Multi BMS Design

[Uksa007]

Discussion on Multi BMS Design

I completed and implement the initial design, the data is now combined from the slaves as per the design.
I have made many improvements and made the code more robust.

Muli BMS support will allow 10 BMS/Batteries to be connected to ONE inverter, 1 Master and 9 slaves
Future plans is to allow a mix of JK-BMS and JBD/Overkill to be slaves, currently just JK-BMS in the Alpha stage.
Update: Hardware interface kit now available with two RJ45s with Termination disabled for Multi BMS use.

It uses a Master and Slave model similar to Pylon and other battery vendors.
One ESP32 connected to each BMS, one will be designated the Master, the rest Slaves.
The Master will collect all the data from the Slaves and combine the data and send to the inverter over CAN bus.

CAN bus design, note only the last device needs 120 ohm termination.
The CAN bus should always be ~60 ohms regardless of the number of devices.

[Uksa007]

Starting to put more advanced code together, need to make some hard decisions on how the Master combines the data from the slaves
Current thinking is below, thoughts?

359 OR the Alarms (this may have some negative impacts but seems the safest option) IMPLEMENTED
351 Use Master for BMS instructs inverter: Charge Volts(CVL), Min voltage IMPLEMENTED
Master added with slaves for Charge Amps(CCL), Discharge Amps (DCL)IMPLEMENTED in V2.26
355 Take Average of SOC, SOH IMPLEMENTED in V2.16*
356 Use Master Actual Pack Voltage, IMPLEMENTED
Add ActualAmps, IMPLEMENTED
Use highest Temp IMPLEMENTED in V2.16
35C Use Master for Request flag to Enable/Disable: Charge, Discharge, Force charge IMPLEMENTED
35E Use Master for Battery type(protocol) eg GOODWE, PYLON IMPLEMENTED

[denveronly]

Page 20 of the Deye manual suggests the inverter has some parallel ability with parallel CAN ports, if that is the case you just need normal Multi BMS support, as the inverter works it out from there.

I still wonder how it's going to be able to charge a large battery pack, and would seriously look at if it's feasible to actually charge the large battery from one PV inverter.

Thanks for the tip, ive checked that.
I guess i will need to Sync the inverters and they should have BMS info from the Battery Bank connected to the very first inverter

[denveronly]

Is it possible to try the Multi BMS version?
I will be installing second battery in my setup with second JK BMS.
I would be able to test it and provide any needed beta testing data

[Uksa007]

Hi,

At this stage the Alpha code is non functional.
I don't have a second set of batteries so I have been simulating as much as I can, but it's slow going.
To finish development and support it, I need to purchase a second set of batteries.
You can help by joining the Alpha testers tier on Patreon as that is where the Alpha code will be release.

Regards.

[denveronly]

i would like to post my work with 2 batteries maintained by JK BMS+ESP32, and Homeassistant dashboard
its been 2 days since second battery and inverter are in function, some homeassistant data can be abnormal on screenshot, because of multiple resets of HA, Solar assistant and ESP32 during setup work
I want to thank Uksa007 for his work, it looks very solid now and im comfortable that it is monitored by Homeassistant
BMS1 is connected to Inv_1 on 1phase in 1phase out
BMS2 is connected to Inv_2 on 2phase in 2+3phase out bridged
Inv_1 is maintainig charge sequence only.

I will be connecting them with parallel com cable, to see what communication can be with them.
Since it usual connection is 3 phase mode (3 inverters) or in 1 phase double inverter mode (to double 1 phase)
I dont know how it will work and will it work without error that there is no 3rd inverter like my setup is.
I asked Sunsynk and Deye centers if i can connect them my way and waiting for a reply, i dont want to burn anything

[torsteinelv]

it would be great to add also multi inverter support i have now two batteries (2bms) in one battery bank and 2 Deye inverters. JK1 in Deye1 , JK2 in Deye2, and they dont know if something is wrong with other tide. Deye1 doesnt know anything about JK2 eg.

With deye multi inverter i think its enough to just connect this to a single inverter if they are in paralell setup

[denveronly]

Maybe at the first start it is possible to make error transition between 2 BMSs-ESP32s
BMS2 error transition to inverter1 using ESP32_1 and ESP32_2
BMS1 error transition to inverter2 using ESP32_1 and ESP32_2

Im happy with 2 ESP32 because i have monitoring in Homeassistant,
So the code could be between two ESP32's GPIO pins using RS485 or CAN protocolor something like that

[Uksa007]

What you describe sounds very similar to what I'm doing, but designing it to scale for more than 2 BMS and data combined into one CAN data stream for ONE inverter (or multi inverters in Deye case)

[torsteinelv]

Is there any place where we can follow the work done?

[Uksa007]

I write a Monthly update report of my progress on various different activities that I'm working on for my Patreon members.

Consider becoming a Patreon member, the sooner we meet the funding goal, the sooner I can purchase a second set of batteries which will allow me to finish multi bms support and provide support for it.

[Uksa007]

I have been making some good progress, but it's slow going, as everything has to be simulated, so essentially write extra code to try to simulate real data.
So far I have:

• A working proof of concept hardware design that allows Master and Slaves to be connected on the same CAN bus.
• Alpha code which contains both master and slave code, user selectable master or slaves.
• Have working proof of concept salve code that sends data from slaves to master over CAN bus.
• Have working proof of concept Master code that:
  • Receives slaves data
  • Detects when one/multiple slave are present (this is important when averaging values)
  • Detects when one/multiple slave are no longer sending data (this is important when averaging values)
• Work continues on:
  • Slave code sending all of the required data in the correct format
  • Master code collecting all the data from slaves
  • Master code combining all the data from the slaves
  • Master code combining its locally attached bms data with slave data
  • Master sending combined data to inverter

[Uksa007]

Multi BMS Alpha POC code
What you have all been waiting for!
It is available for my Patreon Alpha tester members!

I have been working pretty hard on this for the last few months, it is very early Proof of Concept Multi BMS code.
This release is very early code, I have tested using simulated data as I don't have the hardware to fully test it.
Not all functionally has been completed but it should be functional.

At this stage it does the following:

• Combines the alarms from the slaves
• Adds the current from the slaves
  Note: All other data at this stage comes from the Master!

[torsteinelv]

Will it work with 3x 14s li ion?

[Uksa007]

Yes, It should work with any supported JK-BMS battery configuration.

The code configuration items may need some minor tweaking, as it is setup for 16s Lifep04.
But I'm happy to help with that.
There are a couple of people using it standalone with 14s li-ion

[Uksa007]

What make and model inverter are you using?

Have you tried the normal standalone version with your inverter, does it work well?

[Uksa007]

I completed and implement the initial design, the data is now combined from the slaves as per the design.
I have made many improvements and made the code more robust.
For anyone interested there is additional 40 hrs work coding and testing the below improvements.
Alpha V2.16 available for my Patreon Alpha testers members.

• V2.16 Clean up slave code
• V2.15 Clean up can rx code and master logs received slave data
• V2.14 Set slave Charging mode to Alarm if connected BMS has an alarm
• V2.13 Set slaves Charing status to Slave
• V2.12 Clean up slave code, stop sending slave CAN IDs 0x351 0x35C
• V2.11 Any alarm including slaves will now set 0A for Discharge and Change and turn off all flags.
• V2.10 Improve BMS Current now signed and standardised code
• V2.09 Temperature now finds max of slaves and master then sent to inverter.
• V2.08 Ignore CAN id's out of range
• V2.07 Improve slave detection time now 3s
• V2.06 Added startup delay 5s, slave data more robust
• V2.05 SOC and SOH averaged with all slaves and master then sent to inverter

[Uksa007]

Multi BMS Alpha BYD Battery Box LV CAN Support

New version V2.18:

I have added a new CAN protocol for BYD Battery BYD Battery Box LV CAN.

This should have better support for inverters that don't support Goodwe batteries

NOTE: This version will enable BYD battery protocol by default you can change it:

• CAN Protocol: 1 for GOODWE, 2 for PYLON, 3 For BYD Battery Box LV

protocol: "3"

• V2.18 Add CAN Protocol 3 for BYD Battery Box LV support, add Battery total capacity

• V2.17 Text senor for software version

[Uksa007]

Some improvements I'm working on:

• Charge and Discharge current multiplied by the numbered slave +1(master). Implemented in V2.22
  • This allows inverter to scale current when additional slave battery/BMS are added
  • eg 1 master = 100 A, 1 master and 1 slave = 200A ... etc
• Slaves to send Charging status to master
• Master to reduce charge/discharge current when:
  • Slave no longer connected Implemented in V2.22
  • Master or Slave in Alarm
  • Master or Slave has completed charging, in status wait. (Charge current only)

Thoughts?

[dfch]

I am aware of that. I think nevertheless this feature might be useful, as cells are degrading over time (and not necessarily equally fast across multiple banks). Furthermore, imagine when adding additional batteries (maybe 2 years later): these newer cells will most certainly have a different capacity from the existing ones (even when they have the same rating).

[Uksa007]

If the capacities are different due to aging, it wouldn't affect their change and discharge current specification.
e.g. if it's a LF280 * 0.5C its 140A no matter of it's capacity/age.

All that would change would be how quickly it charges and discharge, due its reduced capacity compared with new cells.

In reality what happens is each bank charge/discharge currents will be slightly different, depending on a bunch of factors, eg SOC, ESR, age/capacity etc. It results in them keeping pretty close to each other in terms of SOC as they are governed by the common overall voltage of all the parallel packs.

[Off-Grid-Garage]

The current for each pack can be very different. With LiFePO4 chemistry, the voltage of one pack will only ever rise at the end of the charge curve, not before that. So, even if the capacity is the same or very different, you can see some dramatic differences in charge current when one pack is almost saturated but another is not an takes almost all the current. I have 304Ah, 280Ah and 280Ah in parallel and one of the 280Ah packs will take 80A alone when charging with say 100A. The other two are are basically idling until this one pack catches up.

I also have 280Ah batteries in parallel with 100Ah batteries and this works great. There are no huge currents between the packs and all packs charging/discharging analogue to their capacities. The 280Ah pack takes/delivers 3/4 of the current if in parallel with a 100Ah pack for example.

[alanv72]

Is there a reason you can't send each individual bms pack info on can? The pylon can protocol seems to have a provision for packnumber.

I am a little confused why you are summarizing anything on the master.

[Uksa007]

Is there a reason you can't send each individual bms pack info on can? The pylon can protocol seems to have a provision for packnumber.

The Pylon CAN protocol has the provision to send the number(max) of modules, not individual modules .

[nik-bok]

Hi, great work. I would like to order 3pcs kit for 3 JK BMSes ( 1 master + 2 slaves), how do I proceed?

[Uksa007]

Hi,
I will have to create a link for you. The current kits it just 1 Master and 1 slave.
Do you want then all to be all the new boards with the DCDC converter?

[nik-bok]

Yes please, that would be more elegant solution.

[chaosnature]

Hi, great work. I would like to order 2pcs kit for 2 JK BMSes ( 1 master + 1 slaves),
how do I proceed?

regards

[tatarinovD]

Please help me find a solution.
I have 4pcs. 200Ah batteries connected in parallel. The batteries support rs458 stacking and have a CAN output towards the inverter, but my DEYE inverter is not supported by these batteries.
Can I, with your help, make a gateway between the CAN ports of my batteries and the inverter?

[Uksa007]

Hi,

There doesn't seem to be much information on this battery, do you have any information on what CAN protocols it supports?

Regards

[tatarinovD]

I have instructions for it, I also requested technical documentation from the manufacturer - but I have little hope for this. They usually don't give anything. They also promised me a photo of the BMS board in a few weeks!! maybe this will help us somehow
Li-Tower_Lithium Battery Power Bank Product Manual_20220824.pdf
The protocol of this BMS is supported only by inverters from the SAKO plant

[Uksa007]

The battery manual suggests the battery communicates using the RS485 Lib protocol, Lib has been around for a while.
You could build a cable that connects the battery RS485 A/B pins to the Inverters RS485 A/B pins, then try all the RS485 protocols the inverter supports, you might get lucky.

[tatarinovD]

I'll try - thanks. I'll tell you about the results of this attempt

[tatarinovD]

Hello, I checked the compatibility of the inverter and battery in RS485 mode.
The best I could see was this incorrect data:

[Uksa007]

Probably the most cost effective way to get it communicating would be to replace the BMS with a JK-BMS.

[tatarinovD]

I have two battery packs, one has 5 batteries and the other has 4 batteries, that's 9 pcs. JK-BMS 120USD pcs. and completely disassemble and remake all batteries. In addition, new BMS cannot be installed in standard installation locations, which means that the housings will need to be redesigned or replaced!!!! This option is not suitable :(

I could connect a laptop with a modbass master and give you access to it to view what is coming from their BMS batteries. Maybe it's not so bad there? and the solution may be using your equipment.

I can also forward a USB port with a MODBUS master connected to it for greater convenience

[Uksa007]

My hourly rate is $150USD for bespoke solutions, probably a week or two work to reverse engineer the protocol and write a converter to CAN, another week to design hardware. It adds up fast.

[tatarinovD]

In this case, you are right, it is cheaper to replace the BMS :( it’s a pity that there is no simpler solution!!! thank you for your answers