Added explanation on how to add new meters.

pull/5/head
weetmuts 2018-04-20 11:55:57 +02:00
rodzic bd5cd4a1de
commit e5b0cbb018
2 zmienionych plików z 78 dodań i 6 usunięć

Wyświetl plik

@ -0,0 +1,75 @@
To add a meter type, make a copy of meter_multical21.cc to
meter_yourmeter.cc and add your meter to meters.h (eg
X(YOURMETER_METER) then adjust main.cc and meters.cc to know about
this new meter.
In meter_yourmeter.cc, search and replace multical21 for yourmeter,
possibly adding a new meter type in meters.h if your meter is not a
water meter and inheriting from that new meter type instead.
Then assuming that the payload is encrypted using the standard AES
encryption, and the decrypted content follows the standard DifVif
format, then you only have to change processContent in
meter_yourmeter.cc. For Multical21 when the frame type is 0x78, it
means that the content contains both headers and data. Thus after
parseDV is called, you can access the extracted data from the map.
For example: extractDVdouble(&values, "0413", &offset, &total_water_consumption_);
will retrieve the DifVif data 0x04 0x13 which means: 32bit/integer
storing Volume in litres. (If there are multiple 0413 data fieds in
the content, the second found will be stored under the key 0413_2,
0413_3 etc. )
When the frame is 0x79 it means a compact frame in the C1 protocol
that leaves out the DifVif headers! You have to know the format to
decode. Since I have yet to get the crc calculations to work, this
means that the format for multical21 is currently hardcoded, as can be
seen in the source line: hex2bin("02FF2004134413", &format_bytes); and
the comments above it. (You can see 0413 in this string.)
(The proper solution is to understand the crc and create a small
database of crcs to headers encoded into wmbusmeters. Thus we can
decode the compact frames immediately. Or if the crc is unknown we
have to wait until the long frame arrives to gather the crc->header
data and use that info for future compact frames.)
There are also vendor allowed extension points, like 02FF20 above,
that stores vendor specific bits of data, that happen to describe the
status bits of the meter, DRY, BURST, etc. Fortunately, it is possible
to parse and extract the data, even though you might not yet know what
it is.
Then unfortunately, in Multical21 it seems like they also truncate the
format in a non-standard way. The value 4413, is a vendor specific
code point, that is 32 bits in the full frame, but 16 bits in the
compact frame, even though the DifVif header is the same. I suppose
that is typical vendor specific quirk that has to be found out simply
by trial and error. The quirk is handled in the source code
by passing a lambda to parseDV that will override the length calculation
for a specific dif/vif combo.
But if you are lucky, a new meter might be entirely compliant with the
DifVif headers and makes no or very little use of vendor
extensions. (There is a difference in the frame format between C1 and
T1 see the omnipower stub). But as soon as you have found the difvif
data, it should be straight forward.
If you pass --debug to the commandline, wmbusmeters will print the
parse results of the data package on stdout. You can see what DifVif
data there is and adjust your code to fit. If your meter is using C1
mode, then simply try the multical21 meter and --debug. It will print
the parsing of the DifVif data and tell you the DifVif pairs
(DifVifVife triplets, or DifDifeVifVife quadruplets (though
quadruplets are not yet implemented)) even if the data comes from a
different type of meter.
It is very helpful if you have problems to email me (oehrstroem@gmail.com)
the --debug output, or just simply the telegram=|.........|.......| lines from the
debug output. These contain the decrypted output from the meter and
are very useful to replay the messages. The +... at the end can be
used to replay with the same timing as the messages arrived when
logging. This can be useful for testing the higher levels of the software
stack where the data is actually displayed to the user or alarms triggered.
Check in simulation.txt and test.sh for how to use this feature.

Wyświetl plik

@ -126,15 +126,12 @@ created: either `/dev/im871a` or `/dev/amb8465`. These symlinks are
necessary if you want to pass "auto" to wmbusmeters instead of the
exact serial port /dev/ttyUSBx.
# Limitations
Two usb wmbus receivers are supported: IMST im871A and Amber Wireless AMB8465.
One supported meter: Multical21 (water meter)
And two meters that are work in progress: Multical302 (heat energy meter) and Omnipower (electricity meters).
# Source code
The source code is modular and it should be relatively straightforward to add more receivers and meters.
Read for example the text file: HowToAddaNewMeter.txt
# Good documents on the wireless mbus protocol:
http://www.m-bus.com/files/w4b21021.pdf