I like the idea of the iSpindel, but I wanted something smaller; ideally, something which I could use in my glass carboys. This contraint means I need to get it into a tube with an outside diameter of approximately 25mm.

I've also been trying to get away from Li-ion as a power source and stick to simple NiMH rechargeables. Using ESP-NOW as a communications protocol makes a single NiMH cell with a boost converter a realistic option. However, it turns out that a 25mm tube containing a AA NiMH cell doesn't have the necessary buoyancy to actually float, much less tilt, so I've been forced to switch to an AAA cell. In a nod to practicality, I've also included the option to use a 10440 (AAA-size) Li-ion cell with a conventional LDO regulator, which seems far more stable at this point.

Stripping out all the bits of the iSpindel that I can't use or don't need, I'm left with an ESP-12 module, the MPU-6050 module, the power circuitry, assorted passives, and a single battery cell. Even the DS18B20 temperature sensor is dropped in favour of the MPU-6050. Programming the ESP is handled by either removing it (I mount mine on pin headers) or some sort of TBD pogo pin jig.

weeSpindel is running a stripped down firmware which transmits reports using an ESP-NOW broadcast rather than WiFi. ESP-NOW allows for low-overhead transmission, which reduces the wake-read-send-sleep cycle to approximately 100-120ms (we only bother with five samples from the MPU, which honestly might even be excessive). A ESP-NOW-to-MQTT (ESP32 with an W5500 ethernet module) gateway makes the sensor results available in Home Assistant, etc. Configurability is... minimal.

Note that we don't bother with something like a MOSFET to cut power to the MPU module; between it and the (on my board, at least) Torex XC6204 LDO it'll use maybe 5uA during sleep. We do remove the LED in the usual fashion.

The prototype is done on a PCB mill, and that naturally limits what sort of component footprints I can comfortably get away with. Future design using a manufactured PCB opens up a few possibilities.

Lose the MPU-6050 module and put the chip directly on the board.

The failure mode of the HT7733SA boost regulator when the voltage drops too low (under 1.2v with a AAA NiMH cell) is to go into a approx 200mA battery draining loop. This is bad. In extreme cases, it'll reverse-charge the NiMH cell. The fix for this would be to change to a boost regulator with under-voltage lockout (UVLO). Something like the TPS61097 should do the trick.

I'd love to get a USB controller onboard for programming. Probably the CH340E.