Code for letting an ESP32 read sml data from a smart meter via its IR interface.
I wrote this little hack because I didn't want to wire up my raspberry down the cellar and to the smartmeter - besides, there was a smartmeter but no power supply, so I had to use a powerbank which the raspi would drain within short time.
Thus now an ESP32 is reading the data and writes it to an InfluxDB (Cloud) and to a Volkszaehler instance (raspberry at home).
This code is released under the MIT license. See license.txt.
All configuration is done in config.h. Define user and password for wlan here, as well as target databases, baud rate of your sensor etc.
The sensor is a simple IR photo transistor (TEKT5400S-ASZ or similar) that's hooked up to the ESP32, using an 1K pullup and an 10K pulldown resistor to give a decent signal. An image showing the circuit is added to the projekt as "wiring.jpg". (The sensor is based on a simple IR reader like described at https://wiki.volkszaehler.org/howto/simpler_ir_leser .)
Basically, it's a 1:10 voltage divider, keeping the level high (1) by default. Once the IR photo transistor gets triggered, it shortcuts the pulldown resistor and thus pulls the level to low (0).
The data is read using one of the three hardware serial interfaces the ESP32 has. Initially I tried an ESP8266 with software serial interface on some freely configurable GPIO, but I experienced too much data loss on that, so I decided to go for ESP32 and hardware serial, even if the ESP8266 has the better sleep modes.
I've got to power my ESP32 with a power bank. To save some power, a) it's going to deep sleep for 57s and then starts reading from the smartmeter again, and b) it powers the sensor over GPIO4 so the sensor is switched off during deep sleep.
You may define which fields/values you want to read from the smartmeter. I'm reading
- 1.8.0 : energy in kWh from grid so far
- 2.8.0 : energy in kWh to grid so far, from PV unit
- 16.7.0 : current power in W passing the smartmeter You may want to read others. Find a list of OBIS identifiers at https://de.wikipedia.org/wiki/OBIS-Kennzahlen .
The first version supports three backends as target:
- Influx tineseries DB, e.g. at corlysis.com or at own site. You may use Grafana to visualize the data later.
- Volkszähler.org instance, presumably on a raspberry at home. Find more info at https://www.volkszaehler.org/
- Plain web server, in my case just to have the values in the web servers error log, for debugging purposes. Of course you may use this with an own PHP or Python script to write the data to a database.