GODENERG

Atersa / Axpert / Master Power / Voltronics Inverter python library / interface / tool

The idea is to develop a tool that works better that the dubious WatchPower program that comes with the inverter and improves in practicality by been usable through the network.

The Watchpower is a badly coded piece of Java that works as a GUI tool and has no network functionality when used with USB or Serial connections.

Godenerg should be usable in a Raspberry Pi and could be left connected to a house LAN acting as a bridge to the inverter. For instance, right now I am using Godenerg provide the real-time data I need to display through HTTP into a Raspberry Pi in my living room that uses a OLED to display real-time stats.

Current features:

• Solid daemon that can be left unsupervised working continuously.

• Datalogging working.

• Stats plotting over HTTP working.

• Real-time data over HTTP working.

• Charger process implemented and working better than the inverter's original charger.

TODO:

• More sophisticated charging

• Web based, real-time visualization tool (like watchpower but web based)

• Web based settings setter trough forms (like watchpower but web based)

• Query interface for datalogger (besides the existing querystring interface).

Charging:

Besides offering realtime data like WatchPower, this software is intented to by-pass the big elephant in the room with this inverter, the charger:

1. The firmware in the inverter has a massive bug that the developers are not addressing. When the inverter is charging, if the available amperage to charge, drops below 1/5 of the max charging amperage set in the settings, it stops cold the charge process (in less than a minute). If you happen to get a couple of clouds and have a rice cooker on, for instance, you can forget about getting a full charge. The inverter will stay in float voltage and the day's charge will be over.

2. Because of the mentioned bug you have to lower the max charging amperage if you want to minimize the issue and if you happen to have quite a bit of PV power like I have (3.8KW), you are not using the available energy.

3. Cloudy days, instead of charging with whatever energy there is (no worries for overcharging there), the inverter sits there at float voltage.

4. They are some patched firmwares that try to address this problems, but the are patched firmwares by individuals and no source code is released. I could give them a try but I fear bricking the inverter since my only means of energy is the PV array.

5. Latest releases of the official firmware just wait 10 minutes before dropping to float voltage instead of under a minute. I guess they tried boging a fix but is still not correct.

To implement a charger from Godenerg:

• I can dynamically set the float voltage setting.

• I can monitor the charge amperage and the battery voltage.

• I can gather stats and analyse the amperage and voltage overtime.

Therefore:

1. I can wait for the desired absorption voltage to be reached (bulk phase), forcing float voltage to be equals to absorption voltage.

2. Then I can monitor the average voltage and charge current through time as the absorption phase is happening.

3. When the amperage is low enough I can change the float voltage to the real float voltage.

With this mechanics I mimic the desired charging curve of volts vs amps described in all the charging documentation I find for sealed lead acid batteries.

(graph generated with godenerg: http://host:8890/graph?from=2017111807&to=2017111819&col_2=batt_charge_amps&col_1=batt_volt)

Current project status:

• I have created this project in a hurry and out of necessity, because the charging was starting to become very poor now that autumn has arrived and I had a deficit of energy. I needed to monitor and imporve charging.

• The daemon takes care of watching over himself and restarts gracefully if any issues happen. The daemon can be left unsupervised and will self fix and restart if any problems happen. I had to to this because the USB interface with the inverter is fairly unstable; the firmware sometimes fails to properly respond and other times has small hipccups so anything connected continuously to the inverter has to be able to handle failure.

• Python 3 compatible. Just tested on Linux. No hope to get it tested on Mac or Windows unless someone helps with that.

• So far tested on usb connections, whenever I get my hands on a usb to serial adapter I will test on serial connections.

• Project in its infancy, alpha at best but already usable.

• Charging is crude, is based in time of the day and not dynamic. A lot could be refined in the charging process.

• Right now command line tool is not up to date since I am putting all energies into the daemon.

Current inverter setup:

• Inverter is an 'Atersa 5KVA-48-PAR-A' model.

• Absorb voltage is set to max 58.4V

• Float voltage is set to 52.8V which is what is recommended on the batteries data-sheet.

• I have 12 x 320W PV panels wired in 4 paralel strings of 3 panels each in series.

• 8 x 6 550 Amp/h sealed maintenance free lead acid batteries.

• Connecting from debian linux to the inverter using the hidusb interface.

• Uncompressed the Watchpower JAR and decompiled the source to learn how to talk to the inverter, along side info found in a couple of forums.

Testing:

I will put more energies into testing during the next weeks. Since I needed this software in an urgent manner, I have not been the best testing for everything.

 $> pytest -v --pyargs axpert

Run as daemon

 $> python3 axpert/main.py --usb -d /dev/hidraw0 --daemon

• Datalogger starts up and starts logging QPIGS command into a sqlite database. The loggin interval is configured via the setup file found in the file axpert/settings.py inside the dictionary datalogger_conf:

    datalogger_conf = {
        'db_filename': '/home/ups/godenerg/godenerg.db',
        'interval': 15,
        'port': 8890
    }

  • Datalogger HTTP server for graphing datacharts (for the moment). The server starts in the port specified under the 'port' key in the datalogger_conf specified above.

  • Charting of one or two metrics are allowed.

  • If more of 2048 points are present in the chart averaging is done dinamicaly.

  • Date or datetime ranges are possible with the following querystring parameter formats, you can specify YYYYMMDD/YYYYMMDDHH/YYYYMMDDHHMM/YYYYMMDDHHMMSS:

    • From 2017-11-01 00:00 to 2017-11-04 00:00

    from=20171101&to=20171104 
    

    • From 2017-11-02 15:57 to 2017-11-02 19:00

    from=201711021557&to=2017110219
    

  • First desired column format: col_1=bat_volt

  • Second desired column format (optional): col_2=pv_watts

  • Example 1, graphing batt_volt for third of November:

    http://machine_ip:8890/graph?from=20171103&to=20171104&col_1=batt_volt

  • Example 2, graphing batt_volt vs pv_watts for the fourth of November to the eighteenth.

    http://machine_ip:8890/graph?from=20171104&to=20171119&col_1=batt_volt&col_2=pv_watts

  • Available values are:

    • grid_volt
    • grid_freq
    • ac_volt
    • ac_freq
    • ac_va
    • ac_watt
    • load_percent
    • bus_volt
    • batt_volt
    • batt_charge_amps
    • batt_capacity
    • temp
    • pv_amps
    • pv_volts
    • batt_volt_scc
    • batt_discharge_amps
    • pv_watts

• HTTP Server for JSON real-time data usage. Since the nature of the USB / serial communications is limited to a single client. Calls block until the serial / USB is free. So far just status (QPIGS) and (QMOD) commands are implemented. Adding other query commands or set commands will come soon since is just a matter of defining the specifications to a descriptive structure already defined.

  • Status(QPIGS) as JSON:

  http://machine_ip:8889/cmds?cmd=status
  

  • Operation Mode (QMOD) as JSON:

  http://machine_ip:8889/cmds?cmd=operation_mode
  

  • Both combined as JSON (indexed in 2 different keys)

  http://machine_ip:8889/cmds?cmd=operation_mode&cmd=status
  

  • Both combined as JSON in a single with all key/values merged

  http://machine_ip:8889/cmds?cmd=operation_mode&cmd=status&merge=1
  

Run as command line tool

Get current status values (QPIGS command)

> python3 axpert/main.py --usb -d /dev/hidraw0 --status

(000.0 00.0 230.0 50.0 0322 0221 006 425 52.80 011 100 0040 0016 100.6 52.78 00000 01110110 00 00 00844 010

Get current status values as JSON

> python3 axpert/main.py --usb -d /dev/hidraw0 --status --json

   {"ssc_firmware_updated": false, "grid_volt": 0.0, "raw_status": "01010000", "batt_volt": 49.7, "ac_volt": 229.9, "batt_capacity": 75, "configuration_changed": true, "pv_amps": 0, "bus_volt": 400.0, "ac_freq": 50.0, "ac_va": 298, "mask_d": 10, "batt_charge_amps": 0, "batt_volt_scc": 0.0, "load_status": true, "sbu_priority_version": false, "pv_volts": 0.0, "temp": 45, "load_percent": 6.0, "charge_source": ["not_charging"], "batt_discharge_amps": 5, "pv_watts": 0, "batt_volt_to_steady": false, "mask_c": 0, "grid_freq": 0.0, "mask_b": 0, "ac_watt": 241}

Get operation mode (QMOD command):

 > python3 axpert/main.py --usb -d /dev/hidraw0 --op-mode

Get operation mode as json:

 > python3 axpert/main.py --usb -d /dev/hidraw0 --op-mode --json

    {"mode": "BT"}

Change Float Voltage to 53.0 V

> python3 axpert/main.py --usb -d /dev/hidraw0 --cmd PBFT -v 53.0  -s 8
(ACK

Change Utility MAX charge current to 10 amps for first paralel device

> python3 axpert/main.py --usb -d /dev/hidraw0 --cmd MCHGC -v 010 -s 8 
(ACK

Extract data from datalogger

> python3 axpert/main.py --usb -d /dev/hidraw0 --extract-csv-data 20171031000000-20171031235959 --col datetime --col batt_volt --col batt_charge_amps --extract-file stats.csv