A code for ESP32 controller and a set of web-tools to gather, analyse, monitor and view stat and graphs for Peacefair PZEM-004Tv30 PowerMeter.
- WebUI via self-hosted HTTP/WebSocket server
- Real-time gauges and power charts
- MQTT publishing metrics data in json format
- 3 level Tiered TimeSeries metrics collector in controller's memory pool
- data/metrics export via json
- compressed OTA updating via esp32-flashz lib
- espem could be build to run via DummyPZEM emulator for prototyping and firmware testing without connecting to a real PZEM device
| 
| 
| 
| 
|
To find EspEM in your WiFi network you can use mDNS Discovey app for Android or any other that can work with mDNS or Bonjour for iPhone.
Just find http enabled service in the app and open with a browser.
MQTT server publishing and connection could be configured via "Settings" - "MQTT".
It will publish pzem metrics to topic ~/pub/pzem/jmetrics on each update cycle. Where ~ is topic prefix, default is EmbUI/[DeviceID]/.
Message structure is a dictionary with PZEM metrics in an integer format in a way PZEM sends it via MODBUS.
[{"stale":false,"age":999,"U":2180,"I":503,"P":778,"W":560,"Pf":71,"freq":505}]
I.e.
U - denoted voltage in decivolts
I - current im mA
P - power in deciwatts
W - energy in Wh
Pg - power factor in percents
freq - is frequency in dHz
stale - denotes if data is stale, i.e. has not been updated recently
age - data age in ms
Controller will keep a history of previous data received from PZEM in it's memory in a tiered memory pool. It is commonly used for time series data where the longer the data age then less frequent is sampling rate of the data to keep. All data resides in volatale memory, so it will be lost on power cycle or reset.
By default there are 3 levels of TimeSeries in a pool
| Level | Samples count | Sampling interval | Time range |
|---|---|---|---|
| L1 | 900 | 1 sec | 15 min |
| L2 | 1000 | 15 sec | 250 min (abt 4 hrs) |
| L3 | 1000 | 300 sec | ~83 hrs |
Number of samples and interval could be adjusted per each level via "Espem setup" - "TimeSeries collector" configuration.
For levels with sampling interval more that 1 second an averaging function is used to calculate mean value between intervals. I.e. if sampling interval is set to 60 sec and you power level was about 5 watt for 55 seconds and became 100 watts for the last 5 sec when sampling was taken, then resulting power value would be slightly over 5 watts, but not 100 W.
1000 samples takes about 25 KiB of RAM memory (one sample takes 28 bytes), so plan you pool accordingly. If your board has SPI-RAM then you can have a huge pool worth monthes of data to be kept :)
TimeSeries Data could be exported in json format per each tier level
Tier 1 URL - http://espem/samples.json?tsid=1
Tier 2 URL - http://espem/samples.json?tsid=2
Tier 3 URL - http://espem/samples.json?tsid=3
An example of exported data:
[
{"t":1701876803000,"U":216.60,"I":0.51,"P":88,"W":7,"hz":49.7,"pF":0.79},
{"t":1701877103000,"U":225.00,"I":0.52,"P":90,"W":14,"hz":50.1,"pF":0.77}
]
where:
t - is a unix timestamp in milliseconds (prefered for js processing)
other keys are PZEM metrics in float format
An older ESPEM version 2 was based on 3rd party lib. It's code still available under 2.x branch. ESPEM Ver 3 switched to it's own library pzem-edl. I wrote this lib to overcome limitations of the classic olehs's and mandulaj's libs. Being versatile those libs provided only basic functions talking to PZEM's using Arduino's blocking IO via serial port. New lib uses event-driven approach and provides extendable design API for multiple PZEM communication over single port.
| Version 2.x | Version 3.0 | |
|---|---|---|
| ESP8266 | YES | NO |
| ESP32 | YES | YES |
| PZEM004v30 | YES | YES |
| old PZEMv30 | YES | NO |
| Time Series Charts | YES (Basic) | YES (extendable) |
| TS in PSRAM | NO | YES |
Under /www there is a set of php/sql scripts that could be hosted undel LAMP to gather and calculate stats over long-term periods. Little bit outdated but still usable. No need for any cloud services, spyware etc... just a raspberry/orangepi running web-server with sqlite/mysql DB. It's possible to collect data from any number of PZEM monitors and store it in the DB for a long-term stats or get a PowerChart sampled data from the espem itself.
An example of exernal daily stats dashboard
This code is available under the GPLv3 license