DIY BMS Help

mateusleo · Sep 14, 2016

Hello folks,

Me and some friends, along with a teacher, have been porjecting an eletric fsae car. We received resources to buy the motor, controller, tires and suspension, but not enought money for battery and BMS, so we'll have to do our own. The main problem is that we have no experience, but we think we can break that barrier with our effort. We will certainly take sometime to do a trustworthy BMS. About the BMS design, we have choose to go for switched capacitors topology, which, long story short, conncets a capacitor in parallel to cell 1, then to disconnects cell 1 and conncets to cell 2, then another capacitor does the same to cells 2 and 3, and so there will be N-1 capacitors for N cells. This way, all cells will reach same voltage. But, i've had a hard time thinking of ways i could monitor all cell voltages, there will be about 70 cells in series, theres no microcontroller with so many analog inputs to read all the cells voltage.

I've been looking for a IC that could read several analog voltages and send them by a serial port to the microcontroller, but couldn't find any. So, what i've come here to ask is how would you measure the individual voltages in this case? Thank you!

flangefrog · Sep 14, 2016

I think you might find it hard/expensive to make a discrete circuit to switch the capacitors and drive the switches from a ground referenced microcontroller. I would be looking at the existing chips for active balancing using inductors such as the EMB1428Q. There's a chance this chip could be adapted to help with capacitive balancing.

Even if you were wanting to roll your own solution I do think an inductive balancing system using the cell to battery or battery to cell topology would be much easier to make than a capacitive one. You could wind your own transformer with one primary and 72 secondary windings (or several transformers with less windings). And you would need isolated gate drivers to switch the mosfets for each cell and the battery. Or you can do it like the EMB1428Q and use two mosfets for each cell to connect the cells to a single transformer (one primary, one secondary) for each set of say 7 cells.

You can't just use a microcontroller with lots of inputs to measure the cell voltages. You need an AFE (Analog Front End) which will transpose the cell voltages so they are ground referenced for readout by an ADC. Here's a list I made of some of the AFEs and related ICs available. There are lots of others that measure only around 6 cells per chip but most of these are for >12 cells. Most AFEs can be stacked.

AD7280A
AD8280
ISL94203
ISL94212
ISL78600
ISL94208
EMB1432Q
BQ76PL455A-Q1
BQ76940
BQ34Z100-G1
BQ76PL536A
BQ77904
BQ78350

There are also some options here although it's a bit outdated: http://liionbms.com/php/bms_chips_options.php

dmwahl · Sep 15, 2016

If you're budget limited, building your own BMS will not help. Although it is a fun exercise (speaking from both personal and professional experience). Don't bother with capacitive or switched inductors, way too complicated and unless you use old laptop cells or hobby lipo (bad idea) your pack will not require a large balance current after the initial balance. Resistive balancing works very well in a digital system (much better than analog), and is by far the simplest to implement.

mateusleo · Sep 16, 2016

flangefrog said:
I think you might find it hard/expensive to make a discrete circuit to switch the capacitors and drive the switches from a ground referenced microcontroller. I would be looking at the existing chips for active balancing using inductors such as the EMB1428Q. There's a chance this chip could be adapted to help with capacitive balancing.

Even if you were wanting to roll your own solution I do think an inductive balancing system using the cell to battery or battery to cell topology would be much easier to make than a capacitive one. You could wind your own transformer with one primary and 72 secondary windings (or several transformers with less windings). And you would need isolated gate drivers to switch the mosfets for each cell and the battery. Or you can do it like the EMB1428Q and use two mosfets for each cell to connect the cells to a single transformer (one primary, one secondary) for each set of say 7 cells.

You can't just use a microcontroller with lots of inputs to measure the cell voltages. You need an AFE (Analog Front End) which will transpose the cell voltages so they are ground referenced for readout by an ADC. Here's a list I made of some of the AFEs and related ICs available. There are lots of others that measure only around 6 cells per chip but most of these are for >12 cells. Most AFEs can be stacked.

AD7280A
AD8280
ISL94203
ISL94212
ISL78600
ISL94208
EMB1432Q
BQ76PL455A-Q1
BQ76940
BQ34Z100-G1
BQ76PL536A
BQ77904
BQ78350

There are also some options here although it's a bit outdated: http://liionbms.com/php/bms_chips_options.php

Really good info and nice tips, i will certainly look into all of the ics you suggested

Enviado de meu BLU STUDIO C HD usando Tapatalk

mateusleo · Sep 16, 2016

dmwahl said:
If you're budget limited, building your own BMS will not help. Although it is a fun exercise (speaking from both personal and professional experience). Don't bother with capacitive or switched inductors, way too complicated and unless you use old laptop cells or hobby lipo (bad idea) your pack will not require a large balance current after the initial balance. Resistive balancing works very well in a digital system (much better than analog), and is by far the simplest to implement.

We are using old laptop cells, so they will probably be out if capacity balance, which means we need in— use energy transfer, so good cells will charge bad cells. We have to build the bms, this kind of stuff is extremely expensive in Brazil, we will not be able to buy new one for the voltage we need

Enviado de meu BLU STUDIO C HD usando Tapatalk

flangefrog · Sep 16, 2016

mateusleo said:
dmwahl said:

If you're budget limited, building your own BMS will not help. Although it is a fun exercise (speaking from both personal and professional experience). Don't bother with capacitive or switched inductors, way too complicated and unless you use old laptop cells or hobby lipo (bad idea) your pack will not require a large balance current after the initial balance. Resistive balancing works very well in a digital system (much better than analog), and is by far the simplest to implement.

Click to expand...

We are using old laptop cells, so they will probably be out if capacity balance, which means we need in— use energy transfer, so good cells will charge bad cells. We have to build the bms, this kind of stuff is extremely expensive in Brazil, we will not be able to buy new one for the voltage we need

If you haven't already bought the motor and controller I would go for a far lower voltage system. You will get best efficiency between 12S and 20S and it will make your battery and BMS much simpler. You may find it better to match your parallel cell banks for similar capacity/impedance rather than using an active balancing BMS.

Depending how big your batteries are, it might be cheaper to buy better batteries than build a 72S BMS.

dmwahl · Sep 18, 2016

Old laptop cells aren't intended for high current even when they're new. You'll need a huge pack to get any decent power and then the weight and size may be a problem. I'd suggest checking out okashira's sale thread for Tesla cells. He can build the pack for a very reasonable price and they're good cells, not old laptop cells.

Typically higher voltage systems are more efficient, so I don't know where the 12-20S sweet spot comes from. Maybe cost efficient? Depending on the competition rules, you might be able to use a few smaller (16-24S) BMSs in series and design your own LVC and overcurrent protection since the onboard stuff won't handle the full voltage.

flangefrog · Sep 18, 2016

dmwahl said:
Typically higher voltage systems are more efficient, so I don't know where the 12-20S sweet spot comes from. Maybe cost efficient?.

As long as the wires are sized appropriately and the correct winding motor is chosen there is no theoretical difference in system efficiency at high or low voltages. The 12S-20S sweet spot is where the currently available mosfets are most efficient. I'm not sure what sort of power we're talking about here though so if this is 100s of kWs then it wouldn't be easy to use such a low voltage.

dmwahl · Sep 19, 2016

As I recall there are all sorts of rules for the FSAE. I was part of it in college, as well as the hybrid vehicle team, and for some parts the rules are such that they basically force you into a specific voltage if you want a fighting chance. Like max power, max battery voltage, max fused current, etc so you can either go with their prescribed voltage and current or lose the race.

DIY BMS Help

mateusleo

1 kW

flangefrog

1 kW

dmwahl

10 kW

mateusleo

1 kW

mateusleo

1 kW

flangefrog

1 kW

dmwahl

10 kW

flangefrog

1 kW

dmwahl

10 kW

Similar threads