search

ai-republic / bms-to-inverter

Use, monitor and control any battery brand with any inverter! Application to read and monitor data from BMSes and write data to an inverter using any protocol (e.g. UART, RS485, ModBus or CAN)
https://github.com/ai-republic/bms-to-inverter
Other
32 stars 6 forks source link

Outback Power Skybox Inverter BMS Canbus #20

Closed carsonfee closed 2 months ago

carsonfee commented 3 months ago

outback has a limited list of batteries it supports. It has support for Simpliphi, Blue Ion, Discover, Fortress, Active BMS, Energycell PLR and XLC, Kilovault.

Skybox runs on an industrial beagleboard and has an RJ12 connector to connect to the BMS RJ45 Can port.

carsonfee commented 3 months ago

I have heard mumblings that Skybox uses Sunspec Modbus but I am not entirely sure. Outback tech support is abysmal.

Simpliphi's documentation says this:

It is also possible to communicate with OutBack devices using the Modbus protocol and SunSpec client software as described in OutBack AXS Port Owner’s Manual. The SunSpec Interface menu item enables or disables this type of data stream from the MATE3

ai-republic commented 3 months ago

Hi @carsonfee, so you are trying to connect the Skybox to some other (not supported) BMS using the CANBus right? I would need some CAN specification from Outback or any supported battery manufacturer to create a binding. Unless its speaks a more common specification like the PylonTech CAN specification.

Another way would be to 'listen in' on the communication of a running system (Skybox and supported BMS) and try to reverse engineer the protocol specification. For that you would need a CAN device hooked up on the CAN bus of the system, monitor the communication and protocol the current values (like SOC, current, voltage, etc.).

carsonfee commented 3 months ago

I have that 2 Channel Can hat for my RPi. Can I just plug in to the inverter and capture the traffic? The supported batteries are in my top post ^^ The batteries I have are sungold power. I think it uses the same bms as EG4 and jaikiper.

ai-republic commented 3 months ago

Yes, you can connect the 2CH CAN hat to the inverter and use candump to capture the traffic.

If you can provide CAN or RS485 or ModBus specifications for your inverter or supported batteries it would make life easier. Maybe inquire at the manufacturer (inverter or any supported battery manufacturer) with your exact inverter model and be persistent that you cannot use it without the CAN/RS485/ModBus communication specification.

carsonfee commented 3 months ago

My batteries use a PACE bms. Sungold power

ai-republic commented 3 months ago

Hi Carson, can you configure your PACE BMS to use the Pylontech CAN protocol? I've read somewhere that you can configure it to use it?

carsonfee commented 3 months ago

Yes I can. It supports a few different protocols, Pylon, Pace, Growatt and a couple of others. Trying to connect to the battery now, it is supposed to have a canbus b type of cable using pins 4 & 5 but so far no connection happening. The SPI says the can hat is initialized, going to try a loopback test to make sure it's working then see if I can find another pin out for the cable.

On Sun, Mar 17, 2024, 2:04 AM Torsten Oltmanns @.***> wrote:

Hi Carson, can you configure your PACE BMS to use the Pylontech CAN protocol? I've read somewhere that you can configure it to use it?

— Reply to this email directly, view it on GitHub https://github.com/ai-republic/bms-to-inverter/issues/20#issuecomment-2002337234, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAN4HFOO4MPIRWRBYOOQOPLYYU56VAVCNFSM6AAAAABEER66LOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMBSGMZTOMRTGQ . You are receiving this because you were mentioned.Message ID: @.***>

carsonfee commented 3 months ago

getting somewhere, can bus needed to be set to 500kbps image getting this error, so it looks like they are talking just not the same language. I have the config set to PYLON_CAN on the Pi and BMS Can bus protocol is set to PYLON on the battery would a candump help ?

carsonfee commented 3 months ago

candump looks more or less right I think? 35E shows LEOCH as the manufacturer ID. That looks like the frames have correct ids but data is wonky/different from PYLON ?

can0 351 [8] 00110100 00000010 11010000 00000111 10100000 00001111 10110000 00000001 '4.......' can0 355 [8] 00110011 00000000 01100100 00000000 00000000 00000000 00000000 00000000 '3.d.....' can0 356 [8] 11110010 00010100 10101000 00000010 11110000 00000000 00000000 00000000 '........' can0 35C [8] 11000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 '........' can0 35E [8] 01001100 01000101 01001111 01000011 01001000 00100000 00100000 00100000 'LEOCH ' can0 359 [8] 00000000 00000000 00000000 00000000 00000100 01010000 01001110 00000000 '.....PN.'

carsonfee commented 3 months ago

is the endianess right? 356 Should be Pack Voltage so 11110010 00010100 is F214 which looks like garbage (-3,564) but 14F2 = 5362 * .01 which would be 53.62 which seems about right

ai-republic commented 3 months ago

ok, cool - that's definitely the Pylon protocol! Could you do the candump in standard hex format next time please. Binary is a bit of a hassle reading and converting ;)

So for command: 0x351: Max voltage = 56.4V Charge limit = 200.0A Discharge limit = 400.0A Min voltage = 43.2V

0x355: SOC: 51% SOH: 100%

0x356: Pack voltage: 53.62V Pack current: 6.8A Avg Cell Temp: 24.0C etc.

So that looks all good!

I've added a fix for the SOC and SOH and more logging, so please do a Clean install with the Configurator. Could you also set the log-level set to debug and provide the log file, then I can see what the application is getting.

carsonfee commented 3 months ago

can0 35E [8] 4C 45 4F 43 48 20 20 20 'LEOCH ' can0 359 [8] 00 00 00 00 04 50 4E 00 '.....PN.' can0 351 [8] 34 02 D0 07 A0 0F B0 01 '4.......' can0 355 [8] 2A 00 64 00 00 00 00 00 '.d.....' can0 356 [8] 63 14 0F FF EB 00 00 00 'c.......' can0 35C [8] C0 00 00 00 00 00 00 00 '........' can0 35E [8] 4C 45 4F 43 48 20 20 20 'LEOCH ' can0 359 [8] 00 00 00 00 04 50 4E 00 '.....PN.' can0 351 [8] 34 02 D0 07 A0 0F B0 01 '4.......' can0 355 [8] 2A 00 64 00 00 00 00 00 '.d.....' can0 356 [8] 63 14 18 FF EB 00 00 00 'c.......' can0 35C [8] C0 00 00 00 00 00 00 00 '........' can0 35E [8] 4C 45 4F 43 48 20 20 20 'LEOCH ' can0 359 [8] 00 00 00 00 04 50 4E 00 '.....PN.' can0 351 [8] 34 02 D0 07 A0 0F B0 01 '4.......' can0 355 [8] 2A 00 64 00 00 00 00 00 '.d.....' can0 356 [8] 63 14 3D FF EB 00 00 00 'c.=.....' can0 35C [8] C0 00 00 00 00 00 00 00 '........' can0 35E [8] 4C 45 4F 43 48 20 20 20 'LEOCH ' can0 359 [8] 00 00 00 00 04 50 4E 00 '.....PN.' can0 351 [8] 34 02 D0 07 A0 0F B0 01 '4.......' can0 355 [8] 2A 00 64 00 00 00 00 00 '.d.....' can0 356 [8] 63 14 35 FF EB 00 00 00 'c.5.....' can0 35C [8] C0 00 00 00 00 00 00 00 '........'

ai-republic commented 3 months ago

Ok, so its the short Pylon protocol version.

What does the application do? It should read values now from the BMS.

Could you please update to the lastest version (Clean install in Configurator) and attach the log file (please with debug log-level set).

ai-republic commented 3 months ago

Hi, I wrote some tests with the candump data and it is read and processed correctly. I think the problem you might have lies in the communication between your webserver and the application's MQTT broker.

Could you update your application and send me the BMS-to-Inverter.log and the _BMS-to-InverterWebserver.log from the installations log folder and if possible a screenshot of your browsers debug console output.

Please start the webserver sepeartely after the application is logging some decent data.

First start the application from your installation folder (in terminal 1):

java -DconfigFile=config/config.properties -Dlog4j2.configurationFile=file:config/log4j2.xml -jar lib/bms-to-inverter-main-0.0.1-SNAPSHOT.jar

Then wait until it reads data from the BMS and after that start the webserver (in terminal 2):

java -jar lib/webserver-0.0.1-SNAPSHOT.jar --spring.config.location=file:config/config.properties
carsonfee commented 3 months ago

just running the application it looks like its reading garbage from the can bus.

2024-03-21 11:46:47.406 | INFO | Thread-3 | verter.BmsToInverter:190 | Sending to inverter NONE on can1... 2024-03-21 11:46:47.407 | DEBUG | Thread-3 | verter.core.Inverter:79 | Inverter received: Buffer (HEX): [null]

2024-03-21 11:46:48.144 | INFO | Thread-2 | verter.BmsToInverter:164 | Reading BMS #0 PYLON_CAN on can0... 2024-03-21 11:46:48.145 | INFO | Thread-2 | stoinverter.core.BMS:120 | ---------------------------------> Thread 43 2024-03-21 11:46:48.145 | DEBUG | Thread-2 | ocol.can.JavaCANPort:66 | CAN frame read... 2024-03-21 11:46:48.146 | DEBUG | Thread-2 | ocol.can.JavaCANPort:72 | CAN read frame Buffer (HEX): [0x56, 0x03, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x9E, 0x14, 0xF9, 0xFE, 0xF7, 0x00, 0x00, 0x00]

2024-03-21 11:46:48.147 | DEBUG | Thread-2 | PylonBmsCANProcessor:98 | Pack V Pack A Avg Temp 20.48 0.0 -2506.8 2024-03-21 11:46:48.147 | INFO | Thread-2 | verter.BmsToInverter:216 | BMS SOC V A CellMinV CellMaxV CellDiff

1 204.8 204.8 0.0 0.0(#0) 0.0(#0) 0.0

carsonfee commented 3 months ago

Pylon protocol is little endian, seems like you're reading big endian but the BMS-to-inverter.log the byte buffer it is displaying is entirely different from what candump outputs

carsonfee commented 3 months ago

BMS-to-Inverter.log

carsonfee commented 3 months ago

BMS-to-Inverter_Webserver.log

ai-republic commented 3 months ago

Hi, I just committed a fix. Could you please do a Clean install and update your installation. Let's see what we get :)

carsonfee commented 3 months ago

BMS-to-Inverter.log

New log. Looks somewhat closer but still not right

carsonfee commented 3 months ago

your change to little endian for the byte buffer looks good but the actual data that's in the buffer is out of order or maybe padded somehow? the first 2 bytes look correct e.g. CAN read frame Buffer (HEX): [0x55, 0x03, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x1E, 0x00, 0x64, 0x00, 0x00, 0x00, 0x00, 0x00]

candump reads right, the values in your buffer are the same but there's all those extra 0 bytes. 355 [8] 1E 00 64 00 00 00 00 00

carsonfee commented 3 months ago

CAN read frame Buffer (HEX): [0x55, 0x03, | 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, |0x1E, 0x00, 0x64, 0x00, 0x00, 0x00, 0x00, 0x00]

I think the 6 bytes after the ID is the DLC and other parts of the can arbitration field

ai-republic commented 3 months ago

yes, that is just the extended CAN frame 4 byte frame id, length and flags(3 bytes after length), then comes the 8 byte data. But good point, forgot to skip the next 4 bytes before reading the data. I fixed it now - this will give the correct values now! :)

carsonfee commented 3 months ago

That fixed it!

now to figure out if the skybox can actually give any useful output from a candump

ai-republic commented 2 months ago

closing since this issue has not been actively continued. Please reopen if issues arise.