar when I upgraded my own setup last month. I was worried about the continuous load because I had several relays controlling the pumps and lights, similar to your aquarium setup. The components felt fine, but I started noticing small intermittent drops in the Home Assistant telemetry, especially when the pump cycled on. It turned out the power supply I was using was just a cheap phone charger and couldn't handle the sustained current draw of the ESP32 plus all the accessories simultaneously. I had to switch to a dedicated 5V regulated power brick, one with a higher amperage rating, to keep the voltage consistent. This solved the intermittent logging hiccups completley. Even when the simulated sun and the pH sensor were running their full cycles, the readings remained tight. It’s funny how cheap power supplies can be the biggest hidden weakness in a seemingly stable project.

For the water level monitoring, I also had to adjust the debounce time on my code. Even if the GPIO reading was correct, the quick transition when the water hit a new level woud cause a few spurious triggers unless I built in a small delay. It made the UI much cleaner once I fixed that part of the logic.

I actually ran into a headache not with the physical wiring, but with the software timing. Your stability sounds great, but my issue was the state machine logic when the water level sensor flipped. I was using a float switch setup, and if the state transition happened too quickly, Home Assistant would briefly register both states before settling, causing the wrong automation to trigger. I spent ages fiddling with the debounce time in the YAML. After trying various intervals, settling on 50 milliseconds gave me the smooth handoff I needed. It stopped those spurious triggers when the pump started cycling.

For the simulated sun trigger, I found that relying purely on a simple state change wasn't precise enough. The transition from dusk to dawn needed a buffer to look natural. I added a small delay in the ESPHome configuration that checks the sensor state multiple times before confirming the change. It makes the whole flow feel more robust, almost like a professional setup, instead of just a reactive piece of code. It’s all about making sure the sensor readings are settled before the automation decides what to do next.

I'm intrested in how much CPU power the ESP32 uses when it's juggling all those tasks. Since you've got the live pH reading and the simulated sun running, it's gotta be doing a lot of internal processing. I've seen setups get bogged down becuase they're trying to log data every single second. Are you running the sensors on an interval, or are you letting Home Assistant poll them constantly? Just curious how efficient the ESPHome internal scheduling is.