View on GitHub


Jay Slater

Getting back to work

I haven’t done much on this project for almost a year—that’s life, sometimes. I was planning on returning to it this summer anyway, but that return was somewhat delayed by some projects around the house, including (helpfully for this project) moving brewing operations from my friend’s kitchen to my basement.

The latest piece of news is that I’m planning on switching from the ESP8266 to the ESP32. The big upside to this is that the ESP32 has enough memory both to load the scale program and do SSL requests, unlike the ESP8266, which means that the ESP32 version will be able to log updates directly to Google Sheets without an intermediary. The circuit design is a little simpler, too—the reference circuit indicates fewer pullup resistors, because more of them are on the module (the ESP-32-WROOM or pin-compatible ESP-32S). In exchange, it’s a bit harder to solder—it’s about the same size, but has more pins with finer pitch.

An ESP32 development board and a few bare modules are on their way from China. Once they arrive (September or so?) I can start on the process of updating the software and ensuring my board design is correct. In the interim, I hope to use the ESP8266 boards I have, along with the one-cell test scale I used last summer, to start isolating the problems with the full-size scale, which could be one of several things:

  1. Bad crimps on the connectors on the load cell wires.
  2. Bad soldering on the pin headers on the board.
  3. Bad load cells in the full-size scale.
  4. Poorly matched load cells on the full-size scale, such that their output is invalid when they’re wired in parallel.
  5. Incorrect understanding of how to wire load cells in parallel. I’m using four-wire full-bridge sensors, and simply connecting the excitation and sense lines in parallel is supposed to be enough, as far as I can tell, but I could be wrong.

Updates to come as I make progress.

Written on August 4, 2019