DIY BMS monitoring chip options?

Hello EndlessSphere,

I'm building a BMS to monitor a ~400V battery pack for an electric motorcycle, which would be broken into 4 packs of ~100V each, with about 20-24S in each of the 4 packs. At the moment I'm considering using LTC6802-2 cell monitoring chips and using 2 on each pack to monitor each pack independently. Ryan Biffard from Zero recommended using the Texas Instruments BQ76PL455A-Q1 stackable cell monitoring chips, with again using about 2 on each pack. I was planning to program them to run some switching relays to convert the 4, 100V packs from a 400V series mode to a 100V parallel.

I wanted to know what are your thoughts on each or what other good options that you have tried or tested? Or perhaps any critique of those chips? Of if you could link me to any threads for building a custom BMS, as my searches have come up empty.

Cheers,

I do not know much about this line of questioning, but the switching of such a high voltage from four 100V from series 4xAh to parallel 400V would be quite the feat. Of course you'd be doing that with no current running through it so it may be very easy.

Anyone who has used the BMS chips has commented that it is much, much harder than they expected.

Good luck in your project. Budget plenty of time and money for it.

VietJr said:

I'm building a BMS to monitor a ~400V battery pack for an electric motorcycle, which would be broken into 4 packs of ~100V each, with about 20-24S in each of the 4 packs. At the moment I'm considering using LTC6802-2 cell monitoring chips and using 2 on each pack to monitor each pack independently.

Click to expand...

Should work OK. I have some experience with them on a 57 volt pack and they work OK. However, you can not daisy chain them - you will need an isolated interface for the top one (probably both of them.) You might consider the LTC6813 if you want an easier interface design. (They use the IsoSPI interface that Analog Devices came up with.)

I was planning to program them to run some switching relays to convert the 4, 100V packs from a 400V series mode to a 100V parallel.

Click to expand...

Woah there! Why the heck would you want to do that? That makes BMSes MUCH harder to implement. For example, if you use the LTC6802-2, you'll now need four, and you'll need to isolators to deal with massive 100 volt transients.

billvon said:

Should work OK. I have some experience with them on a 57 volt pack and they work OK. However, you can not daisy chain them - you will need an isolated interface for the top one (probably both of them.) You might consider the LTC6813 if you want an easier interface design. (They use the IsoSPI interface that Analog Devices came up with.)

Click to expand...

I thought that the LTC 6802-2 were able to be daisy chained? Or do you mean that they are just difficult to interface/program if they are daisy chained?

billvon said:

Woah there! Why the heck would you want to do that? That makes BMSes MUCH harder to implement. For example, if you use the LTC6802-2, you'll now need four, and you'll need to isolators to deal with massive 100 volt transients.

Click to expand...

I only have a single >100V charger, so I was thinking that I would make a relay switching system to convert the higher voltage to a lower one to charge all the packs at once. My idea was basically having each pack's BMS be semi-independent, with 2 on each pack, so that they shouldn't have to deal with any huge voltage drops/spikes. Ryan Biffard from Zero pointed out that I would be able to do the same thing by just charging a 100V section of the pack while they're in series though, so that would be an option I'm considering now.

This is the initial relay system I had, with a mode selector set with a timer to avoid connecting the batteries in series and parallel at the same time. The relays also controlled the contactors between each pack when "charging" mode would be enabled, and the BMS would control the contactors during use to cut them off if they got too hot/low. That said, as Ryan Biffard pointed out, this may be too much added complexity, so I might just have the BMS chips monitor the pack in series, and just charge them as 100V sections in series as well.

Alan B said:

Anyone who has used the BMS chips has commented that it is much, much harder than they expected.

Good luck in your project. Budget plenty of time and money for it.

Click to expand...

Got any threads for anything like this? Or should I just look around for some similar projects?

VietJr said:

I thought that the LTC 6802-2 were able to be daisy chained? Or do you mean that they are just difficult to interface/program if they are daisy chained?

Click to expand...

They _can_ be daisy chained, but not natively. They need a bunch of diodes/resistors/transistors to do it (to translate levels etc.) See page 34 of the data sheet. The 6813 is a lot easier to daisy chain.

I only have a single >100V charger

Click to expand...

Are you really saying you are willing to spend a hundred hours or so designing a BMS, but you're not willing to spend $2K on a new charger? That's not really a recipe for success.

so I was thinking that I would make a relay switching system to convert the higher voltage to a lower one to charge all the packs at once. My idea was basically having each pack's BMS be semi-independent, with 2 on each pack, so that they shouldn't have to deal with any huge voltage drops/spikes. Ryan Biffard from Zero pointed out that I would be able to do the same thing by just charging a 100V section of the pack while they're in series though, so that would be an option I'm considering now.

Click to expand...

That's doable provided your charger is 100% isolated AND there's no active balancing during the time that you are intentionally unbalancing the pack. Sounds like a pain in the butt, though. You have to charge four times?

This is the initial relay system I had, with a mode selector set with a timer to avoid connecting the batteries in series and parallel at the same time.

Click to expand...

Until a contact sticks and you blow the traces off the board.

If you are serious about this, get the appropriate charger. If not, stick with what you have.

billvon said:

They _can_ be daisy chained, but not natively. They need a bunch of diodes/resistors/transistors to do it (to translate levels etc.) See page 34 of the data sheet. The 6813 is a lot easier to daisy chain.

Click to expand...

Ah, I see. Thanks for the suggestion! I'll keep that in mind.

billvon said:

Are you really saying you are willing to spend a hundred hours or so designing a BMS, but you're not willing to spend $2K on a new charger? That's not really a recipe for success.

Click to expand...

At the moment this project is going to take up most of the scarce funding given by my university (student team trying to design this bike), so we have more manpower than money at the moment. I would like to consider all options though, and I do see the problems with putting so much time into just the BMS. Any recommendations for a decent high voltage charger then?

billvon said:

That's doable provided your charger is 100% isolated AND there's no active balancing during the time that you are intentionally unbalancing the pack. Sounds like a pain in the butt, though. You have to charge four times?

Click to expand...

billvon said:

If you are serious about this, get the appropriate charger. If not, stick with what you have.

Click to expand...

My university set the rule such that I'd have to completely monitor the pack while it's charging anyways, and having a long charging time isn't a huge drawback for me either, so having to charge 4 times wouldn't be too much of an inconvenience. That said, if I had the extra funding available I would definitely just spring for a higher voltage charger. Know anyone that would give a good discount for students?

billvon said:

Until a contact sticks and you blow the traces off the board.

Click to expand...

Hmmm, that would be a problem if it were to occur. My other option is just removing each 100V pack when I'm charging and store them separately, and just hooking them back up in series for testing. Again, I have an excess of student manpower ready to slave away.

VietJr said:

At the moment this project is going to take up most of the scarce funding given by my university (student team trying to design this bike), so we have more manpower than money at the moment. I would like to consider all options though, and I do see the problems with putting so much time into just the BMS. Any recommendations for a decent high voltage charger then?

Click to expand...

Best one out there is probably the Manzanita Micro PFC-20. $2K.

However if you have the time to spare you might consider hacking together a bunch of lower voltage chargers to charge the battery. If you section it into 48V sections you could charge it with 8 Meanwell 48V chargers, for example. You'd just have to bring more wires to the charger connection. For example, eight PSP-600-48's would cost you $960 and give you 5kW of charging. Eight ELGC-300-H-DA's would cost you $600 and give you about 2kW charging.

Note that the above would NOT require unplugging batteries to charge.

Hmmm, that would be a problem if it were to occur. My other option is just removing each 100V pack when I'm charging and store them separately, and just hooking them back up in series for testing. Again, I have an excess of student manpower ready to slave away.

Click to expand...

Physically unplugging them would prevent that risk, yes. But if you are willing to do that, then bringing out separate charging connections for the subgroups shouldn't be hard (see above.)

billvon said:

Best one out there is probably the Manzanita Micro PFC-20. $2K.

Click to expand...

Thanks for the recommendation! I'll take a look into it.

billvon said:

However if you have the time to spare you might consider hacking together a bunch of lower voltage chargers to charge the battery. If you section it into 48V sections you could charge it with 8 Meanwell 48V chargers, for example. You'd just have to bring more wires to the charger connection. For example, eight PSP-600-48's would cost you $960 and give you 5kW of charging. Eight ELGC-300-H-DA's would cost you $600 and give you about 2kW charging.

Note that the above would NOT require unplugging batteries to charge.

Click to expand...

By hacking together lower voltage chargers, would that mean putting the chargers in series to boost voltage somehow? Or did you mean to charge the battery in sections of ~48V, but charging all sections at the same time to try and minimize any balancing issues? Trying to wrap my head around how the batteries wouldn't need to be unplugged to charge.

VietJr said:

By hacking together lower voltage chargers, would that mean putting the chargers in series to boost voltage somehow?

Click to expand...

Sort of.

Wire all the batteries together to generate 400V.

Bring out charger wiring at intermediate points, about 48 volts apart. Choose points so that there is the same number of cells between each wire.

Get a bunch of Meanwell supplies and configure them to provide the charge voltage that each section needs. (Ex - if each section has 12 cells, then set each supply to 50.4) '

Limit current on each supply as needed to protect the supply and batteries. On the LED supplies that's easy; it's just a pot you adjust. On the other supplies you may have to delve into the supply to make the adjustment.

Connect each supply to each group of cells via the charger wiring you brought out.

At this point each 50.4 volt supply is charging only its own 12 cells. They are connected in series, but the power supplies don't know that. Why? Because 1) they are fully isolated, so they don't care if their V- output isn't at ground. And 2) because each supply is connected directly to its own group of cells - so as far as it can tell, it's only charging them.

Some notes on this arrangement:

The wires at the top and bottom of this "stack" carry almost all the current. The intermediate wiring carries only balancing current. However it is still good practice to design everything to not melt down if the intermediate wiring has to carry full current, since failures do happen.

If you start with a fully balanced pack this method will initially unbalance it. No matter how carefully you set the voltage of each supply (and you should set it as carefully as you can) the charge currents will not be well balanced; power supply designers simply don't try to get accurate regulation of current, because for most supplies current limit is only for protection. (LED supplies are an exception.) What this means is that you MUST do a full charge each time. As long as the voltages match, the pack will come back into balance at the end of the charge.

Since failures can occur, be sure to monitor voltages of all outputs to make sure all sections are charging. If you have endless slave labor, a guy with a meter works fine. However, if you do use a meter, design the probe points so they cannot be shorted accidentally.

billvon said:

Wire all the batteries together to generate 400V.

Bring out charger wiring at intermediate points, about 48 volts apart. Choose points so that there is the same number of cells between each wire.

Get a bunch of Meanwell supplies and configure them to provide the charge voltage that each section needs. (Ex - if each section has 12 cells, then set each supply to 50.4) '

Click to expand...

Something along these lines? If I've made any wrong assumptions about the balancing current direction then please correct me. I'm assuming the chargers are acting as if they're in" series" so the total current through the pack is the same but each charger may just be providing a different amount to that total?


I kind of get why the pack would unbalance if the current going out of each charger isn't the same, but can you explain how the pack would come back into balance near the end of the charge cycle? Let's say a pack gets near the 48V full charge threshold, would it's charger just slow down it's charge rate and upon reaching the 48V just cut off? If that were the case, I could see how it would balance if the top/bottom packs hit full charge first and sequentially allows the middle packs to reach full charge.

VietJr said:

Something along these lines?

Click to expand...

Basically yes. Note, though, that since the positive lead of charger 1 and the negative lead of charger 2 are providing similar but opposite currents, the actual current flowing through the wire is very low - the currents cancel per Kirchoff.

I kind of get why the pack would unbalance if the current going out of each charger isn't the same, but can you explain how the pack would come back into balance near the end of the charge cycle?

Click to expand...

Because if you hold li-ion cells at 4.2 volts, they eventually reach a full SOC no matter what the current.

Note that your charger will not be cutting off automatically, at some .1C or .05C level or something - so it's not great for the batteries. (Keeping them at max voltage for longer than normal isn't great for them.) But it's cheap.

Let's say a pack gets near the 48V full charge threshold, would it's charger just slow down it's charge rate and upon reaching the 48V just cut off?

Click to expand...

No. The current will just tail off and go down and down until it's almost zero but it will never stop 100%. You have to decide when to stop charging after all cells are in the state you want.

I'm weighing options at the moment, but the multi-charger setup you suggested seems the idea to learn towards. The cost wouldn't be as high and I can also shift the complexity away from the bike so that's always a good thing. I'll post my progress here after choosing some chargers and building and testing the bank. Thanks for the help billvon!