Build your own BMS

[s/v Jedi]

With all the boats building their LiFePO4 batteries, I think there are plenty with the skills to build their own BMS using simple Arduino modules.

So here’s the first draft of the diagram, based on my BMS which is in a state that can be called “experimental”. I have old threads with some hardware and software but at that time it was for a 12V battery, while I switched to 24V since then.

Please note that the components used can also work on 12V but not on 48V.

The cost of Arduino, INA226 modules, temperature sensors and OLED display is only a handful of dollars and available on Amazon. If you have a little experience with soldering and programming then this isn’t just the best way to go, it’s also fun to do.

I will be posting below with more details.

[wholybee]

I'm toying with this idea in my mind. Mostly a time issue right now. However, I wouldn't want to rely solely on a microprocessor. I would want a simple hardware based BMS that monitored the extremes of the voltage limits, and a microprocessor that was programable with user limits, alarm signaling etc. Still fairly simple, although some component level design is needed. Or, you could probably parallel a cheap $10 module with an Arduino.

It's a fairly simple project for someone with some experience, and the cost is quite low.

[s/v Jedi]

Originally Posted by wholybee

I'm toying with this idea in my mind. Mostly a time issue right now. However, I wouldn't want to rely solely on a microprocessor. I would want a simple hardware based BMS that monitored the extremes of the voltage limits, and a microprocessor that was programable with user limits, alarm signaling etc. Still fairly simple, although some component level design is needed. Or, you could probably parallel a cheap $10 module with an Arduino.

It's a fairly simple project for someone with some experience, and the cost is quite low.

-all- BMS’s rely on the same microcontroller setup. There exists no BMS without it.

[wholybee]

I don't believe that is true. The cheapest ones use simple transistor logic and resistors to control a FET on/off based on voltage inputs. I've come across dozens of schematics that show this. I can't say with absolute certainty that programmable BMS's do or don't maintain a simple hardware failsafe, but I expect they do. The overkill solar BMS does have "hardware" protection limits given in the manual but I've not looked at a schematic to confirm it. If it can be shown that a REC or similar doesn't, that makes the job easier, that's for sure.

[rgleason]

I am interested in this project and will be following it. Thanks for sharing.

[s/v Jedi]

Cell voltage measurements:

The main issue is that you normally need galvanic isolation to do these measurements, making it all complicated unless you use specialized chips. So the way it is done here is to measure between the battery negative and each cell’s positive. This means the first module measures 3.2V, the second 6.4V, then 9.6, 12.8, 16, 19.2, 22.4 and 25.6V. The INA226 can do these voltages easily and it’s up to the microcontroller to substract the previous cell’s voltage to get back to the 3.2V cell voltage. This is a very neat for-next loop in code.

These INA226 modules cost $5 each on Amazon and communicate over the I2C bus with the micro. We only use the Vin to measure voltage, ignoring the shunt and current & power measurements.

Each module needs a unique address on the I2C bus, for which solder pads are available. For 8 modules you need to do a little work; I believe 4 need a link cut, but the PCB is marked on how to do this.

For a 12V version you can also use the INA219 modules, which I just saw for $3 each on Amazon.

The OLED module is another I2C bus device and has standard libraries in the Arduino IDE available. Now you can design nice displays. The picture is from a couple years ago when I was designing the BMS

[T1 Terry]

What I'd like is cell top boards that only measured the voltage of that cell and had a programmable alarm signal for low cell voltage and a separate alarm for high cell voltage.
The issue arises sending this information back to the main control board. CANBUS is the idea method but that pushes beyond the simple and into the relm of those that better understand just how CANBUS works and be able to understand what the signal looks like if there is a problem with the CANBUS itself.
Every system owner would need to be able to read, interpret and diagnose any such problem and this is where the whole thing falls over, if every system owner knew how to do this, there would be no high paid professionals following the career path .... a bit like graphic designers and web designers and those that repaired the photo development machines that processed exposed rolls of film .... still highly skilled but a skill not worth much these days ....

The tricky part is building a BMS out of basic off the shelf components and build in levels of redundancy, so if one part starts giving trouble, there is work around till the faulty part can be replaced ... just getting the faulty part can be the real problem when remote, so simply replacing the whole unit by carrying a spare becomes complex and expensive.

We built our own BMS units some 12 yrs ago, 7 levels of redundancy, the last level the system operator becomes the BMS, the second last level is battery isolation if no one heard or responded to all the alarms and flashing lights. This protects at both high cell voltage and low cell voltage but can determine if the cell voltage alarm is the real thing or a false alarm.

Sadly. a number of the parts required to build this BMS are now becoming near impossible top source, I'm fine with microprocessors, just as long as there is another micro processor further down the line that can step in to protect the battery if an upstream microprocessor drops its bundle .....

T1 Terry

[Zanshin]

I'm also working on a BMS using the INA229 for my 28V system. But if all you want the INA229 for is voltage measurement there are cheaper and better chips on the market. I use the 229 to also measure power, which is what it is built for and excels at. No need to write a library for the Arduino, I've got one that works interchangeably for most INA2xx chips: Zanduino INA Library

[s/v Jedi]

Originally Posted by T1 Terry

What I'd like is cell top boards that only measured the voltage of that cell and had a programmable alarm signal for low cell voltage and a separate alarm for high cell voltage.
The issue arises sending this information back to the main control board. CANBUS is the idea method but that pushes beyond the simple and into the relm of those that better understand just how CANBUS works and be able to understand what the signal looks like if there is a problem with the CANBUS itself.
Every system owner would need to be able to read, interpret and diagnose any such problem and this is where the whole thing falls over, if every system owner knew how to do this, there would be no high paid professionals following the career path .... a bit like graphic designers and web designers and those that repaired the photo development machines that processed exposed rolls of film .... still highly skilled but a skill not worth much these days ....

The tricky part is building a BMS out of basic off the shelf components and build in levels of redundancy, so if one part starts giving trouble, there is work around till the faulty part can be replaced ... just getting the faulty part can be the real problem when remote, so simply replacing the whole unit by carrying a spare becomes complex and expensive.

We built our own BMS units some 12 yrs ago, 7 levels of redundancy, the last level the system operator becomes the BMS, the second last level is battery isolation if no one heard or responded to all the alarms and flashing lights. This protects at both high cell voltage and low cell voltage but can determine if the cell voltage alarm is the real thing or a false alarm.

Sadly. a number of the parts required to build this BMS are now becoming near impossible top source, I'm fine with microprocessors, just as long as there is another micro processor further down the line that can step in to protect the battery if an upstream microprocessor drops its bundle .....

T1 Terry

That was how I started: a module per cel. It is very easy to do with an ATtiny and to communicate back to the central unit is the problem. The best way is to convert the measured voltage to a frequency, which can be pushed through an opto-isolator and decoded at the main module. By that time it is way better to throw $3 or $5 at each cell for an INA226 chip and get 16-bit precision measurements over I2C bus.

[s/v Jedi]

Originally Posted by Zanshin

I'm also working on a BMS using the INA229 for my 28V system. But if all you want the INA229 for is voltage measurement there are cheaper and better chips on the market. I use the 229 to also measure power, which is what it is built for and excels at. No need to write a library for the Arduino, I've got one that works interchangeably for most INA2xx chips: Zanduino INA Library

Nice, using interrupt handlers. I used the Adafruit library but will go over everything again. You needed 20-bit precision?

[s/v Jedi]

Here’s another picture from the prototyping: it has 4 INA219 modules and 3 different types of temperature sensors. This was a working BMS on a breadboard after I added control of a Blue System RBS. I used overkill, too large mosfets for that so will downscale it probably to a transistor. I’m trying to make it work with 3.3V vcc this time so we can use newer Arduinos with wifi and Bluetooth.

For temperature I have changed my mind and am now using the OneWire sensors because of their easy of use. You can buy waterproof models and slap them on and it’s done.

[Rumrace]

My oldest grandson and I have been playing with Ardunio and Raspberry pi. In 2008 Jeanneau came out with a Sailing option so you didn’t list too much during a dinner cruise. Turned out to be a raspberry pi type private board which is exactly what everyone BMS out there. A raspberry or Ardunio is too expensive. Two raspberries can watch each other and take over if one is taxed. We managed to follow Jupiter and build an Ardunio dissolved oxygen tester on a bread board. It recycles every 30 seconds. Got the design from the Ardunio projects forum. It drew 2Amp off one of my 200-400 amp banks and blew up real good. I’ve found it’s best to run either off a on board lithium 2400mA and charger. While the raspberry may shield itself from high amps it doesn’t protect other boards from drawing killer 2.
I use a Raspberry on my 3D printers with software called octopus. I print some high tech stuff with Pa6-11 Nylon for marine use. I make a bunch of ASA items for marine use. One is a waterproof to 1 atmosphere raspberry pi case. Has two ports. Inner carriage to hold one pi and one wireless board one battery.
They are dirt cheap if you need one. Takes 16 hours to print all the parts. If anyone is interested it will save you at least $40.00 just on a case for wet location.
Anyone interested I take some photos or if anyone wants the data I consider it public domain and print it yourself. I use ASA to print it. Water UV and impact strength is hard to beat. The pi in my boat opens a power hatch turns on heat air golden rods in winter. Has its own cell line watches with two cameras (one failed during winter) I’m also 3D printing my 3rd attempt to make a raspberry camera housing which will aim any direction on command. Has two 90 RC servos. The blue one. Cooked both during winter.
. I bought the new Sony pi Camera. WOW.
I’m currently printing a 2 year old life size TRex and a Rasoberry has been helping my optimize the best position to print each bone in. 90 hours to print the Cevical vertebrae.
I rather buy a BMS of a local electronics geek than China so please keep it up.

[goboatingnow]

Originally Posted by wholybee

I don't believe that is true. The cheapest ones use simple transistor logic and resistors to control a FET on/off based on voltage inputs. I've come across dozens of schematics that show this. I can't say with absolute certainty that programmable BMS's do or don't maintain a simple hardware failsafe, but I expect they do. The overkill solar BMS does have "hardware" protection limits given in the manual but I've not looked at a schematic to confirm it. If it can be shown that a REC or similar doesn't, that makes the job easier, that's for sure.

Microcontroller is a much better job.

[goboatingnow]

Nick thanks for posting.

I’m using esp32 as it’s more powerful and can report via wifi Bluetooth or CAN , I’m looking at multiplexing the voltage measurement as it’s used one high performance ADC and One calibration


I’m ploughing through understanding the VE.CAN protocol but I think I’ll use my yacht devices nmea2000 wifi gateway to get monitoring data onto the nmea 2000 bus.

The bigger task is a iPhone app to allow it all to be configured and monitored.

As I am installing a Rut950 lTE gateway it then should be practical to monitor it on internet as I’m not convinced by the Venus Os Victron system.

[s/v Jedi]

Originally Posted by goboatingnow

Nick thanks for posting.

I’m using esp32 as it’s more powerful and can report via wifi Bluetooth or CAN , I’m looking at multiplexing the voltage measurement as it’s used one high performance ADC and One calibration


I’m ploughing through understanding the VE.CAN protocol but I think I’ll use my yacht devices nmea2000 wifi gateway to get monitoring data onto the nmea 2000 bus.

The bigger task is a iPhone app to allow it all to be configured and monitored.

As I am installing a Rut950 lTE gateway it then should be practical to monitor it on internet as I’m not convinced by the Venus Os Victron system.

A BMS doesn’t require a powerful microcontroller but things like CanBus, Bluetooth and wifi are only available on the powerful units. I like the Teensy with on board CANbus. I also like the new Arduino MKR 1010 with the Arm Cortex. You simply build a website as front-end for the BMS.

But none of that is needed, all it needs to do is monitor and protect the battery. The more invisible the better