Tesla model 3 BMB control boards

[johu]

Maybe that's past the noise threshold of voltage measurement. Do you get the same delta when checking a few cells with a multimeter?

[hanskraayeveld]

Yes almost same measurement, within 2mv.

I only do not understand why Tesla model S/X bleeding is thru 4x160 ohm resistance, so 40 ohm.

4v / 40 ohm = 100 mA

And I removed 1 brick wire, one which was/is bleeding, and measured 37 mA.

Why is Tesla 3/Y bleeding less mA? And how is it done? By resistance wire? Because I dont see any resistors or something getting hot. (I have a thermal imager)

[hanskraayeveld]

Hello,

I changed balancing a bit. If deltaV is big, BalHys grows.

Also showed it in parameters list.

But the strange thing is, I can upload it to both boards I bought, but 1 board it works, 1 led red constant, 1 led blinking.
Other board, after uploading, 1 led constant, other led stops blinking.

Message: communication problem....

When changing code back, it works...

So 1 board new code is working, 1 board didnt work.... strange??

If I do any modifications on code, no communication between stm and esp

Boatloader problem?

[tom91]

Spent my after noon with a 23s BMB playing around. Findings are:

1. Cells populate only half if using one of the two connectors
2. All cells and other values work with the 0.5.TA code
3. Balancing works with 0.5.TA code - Same behavior as observed in this thread.

Findings from bench work this afternoon
1. Balancing registers have a reset timer, I am sure its somewhere in a message being sent but have decide to brute force it for now
2. balance resistors and transistors are on the back of the BMBs

Improvements
1. Keep configuring balancing registers during "idle state" until 20 loops before starting measuring to ensure timer runs out
2. Make the measurement interval configurable, one measurement message loop is roughly 800mS, so recommended setting is 1 measurements per 2 seconds or more if you are looking to balance.
3. Alternating cells is required.

As always updates on github: https://github.com/damienmaguire/Tesla- ... 27ed1dcbed

[hanskraayeveld]

How many percent did balancing increase with your updates, do you think?

[tom91]

Depends on what you set the time between measurement to. So up to you now, there will always be the 0.8ish Seconds to cycle the measurement.

Plus the 50% duty cycle due to switching between even and odd.

[hanskraayeveld]

Both boards showing this message.

I can upload both boards your code, but cant see anything...

[hanskraayeveld]

Hello,

After both boards communication problem, I do this:

Upload in state of .hex, .bin to 0x08000000.
After it upload .hex again, and both boards are working now.

I did not understand what the problem was, but both boards working now.
Balancing is fine, 10 sec even, 10 sec odd, 0.8 for measuring voltages.

Thanks

[hanskraayeveld]

I am really pleased about the balance boards. Balancing is I think almost twice faster now.
Sorry for Tom and Damian, but I did again a few modifications, and I'll explain why.

We do a lot of battery repair. Most model S/X but M3 is coming now... Bad cells, cheap, unfortunatly. When a defect pack arrived, I want immediately charge to approx 4.00v cellvoltage, and balance nice. So 1 module is defect, 3 will be well balanced:

Module1 is defect. Bad leaking cells (lithium) which decrease cellvoltage of 2/3 bricks.

So we use your board to balance immediatly the 3 good modules (over a weekend), open battery, and place 1 other (well balanced module) in the battery. We only have to (dis)charge the replaced well balanced module, to same voltage, maybe balance whole battery again, and place in car.

Well balanced for us is very important, because you have a good 0-point (start point).

So I did some modifications, I make cellvoltage 10 times "better", so balancing can be better.

So thanks for all work done, I will improve a little, e.g. balancing per module, because we have 10 modules on a bench, with different voltages, so 1 controller can balance 4 modules with different voltages. (And maybe a sign after the voltage of which he is balancing)

Not important for 99% users, but important for us.

We bought 2 more controllers.

[hanskraayeveld]

I see I have to do some more modifications, umin x 10, avg cellv, etc

[hanskraayeveld]

[hanskraayeveld]

The measurements of the cell voltages is really very accurate... I begin to admire the batman chip...

I know why it has to be very accurate, because the SOC of the LPF. Only one millivolt can be a soc change of~ 3%.

[muehlpower]

I am very pleased to see progress in the development of the M3 BMS. The need to re-solder the Batman chip is still preventing me from purchasing it. I have a donor board, but I am unsure how to proceed. Does it also work with johu's “ESP32 CAN Web Interface”? It will be installed in the battery box, which is not ideal for Wi-Fi.

[tom91]

Does it also work with johu's “ESP32 CAN Web Interface”?

Yes make sure to flash a CAN bootloader before installing it into the battery box. It is an OI project so you can use the open inverter tools.

[hanskraayeveld]

This M3 battery balanced, module1 defect.

Other module1 balancing to 3687 mv, after it replacing.

[linwenling]

I am really pleased about the balance boards. Balancing is I think almost twice faster now.
Sorry for Tom and Damian, but I did again a few modifications, and I'll explain why.

We do a lot of battery repair. Most model S/X but M3 is coming now... Bad cells, cheap, unfortunatly. When a defect pack arrived, I want immediately charge to approx 4.00v cellvoltage, and balance nice. So 1 module is defect, 3 will be well balanced:

Screenshot_20251019_170110_Samsung Internet.jpgScreenshot_20251019_170119_Samsung Internet.jpg

Module1 is defect. Bad leaking cells (lithium) which decrease cellvoltage of 2/3 bricks.

So we use your board to balance immediatly the 3 good modules (over a weekend), open battery, and place 1 other (well balanced module) in the battery. We only have to (dis)charge the replaced well balanced module, to same voltage, maybe balance whole battery again, and place in car.

Well balanced for us is very important, because you have a good 0-point (start point).

So I did some modifications, I make cellvoltage 10 times "better", so balancing can be better.

So thanks for all work done, I will improve a little, e.g. balancing per module, because we have 10 modules on a bench, with different voltages, so 1 controller can balance 4 modules with different voltages. (And maybe a sign after the voltage of which he is balancing)

Not important for 99% users, but important for us.

We bought 2 more controllers.

happy me send me 0.7TA HEX data？ is stm32 data

[muehlpower]

Since Vintage Voltage doesn't have anything on offer yet and is likely to be inflexible and expensive, I now want to continue with Damien's board. I'm not very good at programming, so I'm relying on Tom and Damien's work. I need something that provides me with data on cell voltages and temperatures, perhaps even min/max values. The balancing has to work, and it would be great to have calculations for SOC, max charge, and discharge values. My battery consists of the 4 modules from a 2018 Model 3 LR. Can the current software and hardware handle that? I've seen that the current release is V0.03A. What data, e.g. current, does the board need to function properly, and can this be transmitted via CAN?

Translated with DeepL.com (free version)

[tom91]

it would be great to have calculations for SOC, max charge, and discharge values

It does not.

The balancing has to work

It does in the latest software version not the release V0.03A for some

can this be transmitted via CAN

It is all spotvalues in the OI framework, so can all be mapped onto CAN frames

[muehlpower]

I looked around on GitHub and cloned version 0.08.TC. Can this version balance my battery? In “param_prj.h” I found the following lines:
VALUE_ENTRY(version, VERSTR, 2001 ) \
VALUE_ENTRY(soc, “%”, 2002 ) \
VALUE_ENTRY(chargelim, “A”, 2003 ) \
VALUE_ENTRY(dischargelim,“A”, 2004 ) \

Are these values determined in version 0.08.TC or are they just created for later use?

[tom91]

Can this version balance my battery?

In theory yes, its just super slow

Values for someone to write logic for. Currently not populated

[linwenling]

I can't upload pictures

[muehlpower]

In theory yes, its just super slow

Slow balancing would be OK for me. I could imagine activating balancing after a full AC charge, and only then. So if I see significant differences, I'll just leave it on the AC charger overnight. I read somewhere that balancing can be enabled or disabled via CAN. The maximum charge and discharge rate should be determined based on the internal resistance of the battery and the open-circuit voltage. Perhaps a temperature correction could be added as well. The SOC should be determined approximately by the open-circuit voltage of the battery and a characteristic curve. I could also imagine a +- calculation of the energy going in and out, which is reset when the battery is completely empty or full. How is it done with your BMS?

Translated with DeepL.com (free version)

[tom91]

Feel free to code it in any functions you feel are missing.

[LRBen]

I've create a wiki for this board and put on the MAX chip parts. Model 3 bits I'll leave to someone that is using it with the Model 3 battery, since my board is only running with the MAX chip part.
https://openinverter.org/wiki/Tesla_Model_3_and_MAX_BMS