Home

Awesome

An HTTP interface to MODBUS smart meters

This project provides a http interface to smart meters with a MODBUS interface. Beside the EASTRON SDM series, the software also supports the Janitza B23 DIN-rail meters. The meters provide all measured values over an RS485 connection. The software reads the measurements and wraps them into a HTTP interface, making it very easy to integrate it into your home automation system. Both a REST-style API and a streaming API are available.

NOTE: gosdm is not longer actively developed and has been archived. The development of gosdm is continued at volkszaehler/mbmd under the name mbmd which stands for ModBus Measurement Daemon. mbmd adds support for additional meters and grid inverters.

Supported Devices

The meters have slightly different capabilities. The EASTRON SDM630 offers a lot of features, while the smaller devices only support basic features. This table gives you an overview (please note: check the manuals for yourself, I could be wrong):

MeterPhasesVoltageCurrentPowerPower FactorTotal ImportTotal ExportPer-phase Import/ExportLine/Neutral THD
SDM1201++++++--
SDM2201++++++--
SDM2201++++++--
SDM5303++++++--
SDM630 v13++++++++
SDM630 v23++++++++
Janitza B23-3123++++++--
DZG DVH40133++--++--
SBC ALE33++++++--

Please note that voltage, current, power and power factor are always reported for each connected phase.

Some of my test devices have been provided by B+G E-Tech - please consider to buy your meter from them!

Table of Contents

Installation

The installation consists of a hardware and a software part. Make sure you buy/fetch the following things before starting:

Hardware installation

SDM630 in my test setup

First, you should integrate the meter into your fuse box. Please ask a professional to do this for you - I don't want you to hurt yourself! Refer to the meter installation manual on how to do this. You need to set the MODBUS communication parameters to 9600 8N1. After this you need to connect a RS485 adaptor to the meter. This is how I did the wiring:

USB-SDM630 wiring

You can try to use a cheap USB-RS485 adaptor, or you can build your own isolated adaptor. I did my first experiments with a Digitus USB-RS485 adaptor which comes with a handy terminal block. I mounted the bias network directly on the terminal block:

bias network

Since then, I tested various adaptors:

I started to develop my own isolated adaptor. Please check this link for more information.

Software installation

Using the precompiled binaries

You can use the precompiled releases if you like. Just download the right binary for your platform and unzip.

Installing the software from source

You need a working Golang installation, the dep package management tool and Embed in order to compile your binary. Please install the Go compiler first. Then clone this repository:

git clone https://github.com/gonium/gosdm630.git

If you have make installed you can use the Makefile to install the tools:

$ cd gosdm630
$ make dep
Installing embed tool
Installing dep tool

You can then build the software using the Makefile:

$ make
Generating embedded assets
Generation complete in 109.907612ms
Building for host platform
Created binaries:
sdm630
sdm630
sdm630_logger
sdm630_monitor
sdm_detect

As you can see two sets of binaries are built:

If you want to build for all platforms you can use

$ make release

or, for a single platform like the Raspberry Pi binary, use

$ GOOS=linux GOARCH=arm GOARM=5 make build

Running

Now fire up the software:

$ ./bin/sdm630 --help
NAME:
   sdm - SDM modbus daemon

USAGE:
   sdm630 [global options] command [command options] [arguments...]

COMMANDS:
     help, h  Shows a list of commands or help for one command

GLOBAL OPTIONS:
   --serialadapter value, -s value     path to serial RTU device (default: "/dev/ttyUSB0")
   --comset value, -c value            which communication parameter set to use. Valid sets are
                                         1:  2400 baud, 8N1
                                         2:  9600 baud, 8N1
                                         3: 19200 baud, 8N1
                                         4:  2400 baud, 8E1
                                         5:  9600 baud, 8E1
                                         6: 19200 baud, 8E1
                                            (default: 2)
   --device_list value, -d value       MODBUS device type and ID to query, separated by comma.
                                           Valid types are:
                                           "SDM" for Eastron SDM meters
                                           "JANITZA" for Janitza B-Series DIN-Rail meters
                                           "DZG" for the DZG Metering GmbH DVH4013 DIN-Rail meter
                                           Example: -d JANITZA:1,SDM:22,DZG:23 (default: "SDM:1")
   --unique_id_format value, -f value  Unique ID format.
                                           Example: -f Instrument%d
                                           The %d is replaced by the device ID (default: "Instrument%d")
   --verbose, -v                       print verbose messages
   --url value, -u value               the URL the server should respond on (default: ":8080")
   --broker value, -b value            MQTT: The broker URI. ex: tcp://10.10.1.1:1883
   --topic value, -t value             MQTT: The topic name to/from which to publish/subscribe (optional) (default: "sdm630")
   --user value                        MQTT: The User (optional)
   --password value                    MQTT: The password (optional)
   --clientid value, -i value          MQTT: The ClientID (optional) (default: "sdm630")
   --rate value, -r value              MQTT: The maximum update rate (default 0, i.e. unlimited) (after a push we will ignore more data from same device andchannel for this time) (default: 0)
   --clean, -l                         MQTT: Set Clean Session (default false)
   --qos value, -q value               MQTT: The Quality of Service 0,1,2 (default 0) (default: 0)
   --help, -h                          show help

A typical invocation looks like this:

$ ./bin/sdm630 -s /dev/ttyUSB0 -d janitza:26,sdm:1
2017/01/25 16:34:26 Connecting to RTU via /dev/ttyUSB0
2017/01/25 16:34:26 Starting API at :8080

This call queries a Janitza B23 meter with ID 26 and an Eastron SDM meter at ID 1. It . If you use the -v commandline switch you can see modbus traffic and the current readings on the command line. At http://localhost:8080 you can see an embedded web page that updates itself with the latest values:

realtime view of incoming measurements

Installation on the Raspberry Pi

You simply copy the binary from the bin subdirectory to the RPi and start it. I usually put the binary into /usr/local/bin and rename it to sdm630. The following sytemd unit can be used to start the service (put this into /etc/systemd/system):

[Unit]
Description=SDM630 via HTTP API
After=syslog.target
[Service]
ExecStart=/usr/local/bin/sdm630 -s /dev/ttyAMA0
Restart=always
[Install]
WantedBy=multi-user.target

You might need to adjust the -s parameter depending on where your RS485 adapter is connected. Then, use

# systemctl start sdm630

to test your installation. If you're satisfied use

# systemctl enable sdm630

to start the service at boot time automatically.

WARNING: If you use an FTDI-based USB-RS485 adaptor you might see the Raspberry Pi becoming unreachable after a while. This is most likely not an issue with your RS485-USB adaptor or this software, but because of a bug in the Raspberry Pi kernel. As mentioned there, add the following parameter to your /boot/cmdline.txt:

dwc_otg.speed=1

This switches the internal dwc USB hub of the Raspberry Pi to USB1.1. While this reduces the available USB speed, the device now works reliably.

Detecting connected meters

MODBUS/RTU does not provide a mechanism to discover devices. There is no reliable way to detect all attached devices. The sdm_detect tool attempts to read the L1 voltage from all valid device IDs and reports which one replied correctly:

./bin/sdm_detect
2017/06/21 10:22:34 Starting bus scan
2017/06/21 10:22:35 Device 1: n/a
...
2017/07/27 16:16:39 Device 21: SDM type device found, L1 voltage: 234.86
2017/07/27 16:16:40 Device 22: n/a
2017/07/27 16:16:40 Device 23: n/a
2017/07/27 16:16:40 Device 24: n/a
2017/07/27 16:16:40 Device 25: n/a
2017/07/27 16:16:40 Device 26: Janitza type device found, L1 voltage: 235.10
...
2017/07/27 16:17:25 Device 247: n/a
2017/07/27 16:17:25 Found 2 active devices:
2017/07/27 16:17:25 * slave address 21: type SDM
2017/07/27 16:17:25 * slave address 26: type JANITZA
2017/07/27 16:17:25 WARNING: This lists only the devices that responded to a known L1 voltage request. Devices with different function code definitions might not be detected.

API

Rest API

GoSDM provides a convenient REST API. Supported endpoints are:

Both device APIs can also be called without the device id to return data for all connected devices.

The "GET /last/{ID}"-call simply returns the last measurements of the device with the Modbus ID {ID}:

$ curl localhost:8080/last/11
{
  "Timestamp": "2017-03-27T15:15:09.243729874+02:00",
  "Unix": 1490620509,
  "ModbusDeviceId": 11,
  "Power": {
    "L1": 0,
    "L2": -45.28234100341797,
    "L3": 0
  },
  "Voltage": {
    "L1": 233.1257781982422,
    "L2": 233.12904357910156,
    "L3": 0
  },
  "Current": {
    "L1": 0,
    "L2": 0.19502629339694977,
    "L3": 0
  },
  "Cosphi": {
    "L1": 1,
    "L2": -0.9995147585868835,
    "L3": 1
  },
  "Import": {
    "L1": 0.16599999368190765,
    "L2": 0.10999999940395355,
    "L3": 0.0010000000474974513
  },
  "TotalImport": 0.2770000100135803,
  "Export": {
    "L1": 0,
    "L2": 0.3019999861717224,
    "L3": 0
  },
  "TotalExport": 0.3019999861717224,
  "THD": {
    "VoltageNeutral": {
      "L1": 0,
      "L2": 0,
      "L3": 0
    },
    "AvgVoltageNeutral": 0
  }
}

The "GET /minuteavg"-call returns the average measurements over the last minute:

$ curl localhost:8080/minuteavg/11
{
  "Timestamp": "2017-03-27T15:19:06.470316939+02:00",
  "Unix": 1490620746,
  "ModbusDeviceId": 11,
  "Power": {
    "L1": 0,
    "L2": -45.333974165794174,
    "L3": 0
  },
  ...
}

Monitoring

The /status endpoint provides the following information:

$ curl http://localhost:8080/status
{
  "Starttime": "2017-01-25T16:35:50.839829945+01:00",
  "UptimeSeconds": 65587.177092186,
  "Goroutines": 11,
  "Memory": {
    "Alloc": 1568344,
    "HeapAlloc": 1568344
  },
  "Modbus": {
    "TotalModbusRequests": 1979122,
    "ModbusRequestRatePerMinute": 1810.5264666764785,
    "TotalModbusErrors": 738,
    "ModbusErrorRatePerMinute": 0.6751319688261972
  },
  "ConfiguredMeters": [
    {
      "Id": 26,
      "Type": "JANITZA",
      "Status": "available"
    }
  ]
}

This is a snapshot of a process running over night, along with the error statistics during that timeframe. The process queries continuously, the cabling is not a shielded, twisted wire but something that I had laying around. With proper cabling the error rate should be lower, though.

Streaming API

GoSDM supports both websockets and long polling to transfer status and meter updates to connected clients.

Data read from the smart meter can be observed by clients in realtime: as soon as a new value is available, you will be notified.

Websocket API

Websocket API is available on /ws. All connected clients receive status and meter updates for all connected meters without further subscription.

Long polling API

NOTE Usage of the long polling API is discouraged for performance reasons. The long polling is only supported with sdm630_http, not with the newer sdm630.

We're using HTTP Long Polling as described in RFC6202 for the data transfer. This essentially means that you can connect to an HTTP endpoint. The server will accept the connection and send you the new values as soon as they are available. Then, you either reconnect or use the same TCP connection for the next request. If you want to get all values, you can do the following:

$ while true; do curl --silent "http://localhost:8080/firehose?timeout=45&category=meterupdates" | jq; done

This requests the last values in a loop with curl and pipes the result through jq. Of course this also closes the connection after each reply, so this is rather costly. In production you can leave the connection intact and reuse it. A resulting reading looks like this:

{
  "events": [
    {
      "timestamp": 1490605909544,
      "category": "all",
      "data": {
        "DeviceId": 12,
        "Value": 0.054999999701976776,
        "IEC61850": "TotkWhExportPhsB",
        "Description": "L2 Export (kWh)",
        "ReadTimestamp": "2017-03-27T11:11:49.544236817+02:00"
      }
    }
  ]
}

Please note that the events structure is formatted by the long polling library we use. The data element contains the information just read from the MODBUS device. Events are emitted as soon as they are received over the serial connection.

In addition, you can also use the firehose to receive status updates:

$ while true; do curl --silent "http://localhost:8080/firehose?timeout=45&category=statusupdate" | jq; done

responds each second with the current status, e.g.

{
  "events": [
    {
      "timestamp": 1501163437772,
      "category": "statusupdate",
      "data": {
        "Starttime": "2017-07-27T10:21:04.790877012+02:00",
        "UptimeSeconds": 10.000907389,
        "Goroutines": 22,
        "Memory": {
          "Alloc": 3605376,
          "HeapAlloc": 3605376
        },
        "Modbus": {
          "TotalModbusRequests": 325,
          "ModbusRequestRatePerMinute": 1943.823619582965,
          "TotalModbusErrors": 0,
          "ModbusErrorRatePerMinute": 0
        },
        "ConfiguredMeters": [
          {
            "Id": 26,
            "Type": "JANITZA",
            "Status": "available"
          }
        ]
      }
    }
  ]
}

Stream Utilities

We provide a simple command line utility to monitor single devices. If you run

$ ./bin/sdm630_monitor -d sdm:23 -u localhost:8080

it will connect to the firehose and print power readings for device 23. Please note that this is all it does, the monitor can serve as a starting point for your own experiments.

If you want to log data in the highest possible resolution you can use the sdm630_logger command:

$ sdm630_logger record -s 120 -f log.db

This will connect to the sdm630 process on localhost and serialize all measurements into log.db. Received values will be cached for 120 seconds and then written in bulk. We use BoltDB for data storage in order to minimize runtime dependencies. You can use the inspect subcommand to get some information about the database:

$ ./bin/sdm630_logger inspect -f log.db
Found 529 records:
* First recorded on 2017-03-22 11:17:39.911271769 +0100 CET
* Last recorded on 2017-03-22 11:39:10.099236381 +0100 CET

If you want to export the dataset to TSV, you can use the export subcommand:

./bin/sdm630_logger export -t log.tsv -f log.db
2017/03/27 11:22:23 Exported 529 records.

The sdm630_logger tool is still under development and lacks certain features:

OpenHAB integration

Please note: The following integration guide was written for OpenHAB 1.8. We currently do not have an OpenHAB 2.x instructions, but would appreciate any contributions.

It is very easy to translate this into OpenHAB items. I run the SDM630 software on a Raspberry Pi with the IP 192.168.1.44. My items look like this:

Group Power_Chart
Number Power_L1 "Strombezug L1 [%.1f W]" <power> (Power, Power_Chart) { http="<[http://192.168.1.44:8080/last/1:60000:JS(SDM630GetL1Power.js)]" }

I'm using the http plugin to call the /last/1 endpoint every 60 seconds. Then, I feed the result into a JSON transform stored in SDM630GetL1Power.js. The contents of transform/SDM630GetL1Power.js looks like this:

JSON.parse(input).Power.L1;

Just repeat these lines for each measurement you want to track. Finally, my sitemap contains the following lines:

Chart item=Power_Chart period=D refresh=1800

This draws a chart of all items in the Power_Chart group.

How does it look like in OpenHAB?

I use OpenHAB 1.8 to record various measurements at home. In the classic ui, this is how one of the graphs looks like:

OpenHAB interface screenshot

Everything is in German, but the "Verlauf Strombezug" graph shows my power consumption for three phases. I have a SDM630 installed in my distribution cabinet. A serial connection links it to a Raspberry Pi (RPi). This is where this piece of software runs and exposes the measurements via a RESTful API. OpenHAB connects to it and stores the values, just as it does with other sensors in my home.