Build your own BMS

[AKA-None]

Hmmm so why wouldn’t a mesh system work where you could combine individual cells rather than treat a bunch of cells as a “battery “?

[s/v Jedi]

Originally Posted by AKA-None

Hmmm so why wouldn’t a mesh system work where you could combine individual cells rather than treat a bunch of cells as a “battery “?

Oh I like that thought When a cell fails you just take it out of the pool, another cell taking over.

I think the problem will be the interconnect matrix. I’m not easily backing off costly options, but each cell having multiple disconnect solenoids quickly adds up. Using MOSFET’s introduces a lot of resistance to the circuit as you now get 4 times as many for a 12V or 8 times as many for 24V compared to a single MOSFET disconnect for a whole battery.

Another challenge is the cell charging setup. The interconnecting matrix may need to form a charge bus as well as load bus for the cells.

[goboatingnow]

I just reviewed the TI range of BMS support chips. It’s definitely the way to go , 6-18 cell voltage monitors , 9 temp monitors. I2C , 12-16 bit ADC. Differential cell voltage plus high side drivers for mosfet overcurrent ( and other disconnects. ) high common mode ability ground referenced i2C etc.

All this without any microcontroller input and chip prices from about $5 upwards.

Couple that to a ATTINY , or AVRDA series and you have a very nice BMS building block.

Add a ESP32 coordinating master node for comms ( especially the new C series RISC V based and you have one of the most flexible , all cell monitor safest BMS design around.

[s/v Jedi]

Originally Posted by goboatingnow

I just reviewed the TI range of BMS support chips. It’s definitely the way to go , 6-18 cell voltage monitors , 9 temp monitors. I2C , 12-16 bit ADC. Differential cell voltage plus high side drivers for mosfet overcurrent ( and other disconnects. ) high common mode ability ground referenced i2C etc.

All this without any microcontroller input and chip prices from about $5 upwards.

Couple that to a ATTINY , or AVRDA series and you have a very nice BMS building block.

Add a ESP32 coordinating master node for comms ( especially the new C series RISC V based and you have one of the most flexible , all cell monitor safest BMS design around.

Yes, those chips are nice. I hope you have a microscope to solder them

[goboatingnow]

Originally Posted by s/v Jedi

Yes, those chips are nice. I hope you have a microscope to solder them

SMD reflow oven.

[deepthought]

I am guessing you have already seen https://github.com/stuartpittaway/diyBMSv4



Have used it on 12 and 24v systems.

[goboatingnow]

Originally Posted by deepthought

I am guessing you have already seen https://github.com/stuartpittaway/diyBMSv4



Have used it on 12 and 24v systems.

Yes but I don’t think the ADC is precise enough

[goboatingnow]

Originally Posted by goboatingnow

Yes but I don’t think the ADC is precise enough

It’s only 10 but the new ATTINY series has 12 bit ADCS

[rgleason]

Is this it?

Microchip ATTINY1627 12-bit diff ADC with PGA

https://www.microchip.com/en-us/prod...cus/attiny1627

https://www.microchip.com/en-us/product/ATtiny1627

How to Use the 12-Bit Differential ADC with PGA

The new 12-bit ADC offers true differential measuring capabilities with optional hardware accumulation of up to 1024 samples, which effectively gives up to 17-bits resolution. With sampling rates up to 375 ksps, the analog data is available very rapidly. The Programmable Gain Amplifier (PGA) for the ADC can amplify single-ended and differential analog inputs to make it possible to measure even small amplitude signals efficiently.


Oddly Mouser calls this 8-bit in this Microchip Technology ATtiny1627 Curiosity Nano Kit (DM080104


Prices https://octopart.com/attiny1627-mfr-...yAAEgIJ__D_BwE

[goboatingnow]

Yes. That’s the new series. I used a lot of them

[goboatingnow]

But I will actually move away from battery mounted local monitoring to using a AD7284 or similar analog front end and build the whole battery ie 4S into a ECU style enclosure and run cell wires.

[s/v Jedi]

So… differential input, 12-bit ADC. Input+ and Input- must be at a level between ground and 3.3V. I am assuming it’s an ATTiny per cell, which will work well. What would be the maximum voltage we can possibly see per cell when overcharging? Winston specifies a maximum charge cutoff voltage of 4V so for this circuit a voltage divider 5:3 or so is needed.

Assuming maximum measurement value is 5V then resolution is 5/4096=1.22mV

All the talk about oversampling is only applicable to measuring AC signals, so for this DC application were simply stuck with 12-bit resolution, which isn’t enough to get 1mV precision.

Maybe we can fix that by reducing measurement range to 2.5V to 5V. That won’t be easy but let’s go with it. Now we have an accurate resolution, how do we send that to the central controller with galvanic isolation? Bluetooth?

[goboatingnow]

Originally Posted by s/v Jedi

So… differential input, 12-bit ADC. Input+ and Input- must be at a level between ground and 3.3V. I am assuming it’s an ATTiny per cell, which will work well. What would be the maximum voltage we can possibly see per cell when overcharging? Winston specifies a maximum charge cutoff voltage of 4V so for this circuit a voltage divider 5:3 or so is needed.



Assuming maximum measurement value is 5V then resolution is 5/4096=1.22mV



All the talk about oversampling is only applicable to measuring AC signals, so for this DC application were simply stuck with 12-bit resolution, which isn’t enough to get 1mV precision.



Maybe we can fix that by reducing measurement range to 2.5V to 5V. That won’t be easy but let’s go with it. Now we have an accurate resolution, how do we send that to the central controller with galvanic isolation? Bluetooth?

Well I’m not going for ATTINY front ends now. I’m using a BMS orientated front end chip. These come in internal ADC and external ADC options typically 16 bit. The chip handles the common voltage range but all micro outputs are ground referenced. As I’m building 12v battery a common ground across all banks avoids any ground issues and or i2C ground referencing.

So ATTINY 16 series talking to a 16 bit adc fribt end ( some even have N channel gate generators to facilitate high side mosfet cutoff which is my preferred approach

An ESP32 will act as a comms clearing house for all battery bms.

[s/v Jedi]

Originally Posted by goboatingnow

Well I’m not going for ATTINY front ends now. I’m using a BMS orientated front end chip. These come in internal ADC and external ADC options typically 16 bit. The chip handles the common voltage range but all micro outputs are ground referenced. As I’m building 12v battery a common ground across all banks avoids any ground issues and or i2C ground referencing.

So ATTINY 16 series talking to a 16 bit adc fribt end ( some even have N channel gate generators to facilitate high side mosfet cutoff which is my preferred approach

An ESP32 will act as a comms clearing house for all battery bms.

Man you keep changing things, I can’t keep up

Okay, 16 but is fine, will give 0.5mV real world resolution I think I remember from testing. Working at common ground reference solves all problems around galvanic isolation just like I did with the INA226 modules.

Please start a thread on your BMS design and build. I have studied these bms chips and may adopt for a new version bms as well. Will use a wifi & Bluetooth Arduino though, maybe even Ethernet

[goboatingnow]

I will post some block diagrams

My goals change as I research the project.

Most commercial BMS seem to offer about 10 mV resolution. I’m aiming at +- 1mV.

I’m moving away from battery terminal mount monitors. In my case I am limiting it to a 4S design. ( in essence a form of ultra protected drop in batteries )

Hence my battery will be one pcb. With an analog front end chip to handle the level shifting and common mode issues. This allows differential sensing which seriously improves resolution over ground referenced sensing ( as in the ina226) . High side mosfet switching per battery allows step back capacity failure and retains ground referencing even after disconnect.

bms functionality primary as a safety monitor is quite a simplistic task. My main goal being to merely monitor LVE HVE , over current and over temp. Charge control and load disconnect is not being factored in

Each battery bms then networks Tons communications module that has CAN , wifi and Bluetooth. This requires sophisticated software on many cases. So we’ll see. My priority is CAN as I have a big background in CAN bus.