Chargery BMS16T Teardown and circuit analysis? Anyone?

[Factory400]

I don't believe anyone has started a thread like this - a teardown of a Chargery BMS16T to analyze how it works.

Does this sound interesting?

Nerd Level: 1000

 

[JoeHam]

Mark me as interested

 

[Steve_S]

I've got a dead BMS8T I'donate to someone who wants to tear into it. It got shorted and the magic smoke escaped.

bottom with heatsink ..|.. top no heatsink ..|.. bottom with heatsink removed.

 

[grizzzman]

I've got a dead BMS8T I'donate to someone who wants to tear into it. It got shorted and the magic smoke escaped.

bottom with heatsink ..|.. top no heatsink ..|.. bottom with heatsink removed.
View attachment 29374 View attachment 29375 View attachment 29376

On the picture in the upper right, behind the external power port is U1. This is the chip responsible for external power. If you back feed it with a 25 Amp load, It will supply it tell the smoke Comes out.......

 

[Cal]

On the picture in the upper right, behind the external power port is U1. This is the chip responsible for external power. If you back feed it with a 25 Amp load, It will supply it tell the smoke Comes out.......

That's a major design flaw!

 

[Factory400]

I've got a dead BMS8T I'donate to someone who wants to tear into it. It got shorted and the magic smoke escaped.

bottom with heatsink ..|.. top no heatsink ..|.. bottom with heatsink removed.
View attachment 29374 View attachment 29375 View attachment 29376

Most likely this can be repaired.

 

[Factory400]

I identified and cataloged all the components on my BMS16T. I am not going to reverse engineer the thing, but I can clearly see the block diagram and better understand where some of the performance limitations come from.

 

[Cal]

Can you sketch out the block diagram?

 

[grizzzman]

Intrusting, this 8T I have has a green board and is version 3.1

 

[Factory400]

Can you sketch out the block diagram?

I think so - it is rather simple.

Currently trying to examine the balancing scheme. There is a MOSFET or BJT doing the energy dissipation directly. Perhaps operating in a linear region of the device or maybe PWM. Not sure yet. I don't think this is the approach I would take personally, but it appears to be doing the job. My guess at the moment is that it is not very consistent. The batteries don't really care about that, but the thermal dissipation does.

The heat spreader is not mechanically coupled to the external case - so rather inneficient way to get rid of balancing heat. That all by itself compromises the overall accuracy potential, but there are other signs that point to the instability I see in mine.

The system uses the internal STM32 microcontroller analog to digital converters which is probably why we see performance similar to an ENOB of 10 bits. The internal 12bit ADC peripheral are not great and require considerable care in PCB layout and firmware to get the best effective number of bits. There is an external voltage reference, which is good, but I cannot see that the layout does a good job of protecting the delicate analog signals. Remember we are talking about microvolts in this area - it takes very little to disturb that or bury the signal in noise. It is a multilayer PCB so I cannot see how ground and power planes are managed - just guessing for now.

 

[Cal]

I think so - it is rather simple.

Currently trying to examine the balancing scheme. There is a MOSFET or BJT doing the energy dissipation directly. Perhaps operating in a linear region of the device or maybe PWM. Not sure yet.

Most likely there's a resistor connected to a mosfet. That's the simplest approach.

The mosfet doesn't have to work as a switch. As a switch the resistor takes all the load (power dissipation).

The mosfet can also work as a current sink. Now the power dissipation can be distributed between the resistor and fet. If Vset is 1V and the resistor is 1 ohm, then there's 1A of balance current. If the cell is at 4V, then the resistor and fet dissipate 4W (combined). The resistor consumes 1W and the fet consumes 3W.

Can you check how cell voltages are measured?

 

[Factory400]

That is what I expected, but what I see on the PCB is that the MOSFET is the resistive element working in its linear region. Not yet sure if that is from delicate analog gate control or PWM. They may even be BJT's which would be a little easier to control perhaps.

 

[Factory400]

The cell voltages are sent through a simple mux and then on to the internal ESP32 ADC. From the layout, I am not surprised that the stability and accuracy are not very good. There appears to be very little concern for protecting the signals as they make their way to the ADC. Using the internal ADC for this measurement does not seem to be a good idea. Personally, I would have all of this sequestered in a protected area of the PCB.

 

[Cal]

The cell voltages are sent through a simple mux and then on to the internal ESP32 ADC. From the layout, I am not surprised that the stability and accuracy are not very good. There appears to be very little concern for protecting the signals as they make their way to the ADC. Using the internal ADC for this measurement does not seem to be a good idea. Personally, I would have all of this sequestered in a protected area of the PCB.

There has to be more to it than that. You may have 8 cells, or max 32V. Any voltage above 3.3V will smoke the ESP32. There's some circuit that drops all voltages below 3.3V. Perhaps there's a differential amplifier that takes the difference of the adjacent cell voltage and then sends it (the difference) to the adc. Look for an op amp IC.

 

[Factory400]

There has to be more to it than that. You may have 8 cells, or max 32V. Any voltage above 3.3V will smoke the ESP32. There's some circuit that drops all voltages below 3.3V. Perhaps there's a differential amplifier that takes the difference of the adjacent cell voltage and then sends it (the difference) to the adc. Look for an op amp IC.

There is an opamp that appears to be inline which could give a nice low impedance signal to the ADC. Voltage dividers on each input, but not so sure what is what between the mux and ADC.

The opamp and dividers would really be only for scaling to whatever the Vref is. If I get brave, I may look at the signal chain on the scope to get a sense of what it is doing on the analog side.

 

[Factory400]

Some reference photos to offer some context for future discussion.

 

[Factory400]

 

[Factory400]

U3: MAX3232 RS232 Driver https://www.ti.com/lit/ds/symlink/max3232.pdf?ts=1607110836940
H1-H4: CD4052 MUX https://www.ti.com/product/CD4052B
U5: MC1403 Voltage Reference https://www.onsemi.com/pub/Collateral/MC1403-D.PDF
U6: CH340 USB to UART https://cdn.sparkfun.com/datasheets/Dev/Arduino/Other/CH340DS1.PDF
Q13: MOSFET of some sort - ??
O2: LM358 https://www.ti.com/product/LM358
O1: TLC272 OpAmp https://www.ti.com/lit/ds/symlink/tlc272.pdf?ts=1607480069849&ref_url=https%3A%2F%2Fwww.ti.com%2Fproduct%2FTLC272
U1: LTC3703 DCDC converter controller https://www.analog.com/media/en/technical-documentation/data-sheets/3703fc.pdf

 

[Factory400]

When I get a bit more time, I would like to learn more about the MUX arrangement. It takes a while to beep all the connections and match them up to the datasheets. It looks like the MUX is summed into the LM358 O2 and then on to the TLC272 that is connected through a low pass RC into what appears to be the ADC pin on uC.

 

[Cal]

I don’t think getting scope measurements will help much. I would buzz it out and generate a schematic. The heart of the BMS is the cell voltage measurement.